diff --git "a/checkpoint-5000/trainer_state.json" "b/checkpoint-5000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-5000/trainer_state.json"
@@ -0,0 +1,36221 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.505703422053232,
+  "eval_steps": 100,
+  "global_step": 5000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.9180965423583984,
+      "learning_rate": 3.7974683544303794e-06,
+      "loss": 0.1337,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.145163059234619,
+      "learning_rate": 7.594936708860759e-06,
+      "loss": 0.1635,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.690329551696777,
+      "learning_rate": 1.1392405063291139e-05,
+      "loss": 0.1942,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.002701282501221,
+      "learning_rate": 1.5189873417721518e-05,
+      "loss": 0.1871,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.975184917449951,
+      "learning_rate": 1.89873417721519e-05,
+      "loss": 0.176,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.468452453613281,
+      "learning_rate": 2.2784810126582277e-05,
+      "loss": 0.2465,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.971653938293457,
+      "learning_rate": 2.6582278481012654e-05,
+      "loss": 0.3365,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 15.753518104553223,
+      "learning_rate": 3.0379746835443035e-05,
+      "loss": 0.4271,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 12.906909942626953,
+      "learning_rate": 3.4177215189873416e-05,
+      "loss": 0.6491,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 11.588561058044434,
+      "learning_rate": 3.79746835443038e-05,
+      "loss": 0.8569,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.39958381652832,
+      "learning_rate": 4.177215189873417e-05,
+      "loss": 0.8824,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.734513282775879,
+      "learning_rate": 4.5569620253164554e-05,
+      "loss": 1.0863,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.504823684692383,
+      "learning_rate": 4.936708860759493e-05,
+      "loss": 1.0321,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.382841110229492,
+      "learning_rate": 5.316455696202531e-05,
+      "loss": 1.2056,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.642165184020996,
+      "learning_rate": 5.696202531645569e-05,
+      "loss": 1.2667,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.273102283477783,
+      "learning_rate": 6.075949367088607e-05,
+      "loss": 1.1953,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 9.236750602722168,
+      "learning_rate": 6.455696202531645e-05,
+      "loss": 1.3512,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.5139007568359375,
+      "learning_rate": 6.835443037974683e-05,
+      "loss": 1.1733,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.77164363861084,
+      "learning_rate": 7.215189873417722e-05,
+      "loss": 1.4029,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.658441066741943,
+      "learning_rate": 7.59493670886076e-05,
+      "loss": 1.3057,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.157665252685547,
+      "learning_rate": 7.974683544303796e-05,
+      "loss": 1.3839,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.226153373718262,
+      "learning_rate": 8.354430379746834e-05,
+      "loss": 1.3104,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.118317604064941,
+      "learning_rate": 8.734177215189872e-05,
+      "loss": 1.388,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.612947463989258,
+      "learning_rate": 9.113924050632911e-05,
+      "loss": 1.2505,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.770087242126465,
+      "learning_rate": 9.493670886075949e-05,
+      "loss": 1.4854,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.438778400421143,
+      "learning_rate": 9.873417721518986e-05,
+      "loss": 1.2458,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.019481182098389,
+      "learning_rate": 0.00010253164556962025,
+      "loss": 1.3169,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.281876564025879,
+      "learning_rate": 0.00010632911392405062,
+      "loss": 1.4458,
+      "step": 28
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.845592498779297,
+      "learning_rate": 0.000110126582278481,
+      "loss": 1.2691,
+      "step": 29
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.253188133239746,
+      "learning_rate": 0.00011392405063291139,
+      "loss": 1.3797,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.8740081787109375,
+      "learning_rate": 0.00011772151898734176,
+      "loss": 1.3857,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.267519950866699,
+      "learning_rate": 0.00012151898734177214,
+      "loss": 1.3929,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.905129432678223,
+      "learning_rate": 0.00012531645569620252,
+      "loss": 1.5487,
+      "step": 33
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.3816022872924805,
+      "learning_rate": 0.0001291139240506329,
+      "loss": 1.3274,
+      "step": 34
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.547843933105469,
+      "learning_rate": 0.00013291139240506327,
+      "loss": 1.3835,
+      "step": 35
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.268343448638916,
+      "learning_rate": 0.00013670886075949366,
+      "loss": 1.4118,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.892008304595947,
+      "learning_rate": 0.00014050632911392403,
+      "loss": 1.4075,
+      "step": 37
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.706852912902832,
+      "learning_rate": 0.00014430379746835443,
+      "loss": 1.3339,
+      "step": 38
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.807590961456299,
+      "learning_rate": 0.0001481012658227848,
+      "loss": 1.346,
+      "step": 39
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.422623634338379,
+      "learning_rate": 0.0001518987341772152,
+      "loss": 1.2038,
+      "step": 40
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.568236827850342,
+      "learning_rate": 0.00015569620253164554,
+      "loss": 1.1964,
+      "step": 41
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.055812358856201,
+      "learning_rate": 0.0001594936708860759,
+      "loss": 1.3539,
+      "step": 42
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.37033224105835,
+      "learning_rate": 0.0001632911392405063,
+      "loss": 1.1537,
+      "step": 43
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.628723382949829,
+      "learning_rate": 0.00016708860759493668,
+      "loss": 1.1918,
+      "step": 44
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7917227745056152,
+      "learning_rate": 0.00017088607594936708,
+      "loss": 1.1828,
+      "step": 45
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.8618366718292236,
+      "learning_rate": 0.00017468354430379745,
+      "loss": 1.1783,
+      "step": 46
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.379981994628906,
+      "learning_rate": 0.00017848101265822782,
+      "loss": 1.1991,
+      "step": 47
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.7640790939331055,
+      "learning_rate": 0.00018227848101265822,
+      "loss": 1.3562,
+      "step": 48
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.672415256500244,
+      "learning_rate": 0.0001860759493670886,
+      "loss": 1.2998,
+      "step": 49
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.8382604122161865,
+      "learning_rate": 0.00018987341772151899,
+      "loss": 1.2274,
+      "step": 50
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.9518237113952637,
+      "learning_rate": 0.00019367088607594936,
+      "loss": 1.316,
+      "step": 51
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.213583946228027,
+      "learning_rate": 0.00019746835443037973,
+      "loss": 1.224,
+      "step": 52
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.5902721881866455,
+      "learning_rate": 0.00020126582278481012,
+      "loss": 1.2601,
+      "step": 53
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4067130088806152,
+      "learning_rate": 0.0002050632911392405,
+      "loss": 1.2888,
+      "step": 54
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.537475824356079,
+      "learning_rate": 0.0002088607594936709,
+      "loss": 0.994,
+      "step": 55
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.438396453857422,
+      "learning_rate": 0.00021265822784810124,
+      "loss": 1.2096,
+      "step": 56
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.601135730743408,
+      "learning_rate": 0.0002164556962025316,
+      "loss": 1.2376,
+      "step": 57
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.794816493988037,
+      "learning_rate": 0.000220253164556962,
+      "loss": 1.1684,
+      "step": 58
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9558260440826416,
+      "learning_rate": 0.00022405063291139237,
+      "loss": 0.9734,
+      "step": 59
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.463297128677368,
+      "learning_rate": 0.00022784810126582277,
+      "loss": 1.1822,
+      "step": 60
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.382005214691162,
+      "learning_rate": 0.00023164556962025314,
+      "loss": 1.0938,
+      "step": 61
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.7794811725616455,
+      "learning_rate": 0.0002354430379746835,
+      "loss": 1.2722,
+      "step": 62
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.588163375854492,
+      "learning_rate": 0.0002392405063291139,
+      "loss": 1.0579,
+      "step": 63
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.996206521987915,
+      "learning_rate": 0.00024303797468354428,
+      "loss": 1.2467,
+      "step": 64
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.5940911769866943,
+      "learning_rate": 0.0002468354430379747,
+      "loss": 1.2461,
+      "step": 65
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.347010374069214,
+      "learning_rate": 0.00025063291139240505,
+      "loss": 1.0665,
+      "step": 66
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.515129566192627,
+      "learning_rate": 0.0002544303797468354,
+      "loss": 1.2582,
+      "step": 67
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.51462984085083,
+      "learning_rate": 0.0002582278481012658,
+      "loss": 1.0886,
+      "step": 68
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.4479589462280273,
+      "learning_rate": 0.00026202531645569616,
+      "loss": 1.1587,
+      "step": 69
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9662203788757324,
+      "learning_rate": 0.00026582278481012653,
+      "loss": 0.997,
+      "step": 70
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.8674981594085693,
+      "learning_rate": 0.00026962025316455696,
+      "loss": 1.048,
+      "step": 71
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.7003326416015625,
+      "learning_rate": 0.0002734177215189873,
+      "loss": 1.1272,
+      "step": 72
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.68707275390625,
+      "learning_rate": 0.0002772151898734177,
+      "loss": 0.9754,
+      "step": 73
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3709468841552734,
+      "learning_rate": 0.00028101265822784807,
+      "loss": 1.2285,
+      "step": 74
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.144456624984741,
+      "learning_rate": 0.00028481012658227844,
+      "loss": 1.1785,
+      "step": 75
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3562092781066895,
+      "learning_rate": 0.00028860759493670886,
+      "loss": 1.1158,
+      "step": 76
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.995312452316284,
+      "learning_rate": 0.00029240506329113923,
+      "loss": 1.1312,
+      "step": 77
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.838310480117798,
+      "learning_rate": 0.0002962025316455696,
+      "loss": 1.0565,
+      "step": 78
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.910526752471924,
+      "learning_rate": 0.0003,
+      "loss": 1.3094,
+      "step": 79
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.107647180557251,
+      "learning_rate": 0.0003037974683544304,
+      "loss": 1.007,
+      "step": 80
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1852099895477295,
+      "learning_rate": 0.0003075949367088607,
+      "loss": 1.1375,
+      "step": 81
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.1809871196746826,
+      "learning_rate": 0.0003113924050632911,
+      "loss": 1.0171,
+      "step": 82
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.137723684310913,
+      "learning_rate": 0.0003151898734177215,
+      "loss": 1.2669,
+      "step": 83
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.942014217376709,
+      "learning_rate": 0.0003189873417721518,
+      "loss": 1.0831,
+      "step": 84
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.143235921859741,
+      "learning_rate": 0.00032278481012658225,
+      "loss": 1.2395,
+      "step": 85
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8953638076782227,
+      "learning_rate": 0.0003265822784810126,
+      "loss": 1.0803,
+      "step": 86
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.119853973388672,
+      "learning_rate": 0.00033037974683544305,
+      "loss": 1.2251,
+      "step": 87
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.003676652908325,
+      "learning_rate": 0.00033417721518987336,
+      "loss": 1.0059,
+      "step": 88
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.959336280822754,
+      "learning_rate": 0.0003379746835443038,
+      "loss": 1.0912,
+      "step": 89
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 55.924034118652344,
+      "learning_rate": 0.00034177215189873416,
+      "loss": 2.2782,
+      "step": 90
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 22.1146183013916,
+      "learning_rate": 0.00034556962025316453,
+      "loss": 1.9118,
+      "step": 91
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 10.867006301879883,
+      "learning_rate": 0.0003493670886075949,
+      "loss": 1.4489,
+      "step": 92
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 8.29576587677002,
+      "learning_rate": 0.0003531645569620253,
+      "loss": 1.3153,
+      "step": 93
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 7.485663890838623,
+      "learning_rate": 0.00035696202531645564,
+      "loss": 1.3269,
+      "step": 94
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 44.048763275146484,
+      "learning_rate": 0.000360759493670886,
+      "loss": 2.5517,
+      "step": 95
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 32.70234298706055,
+      "learning_rate": 0.00036455696202531643,
+      "loss": 2.8621,
+      "step": 96
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 24.325580596923828,
+      "learning_rate": 0.00036835443037974686,
+      "loss": 5.4068,
+      "step": 97
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 13.49798583984375,
+      "learning_rate": 0.0003721518987341772,
+      "loss": 6.329,
+      "step": 98
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 61.25861358642578,
+      "learning_rate": 0.00037594936708860755,
+      "loss": 8.9344,
+      "step": 99
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 23.960065841674805,
+      "learning_rate": 0.00037974683544303797,
+      "loss": 7.8611,
+      "step": 100
+    },
+    {
+      "epoch": 0.19,
+      "eval_blimp_filtered_avg": 0.5692537313432836,
+      "eval_blimp_filtered_std": 0.0053149012855573845,
+      "step": 100
+    },
+    {
+      "epoch": 0.19,
+      "eval_blimp_supplement_avg": 0.4956896551724138,
+      "eval_blimp_supplement_std": 0.023015823142024827,
+      "step": 100
+    },
+    {
+      "epoch": 0.19,
+      "eval_vqa_filtered_avg": 0.33,
+      "eval_vqa_filtered_std": 0.047258156262526045,
+      "step": 100
+    },
+    {
+      "epoch": 0.19,
+      "eval_winoground_filtered_avg": 0.63,
+      "eval_winoground_filtered_std": 0.048523658709391,
+      "step": 100
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 32.528648376464844,
+      "learning_rate": 0.0003835443037974683,
+      "loss": 9.1303,
+      "step": 101
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 22.722553253173828,
+      "learning_rate": 0.0003873417721518987,
+      "loss": 8.4384,
+      "step": 102
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 9.758163452148438,
+      "learning_rate": 0.0003911392405063291,
+      "loss": 7.2559,
+      "step": 103
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 134.83558654785156,
+      "learning_rate": 0.00039493670886075945,
+      "loss": 14.6314,
+      "step": 104
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 25.381038665771484,
+      "learning_rate": 0.0003987341772151898,
+      "loss": 7.5377,
+      "step": 105
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 12.414955139160156,
+      "learning_rate": 0.00040253164556962025,
+      "loss": 7.1943,
+      "step": 106
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 9.32969856262207,
+      "learning_rate": 0.00040632911392405056,
+      "loss": 7.7736,
+      "step": 107
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 10.596403121948242,
+      "learning_rate": 0.000410126582278481,
+      "loss": 7.2289,
+      "step": 108
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 8.965174674987793,
+      "learning_rate": 0.00041392405063291136,
+      "loss": 6.9373,
+      "step": 109
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 28.596155166625977,
+      "learning_rate": 0.0004177215189873418,
+      "loss": 7.4455,
+      "step": 110
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 9.235132217407227,
+      "learning_rate": 0.0004215189873417721,
+      "loss": 7.2402,
+      "step": 111
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 8.0965576171875,
+      "learning_rate": 0.00042531645569620247,
+      "loss": 7.2256,
+      "step": 112
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 8.884477615356445,
+      "learning_rate": 0.0004291139240506329,
+      "loss": 7.0055,
+      "step": 113
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 8.821311950683594,
+      "learning_rate": 0.0004329113924050632,
+      "loss": 6.6268,
+      "step": 114
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.541005611419678,
+      "learning_rate": 0.00043670886075949364,
+      "loss": 6.4053,
+      "step": 115
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 9.087919235229492,
+      "learning_rate": 0.000440506329113924,
+      "loss": 6.2662,
+      "step": 116
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 13.510078430175781,
+      "learning_rate": 0.0004443037974683544,
+      "loss": 6.2643,
+      "step": 117
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.864687442779541,
+      "learning_rate": 0.00044810126582278475,
+      "loss": 6.1455,
+      "step": 118
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 17.963611602783203,
+      "learning_rate": 0.0004518987341772152,
+      "loss": 6.4472,
+      "step": 119
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.151864528656006,
+      "learning_rate": 0.00045569620253164554,
+      "loss": 6.1777,
+      "step": 120
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 8.707387924194336,
+      "learning_rate": 0.0004594936708860759,
+      "loss": 5.9939,
+      "step": 121
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.564164638519287,
+      "learning_rate": 0.0004632911392405063,
+      "loss": 6.0207,
+      "step": 122
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.42354154586792,
+      "learning_rate": 0.0004670886075949367,
+      "loss": 5.8775,
+      "step": 123
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 8.103471755981445,
+      "learning_rate": 0.000470886075949367,
+      "loss": 5.9943,
+      "step": 124
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.007039546966553,
+      "learning_rate": 0.00047468354430379745,
+      "loss": 5.9702,
+      "step": 125
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.726551055908203,
+      "learning_rate": 0.0004784810126582278,
+      "loss": 5.8341,
+      "step": 126
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.789422512054443,
+      "learning_rate": 0.00048227848101265814,
+      "loss": 5.9733,
+      "step": 127
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.5309865474700928,
+      "learning_rate": 0.00048607594936708856,
+      "loss": 5.9706,
+      "step": 128
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.5190696716308594,
+      "learning_rate": 0.000489873417721519,
+      "loss": 5.7478,
+      "step": 129
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.7941415309906006,
+      "learning_rate": 0.0004936708860759494,
+      "loss": 5.6085,
+      "step": 130
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.952862501144409,
+      "learning_rate": 0.0004974683544303797,
+      "loss": 5.7979,
+      "step": 131
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4186818599700928,
+      "learning_rate": 0.0005012658227848101,
+      "loss": 5.7719,
+      "step": 132
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.155359983444214,
+      "learning_rate": 0.0005050632911392405,
+      "loss": 5.7195,
+      "step": 133
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.8962011337280273,
+      "learning_rate": 0.0005088607594936708,
+      "loss": 5.7556,
+      "step": 134
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.960118532180786,
+      "learning_rate": 0.0005126582278481012,
+      "loss": 5.7123,
+      "step": 135
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.189756155014038,
+      "learning_rate": 0.0005164556962025316,
+      "loss": 5.6339,
+      "step": 136
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.618518590927124,
+      "learning_rate": 0.000520253164556962,
+      "loss": 5.532,
+      "step": 137
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7134299278259277,
+      "learning_rate": 0.0005240506329113923,
+      "loss": 5.7245,
+      "step": 138
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0144946575164795,
+      "learning_rate": 0.0005278481012658228,
+      "loss": 5.6191,
+      "step": 139
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.746931552886963,
+      "learning_rate": 0.0005316455696202531,
+      "loss": 5.6132,
+      "step": 140
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9589780569076538,
+      "learning_rate": 0.0005354430379746835,
+      "loss": 5.6625,
+      "step": 141
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.227086067199707,
+      "learning_rate": 0.0005392405063291139,
+      "loss": 5.5371,
+      "step": 142
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7142937183380127,
+      "learning_rate": 0.0005430379746835443,
+      "loss": 5.5103,
+      "step": 143
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.4383320808410645,
+      "learning_rate": 0.0005468354430379747,
+      "loss": 5.6392,
+      "step": 144
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.446405053138733,
+      "learning_rate": 0.000550632911392405,
+      "loss": 5.5727,
+      "step": 145
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4427984952926636,
+      "learning_rate": 0.0005544303797468354,
+      "loss": 5.5124,
+      "step": 146
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.1519070863723755,
+      "learning_rate": 0.0005582278481012658,
+      "loss": 5.3888,
+      "step": 147
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9773480892181396,
+      "learning_rate": 0.0005620253164556961,
+      "loss": 5.4587,
+      "step": 148
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8387157917022705,
+      "learning_rate": 0.0005658227848101265,
+      "loss": 5.5391,
+      "step": 149
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5714105367660522,
+      "learning_rate": 0.0005696202531645569,
+      "loss": 5.5563,
+      "step": 150
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.341513752937317,
+      "learning_rate": 0.0005734177215189872,
+      "loss": 5.473,
+      "step": 151
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.1008450984954834,
+      "learning_rate": 0.0005772151898734177,
+      "loss": 5.4656,
+      "step": 152
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.3675187826156616,
+      "learning_rate": 0.0005810126582278481,
+      "loss": 5.4719,
+      "step": 153
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.0577718019485474,
+      "learning_rate": 0.0005848101265822785,
+      "loss": 5.4442,
+      "step": 154
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.348527193069458,
+      "learning_rate": 0.0005886075949367088,
+      "loss": 5.3865,
+      "step": 155
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1914440393447876,
+      "learning_rate": 0.0005924050632911392,
+      "loss": 5.4589,
+      "step": 156
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1833913326263428,
+      "learning_rate": 0.0005962025316455696,
+      "loss": 5.3987,
+      "step": 157
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0698187351226807,
+      "learning_rate": 0.0006,
+      "loss": 5.3418,
+      "step": 158
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6521391868591309,
+      "learning_rate": 0.0005999999431264834,
+      "loss": 5.2667,
+      "step": 159
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0398824214935303,
+      "learning_rate": 0.0005999997725059552,
+      "loss": 5.4198,
+      "step": 160
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.3418430089950562,
+      "learning_rate": 0.0005999994881384801,
+      "loss": 5.4061,
+      "step": 161
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6336771249771118,
+      "learning_rate": 0.000599999090024166,
+      "loss": 5.287,
+      "step": 162
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.727086067199707,
+      "learning_rate": 0.0005999985781631637,
+      "loss": 5.2544,
+      "step": 163
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.2326830625534058,
+      "learning_rate": 0.0005999979525556674,
+      "loss": 5.2273,
+      "step": 164
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.3405826091766357,
+      "learning_rate": 0.0005999972132019142,
+      "loss": 5.1715,
+      "step": 165
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.2242249250411987,
+      "learning_rate": 0.0005999963601021845,
+      "loss": 5.2758,
+      "step": 166
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.1585261821746826,
+      "learning_rate": 0.0005999953932568019,
+      "loss": 5.2253,
+      "step": 167
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5858871936798096,
+      "learning_rate": 0.0005999943126661325,
+      "loss": 5.2005,
+      "step": 168
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.2485495805740356,
+      "learning_rate": 0.0005999931183305866,
+      "loss": 5.1953,
+      "step": 169
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.448172688484192,
+      "learning_rate": 0.0005999918102506166,
+      "loss": 5.1012,
+      "step": 170
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.3728394508361816,
+      "learning_rate": 0.0005999903884267189,
+      "loss": 5.1581,
+      "step": 171
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6083787679672241,
+      "learning_rate": 0.000599988852859432,
+      "loss": 5.2955,
+      "step": 172
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2761732339859009,
+      "learning_rate": 0.0005999872035493387,
+      "loss": 5.3084,
+      "step": 173
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.0587623119354248,
+      "learning_rate": 0.000599985440497064,
+      "loss": 5.052,
+      "step": 174
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2141993045806885,
+      "learning_rate": 0.0005999835637032766,
+      "loss": 5.2542,
+      "step": 175
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2662979364395142,
+      "learning_rate": 0.0005999815731686878,
+      "loss": 5.0592,
+      "step": 176
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.9875713586807251,
+      "learning_rate": 0.0005999794688940527,
+      "loss": 5.0603,
+      "step": 177
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.035649061203003,
+      "learning_rate": 0.0005999772508801689,
+      "loss": 5.2513,
+      "step": 178
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.2239720821380615,
+      "learning_rate": 0.0005999749191278773,
+      "loss": 5.0017,
+      "step": 179
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.2644716501235962,
+      "learning_rate": 0.0005999724736380623,
+      "loss": 5.1028,
+      "step": 180
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.07086181640625,
+      "learning_rate": 0.0005999699144116508,
+      "loss": 5.1158,
+      "step": 181
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.3119109869003296,
+      "learning_rate": 0.0005999672414496134,
+      "loss": 5.0615,
+      "step": 182
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.00112783908844,
+      "learning_rate": 0.0005999644547529635,
+      "loss": 4.9423,
+      "step": 183
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.053025245666504,
+      "learning_rate": 0.0005999615543227576,
+      "loss": 5.1368,
+      "step": 184
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.044136881828308,
+      "learning_rate": 0.0005999585401600955,
+      "loss": 5.0947,
+      "step": 185
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.1976217031478882,
+      "learning_rate": 0.00059995541226612,
+      "loss": 5.1375,
+      "step": 186
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.300540804862976,
+      "learning_rate": 0.0005999521706420172,
+      "loss": 5.0434,
+      "step": 187
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0234649181365967,
+      "learning_rate": 0.0005999488152890159,
+      "loss": 5.0229,
+      "step": 188
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2684745788574219,
+      "learning_rate": 0.0005999453462083886,
+      "loss": 5.0847,
+      "step": 189
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9677169322967529,
+      "learning_rate": 0.0005999417634014504,
+      "loss": 4.938,
+      "step": 190
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5459685325622559,
+      "learning_rate": 0.00059993806686956,
+      "loss": 4.9836,
+      "step": 191
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.115157127380371,
+      "learning_rate": 0.0005999342566141187,
+      "loss": 4.9723,
+      "step": 192
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.280566930770874,
+      "learning_rate": 0.0005999303326365714,
+      "loss": 5.0879,
+      "step": 193
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5884655714035034,
+      "learning_rate": 0.0005999262949384057,
+      "loss": 5.024,
+      "step": 194
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.0160022974014282,
+      "learning_rate": 0.0005999221435211527,
+      "loss": 4.9605,
+      "step": 195
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.1860222816467285,
+      "learning_rate": 0.0005999178783863862,
+      "loss": 4.9734,
+      "step": 196
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5145725011825562,
+      "learning_rate": 0.0005999134995357236,
+      "loss": 4.8301,
+      "step": 197
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6636500358581543,
+      "learning_rate": 0.0005999090069708252,
+      "loss": 5.0631,
+      "step": 198
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2012393474578857,
+      "learning_rate": 0.0005999044006933942,
+      "loss": 5.0297,
+      "step": 199
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4767887592315674,
+      "learning_rate": 0.0005998996807051771,
+      "loss": 4.9196,
+      "step": 200
+    },
+    {
+      "epoch": 0.38,
+      "eval_blimp_filtered_avg": 0.5465671641791044,
+      "eval_blimp_filtered_std": 0.005543515353530305,
+      "step": 200
+    },
+    {
+      "epoch": 0.38,
+      "eval_blimp_supplement_avg": 0.5021551724137931,
+      "eval_blimp_supplement_std": 0.022897193972434348,
+      "step": 200
+    },
+    {
+      "epoch": 0.38,
+      "eval_vqa_filtered_avg": 0.05,
+      "eval_vqa_filtered_std": 0.021904291355759085,
+      "step": 200
+    },
+    {
+      "epoch": 0.38,
+      "eval_winoground_filtered_avg": 0.52,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 200
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.2278707027435303,
+      "learning_rate": 0.0005998948470079637,
+      "loss": 4.943,
+      "step": 201
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.2997976541519165,
+      "learning_rate": 0.0005998898996035865,
+      "loss": 4.9822,
+      "step": 202
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.3038910627365112,
+      "learning_rate": 0.0005998848384939215,
+      "loss": 4.972,
+      "step": 203
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.0779736042022705,
+      "learning_rate": 0.0005998796636808877,
+      "loss": 4.8583,
+      "step": 204
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4388256072998047,
+      "learning_rate": 0.000599874375166447,
+      "loss": 5.1604,
+      "step": 205
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9082525968551636,
+      "learning_rate": 0.0005998689729526046,
+      "loss": 4.9315,
+      "step": 206
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1479289531707764,
+      "learning_rate": 0.000599863457041409,
+      "loss": 4.8325,
+      "step": 207
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.184837579727173,
+      "learning_rate": 0.0005998578274349513,
+      "loss": 4.8588,
+      "step": 208
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.2113370895385742,
+      "learning_rate": 0.0005998520841353661,
+      "loss": 4.8304,
+      "step": 209
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.038355588912964,
+      "learning_rate": 0.0005998462271448312,
+      "loss": 4.8703,
+      "step": 210
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.620894193649292,
+      "learning_rate": 0.000599840256465567,
+      "loss": 4.7287,
+      "step": 211
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.3175890445709229,
+      "learning_rate": 0.0005998341720998377,
+      "loss": 4.8838,
+      "step": 212
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.896052598953247,
+      "learning_rate": 0.0005998279740499498,
+      "loss": 4.7973,
+      "step": 213
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7717210054397583,
+      "learning_rate": 0.0005998216623182537,
+      "loss": 4.7623,
+      "step": 214
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6767300367355347,
+      "learning_rate": 0.0005998152369071423,
+      "loss": 4.8793,
+      "step": 215
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6933014392852783,
+      "learning_rate": 0.000599808697819052,
+      "loss": 4.743,
+      "step": 216
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.600928544998169,
+      "learning_rate": 0.000599802045056462,
+      "loss": 4.7623,
+      "step": 217
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.3260960578918457,
+      "learning_rate": 0.0005997952786218948,
+      "loss": 4.8457,
+      "step": 218
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.344473123550415,
+      "learning_rate": 0.000599788398517916,
+      "loss": 4.9594,
+      "step": 219
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.573988914489746,
+      "learning_rate": 0.0005997814047471342,
+      "loss": 4.8258,
+      "step": 220
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.4122810363769531,
+      "learning_rate": 0.0005997742973122011,
+      "loss": 4.834,
+      "step": 221
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.709999680519104,
+      "learning_rate": 0.0005997670762158115,
+      "loss": 4.7329,
+      "step": 222
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.41435968875885,
+      "learning_rate": 0.0005997597414607034,
+      "loss": 4.7151,
+      "step": 223
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8703913688659668,
+      "learning_rate": 0.0005997522930496578,
+      "loss": 4.6995,
+      "step": 224
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.471590280532837,
+      "learning_rate": 0.0005997447309854988,
+      "loss": 4.8798,
+      "step": 225
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8203284740447998,
+      "learning_rate": 0.0005997370552710936,
+      "loss": 4.7192,
+      "step": 226
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7644460201263428,
+      "learning_rate": 0.0005997292659093527,
+      "loss": 4.8428,
+      "step": 227
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.903369426727295,
+      "learning_rate": 0.0005997213629032292,
+      "loss": 4.5861,
+      "step": 228
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7533273696899414,
+      "learning_rate": 0.0005997133462557196,
+      "loss": 4.7641,
+      "step": 229
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.7285752296447754,
+      "learning_rate": 0.0005997052159698637,
+      "loss": 4.51,
+      "step": 230
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.5392253398895264,
+      "learning_rate": 0.0005996969720487439,
+      "loss": 4.6599,
+      "step": 231
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7242624759674072,
+      "learning_rate": 0.0005996886144954861,
+      "loss": 4.7385,
+      "step": 232
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9797385931015015,
+      "learning_rate": 0.0005996801433132592,
+      "loss": 4.5558,
+      "step": 233
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7230290174484253,
+      "learning_rate": 0.0005996715585052748,
+      "loss": 4.8748,
+      "step": 234
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6344454288482666,
+      "learning_rate": 0.000599662860074788,
+      "loss": 4.6311,
+      "step": 235
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.383590579032898,
+      "learning_rate": 0.0005996540480250971,
+      "loss": 4.7031,
+      "step": 236
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5823423862457275,
+      "learning_rate": 0.000599645122359543,
+      "loss": 4.6794,
+      "step": 237
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4282130002975464,
+      "learning_rate": 0.0005996360830815101,
+      "loss": 4.6274,
+      "step": 238
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0727856159210205,
+      "learning_rate": 0.0005996269301944254,
+      "loss": 4.5817,
+      "step": 239
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8817509412765503,
+      "learning_rate": 0.0005996176637017597,
+      "loss": 4.68,
+      "step": 240
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.5058006048202515,
+      "learning_rate": 0.0005996082836070262,
+      "loss": 4.6106,
+      "step": 241
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.2793718576431274,
+      "learning_rate": 0.0005995987899137814,
+      "loss": 4.6815,
+      "step": 242
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.987833857536316,
+      "learning_rate": 0.0005995891826256251,
+      "loss": 4.655,
+      "step": 243
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4055883884429932,
+      "learning_rate": 0.0005995794617461997,
+      "loss": 4.5517,
+      "step": 244
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.809360146522522,
+      "learning_rate": 0.0005995696272791911,
+      "loss": 4.5298,
+      "step": 245
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7232770919799805,
+      "learning_rate": 0.0005995596792283283,
+      "loss": 4.4667,
+      "step": 246
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.1689988374710083,
+      "learning_rate": 0.0005995496175973826,
+      "loss": 4.4395,
+      "step": 247
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.3216766119003296,
+      "learning_rate": 0.0005995394423901696,
+      "loss": 4.4613,
+      "step": 248
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.419859528541565,
+      "learning_rate": 0.0005995291536105467,
+      "loss": 4.5492,
+      "step": 249
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7091864347457886,
+      "learning_rate": 0.0005995187512624155,
+      "loss": 4.4551,
+      "step": 250
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.1945645809173584,
+      "learning_rate": 0.0005995082353497198,
+      "loss": 4.4362,
+      "step": 251
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.2608637809753418,
+      "learning_rate": 0.0005994976058764467,
+      "loss": 4.4508,
+      "step": 252
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.2600398063659668,
+      "learning_rate": 0.0005994868628466267,
+      "loss": 4.46,
+      "step": 253
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.732710838317871,
+      "learning_rate": 0.0005994760062643331,
+      "loss": 4.3906,
+      "step": 254
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2042415142059326,
+      "learning_rate": 0.0005994650361336819,
+      "loss": 4.44,
+      "step": 255
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0320358276367188,
+      "learning_rate": 0.0005994539524588328,
+      "loss": 4.681,
+      "step": 256
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.592881441116333,
+      "learning_rate": 0.0005994427552439882,
+      "loss": 4.4447,
+      "step": 257
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5317440032958984,
+      "learning_rate": 0.0005994314444933935,
+      "loss": 4.5799,
+      "step": 258
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.4910897016525269,
+      "learning_rate": 0.0005994200202113374,
+      "loss": 4.4919,
+      "step": 259
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.4859870672225952,
+      "learning_rate": 0.0005994084824021514,
+      "loss": 4.5821,
+      "step": 260
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.3539999723434448,
+      "learning_rate": 0.0005993968310702101,
+      "loss": 4.5231,
+      "step": 261
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.37140953540802,
+      "learning_rate": 0.0005993850662199314,
+      "loss": 4.566,
+      "step": 262
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.1939131021499634,
+      "learning_rate": 0.0005993731878557757,
+      "loss": 4.475,
+      "step": 263
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.511049509048462,
+      "learning_rate": 0.000599361195982247,
+      "loss": 4.3655,
+      "step": 264
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.303407907485962,
+      "learning_rate": 0.0005993490906038919,
+      "loss": 4.404,
+      "step": 265
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.3403290510177612,
+      "learning_rate": 0.0005993368717253005,
+      "loss": 4.3591,
+      "step": 266
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.3933802843093872,
+      "learning_rate": 0.0005993245393511054,
+      "loss": 4.6235,
+      "step": 267
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.3070017099380493,
+      "learning_rate": 0.0005993120934859827,
+      "loss": 4.4221,
+      "step": 268
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.419535756111145,
+      "learning_rate": 0.0005992995341346513,
+      "loss": 4.3901,
+      "step": 269
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.47270929813385,
+      "learning_rate": 0.0005992868613018731,
+      "loss": 4.2848,
+      "step": 270
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.5368038415908813,
+      "learning_rate": 0.000599274074992453,
+      "loss": 4.4598,
+      "step": 271
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.4578243494033813,
+      "learning_rate": 0.0005992611752112393,
+      "loss": 4.5297,
+      "step": 272
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.1982636451721191,
+      "learning_rate": 0.0005992481619631229,
+      "loss": 4.3862,
+      "step": 273
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.2984882593154907,
+      "learning_rate": 0.0005992350352530376,
+      "loss": 4.586,
+      "step": 274
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.5547841787338257,
+      "learning_rate": 0.0005992217950859609,
+      "loss": 4.2316,
+      "step": 275
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.5725514888763428,
+      "learning_rate": 0.0005992084414669127,
+      "loss": 4.2919,
+      "step": 276
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9258159399032593,
+      "learning_rate": 0.000599194974400956,
+      "loss": 4.2984,
+      "step": 277
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7338740825653076,
+      "learning_rate": 0.0005991813938931972,
+      "loss": 4.4651,
+      "step": 278
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.344088554382324,
+      "learning_rate": 0.0005991676999487853,
+      "loss": 4.4316,
+      "step": 279
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6079838275909424,
+      "learning_rate": 0.0005991538925729123,
+      "loss": 4.3205,
+      "step": 280
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7363066673278809,
+      "learning_rate": 0.0005991399717708137,
+      "loss": 4.1732,
+      "step": 281
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7125232219696045,
+      "learning_rate": 0.0005991259375477673,
+      "loss": 4.2868,
+      "step": 282
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9736409187316895,
+      "learning_rate": 0.0005991117899090947,
+      "loss": 4.252,
+      "step": 283
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.415002465248108,
+      "learning_rate": 0.0005990975288601596,
+      "loss": 4.2496,
+      "step": 284
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.809054374694824,
+      "learning_rate": 0.0005990831544063695,
+      "loss": 4.1963,
+      "step": 285
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.699670314788818,
+      "learning_rate": 0.0005990686665531744,
+      "loss": 4.3842,
+      "step": 286
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.017895221710205,
+      "learning_rate": 0.0005990540653060676,
+      "loss": 4.3124,
+      "step": 287
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.1736888885498047,
+      "learning_rate": 0.0005990393506705851,
+      "loss": 4.4515,
+      "step": 288
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.751234531402588,
+      "learning_rate": 0.0005990245226523063,
+      "loss": 4.2887,
+      "step": 289
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0288538932800293,
+      "learning_rate": 0.0005990095812568532,
+      "loss": 4.2811,
+      "step": 290
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.3899643421173096,
+      "learning_rate": 0.0005989945264898908,
+      "loss": 4.3356,
+      "step": 291
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0041441917419434,
+      "learning_rate": 0.0005989793583571273,
+      "loss": 4.2614,
+      "step": 292
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.4907017946243286,
+      "learning_rate": 0.000598964076864314,
+      "loss": 4.1688,
+      "step": 293
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.690474033355713,
+      "learning_rate": 0.0005989486820172446,
+      "loss": 4.3153,
+      "step": 294
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.4235070943832397,
+      "learning_rate": 0.0005989331738217566,
+      "loss": 4.3922,
+      "step": 295
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8015050888061523,
+      "learning_rate": 0.0005989175522837297,
+      "loss": 4.1913,
+      "step": 296
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.5248584747314453,
+      "learning_rate": 0.0005989018174090871,
+      "loss": 4.3262,
+      "step": 297
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6526167392730713,
+      "learning_rate": 0.0005988859692037948,
+      "loss": 4.2821,
+      "step": 298
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8668670654296875,
+      "learning_rate": 0.0005988700076738616,
+      "loss": 4.1382,
+      "step": 299
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.565722942352295,
+      "learning_rate": 0.0005988539328253394,
+      "loss": 4.2491,
+      "step": 300
+    },
+    {
+      "epoch": 0.57,
+      "eval_blimp_filtered_avg": 0.5032835820895523,
+      "eval_blimp_filtered_std": 0.005663343846892089,
+      "step": 300
+    },
+    {
+      "epoch": 0.57,
+      "eval_blimp_supplement_avg": 0.5474137931034483,
+      "eval_blimp_supplement_std": 0.022889054092193693,
+      "step": 300
+    },
+    {
+      "epoch": 0.57,
+      "eval_vqa_filtered_avg": 0.15,
+      "eval_vqa_filtered_std": 0.035887028128263734,
+      "step": 300
+    },
+    {
+      "epoch": 0.57,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 300
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8174269199371338,
+      "learning_rate": 0.0005988377446643233,
+      "loss": 4.0708,
+      "step": 301
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.5183809995651245,
+      "learning_rate": 0.000598821443196951,
+      "loss": 4.1305,
+      "step": 302
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7826297283172607,
+      "learning_rate": 0.0005988050284294034,
+      "loss": 4.1841,
+      "step": 303
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.5710526704788208,
+      "learning_rate": 0.0005987885003679041,
+      "loss": 4.146,
+      "step": 304
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.5514923334121704,
+      "learning_rate": 0.0005987718590187201,
+      "loss": 4.3937,
+      "step": 305
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.2432794570922852,
+      "learning_rate": 0.0005987551043881607,
+      "loss": 4.0521,
+      "step": 306
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.2738429307937622,
+      "learning_rate": 0.0005987382364825789,
+      "loss": 4.0893,
+      "step": 307
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6466275453567505,
+      "learning_rate": 0.0005987212553083701,
+      "loss": 4.1039,
+      "step": 308
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5602989196777344,
+      "learning_rate": 0.0005987041608719729,
+      "loss": 4.2788,
+      "step": 309
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6910903453826904,
+      "learning_rate": 0.0005986869531798687,
+      "loss": 4.0318,
+      "step": 310
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6199911832809448,
+      "learning_rate": 0.0005986696322385819,
+      "loss": 4.1986,
+      "step": 311
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7196334600448608,
+      "learning_rate": 0.00059865219805468,
+      "loss": 4.3012,
+      "step": 312
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.500410556793213,
+      "learning_rate": 0.0005986346506347732,
+      "loss": 4.2755,
+      "step": 313
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.675268530845642,
+      "learning_rate": 0.0005986169899855146,
+      "loss": 3.9372,
+      "step": 314
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7276028394699097,
+      "learning_rate": 0.0005985992161136005,
+      "loss": 4.0231,
+      "step": 315
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.4644769430160522,
+      "learning_rate": 0.0005985813290257701,
+      "loss": 4.0626,
+      "step": 316
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0290074348449707,
+      "learning_rate": 0.0005985633287288051,
+      "loss": 4.0342,
+      "step": 317
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1737308502197266,
+      "learning_rate": 0.0005985452152295306,
+      "loss": 4.0109,
+      "step": 318
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.610757827758789,
+      "learning_rate": 0.0005985269885348146,
+      "loss": 3.9087,
+      "step": 319
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.236865282058716,
+      "learning_rate": 0.0005985086486515676,
+      "loss": 4.1038,
+      "step": 320
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2351064682006836,
+      "learning_rate": 0.0005984901955867433,
+      "loss": 4.0982,
+      "step": 321
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.5661743879318237,
+      "learning_rate": 0.0005984716293473386,
+      "loss": 4.0045,
+      "step": 322
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9831912517547607,
+      "learning_rate": 0.0005984529499403928,
+      "loss": 3.9863,
+      "step": 323
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2589645385742188,
+      "learning_rate": 0.0005984341573729883,
+      "loss": 3.9025,
+      "step": 324
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.5023841857910156,
+      "learning_rate": 0.0005984152516522506,
+      "loss": 3.9602,
+      "step": 325
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2653017044067383,
+      "learning_rate": 0.0005983962327853478,
+      "loss": 4.0472,
+      "step": 326
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6561295986175537,
+      "learning_rate": 0.000598377100779491,
+      "loss": 3.9646,
+      "step": 327
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.09831166267395,
+      "learning_rate": 0.0005983578556419344,
+      "loss": 4.1035,
+      "step": 328
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.678687334060669,
+      "learning_rate": 0.0005983384973799747,
+      "loss": 3.9114,
+      "step": 329
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8050166368484497,
+      "learning_rate": 0.0005983190260009518,
+      "loss": 3.7936,
+      "step": 330
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.66739821434021,
+      "learning_rate": 0.0005982994415122485,
+      "loss": 3.9463,
+      "step": 331
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.656948208808899,
+      "learning_rate": 0.0005982797439212903,
+      "loss": 3.8805,
+      "step": 332
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.072005271911621,
+      "learning_rate": 0.0005982599332355457,
+      "loss": 3.9741,
+      "step": 333
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.5150046348571777,
+      "learning_rate": 0.000598240009462526,
+      "loss": 4.0249,
+      "step": 334
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9640929698944092,
+      "learning_rate": 0.0005982199726097856,
+      "loss": 3.7077,
+      "step": 335
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.17708683013916,
+      "learning_rate": 0.0005981998226849214,
+      "loss": 3.7082,
+      "step": 336
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0451955795288086,
+      "learning_rate": 0.0005981795596955735,
+      "loss": 3.7746,
+      "step": 337
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9555314779281616,
+      "learning_rate": 0.0005981591836494248,
+      "loss": 3.8458,
+      "step": 338
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.384778380393982,
+      "learning_rate": 0.0005981386945542008,
+      "loss": 3.9121,
+      "step": 339
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0258917808532715,
+      "learning_rate": 0.0005981180924176703,
+      "loss": 3.8368,
+      "step": 340
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.6506454944610596,
+      "learning_rate": 0.0005980973772476447,
+      "loss": 3.8632,
+      "step": 341
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.5015884637832642,
+      "learning_rate": 0.0005980765490519783,
+      "loss": 3.7228,
+      "step": 342
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.579393982887268,
+      "learning_rate": 0.000598055607838568,
+      "loss": 3.7149,
+      "step": 343
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.4940959215164185,
+      "learning_rate": 0.0005980345536153542,
+      "loss": 3.6288,
+      "step": 344
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 7.536722660064697,
+      "learning_rate": 0.0005980133863903195,
+      "loss": 3.5086,
+      "step": 345
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4062347412109375,
+      "learning_rate": 0.0005979921061714897,
+      "loss": 4.1495,
+      "step": 346
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.5928179025650024,
+      "learning_rate": 0.0005979707129669332,
+      "loss": 4.2505,
+      "step": 347
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.4179788827896118,
+      "learning_rate": 0.0005979492067847616,
+      "loss": 4.0442,
+      "step": 348
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1156680583953857,
+      "learning_rate": 0.000597927587633129,
+      "loss": 4.2731,
+      "step": 349
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9026774168014526,
+      "learning_rate": 0.0005979058555202324,
+      "loss": 4.3129,
+      "step": 350
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3405873775482178,
+      "learning_rate": 0.0005978840104543116,
+      "loss": 4.3819,
+      "step": 351
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0037131309509277,
+      "learning_rate": 0.0005978620524436495,
+      "loss": 4.0151,
+      "step": 352
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6274752616882324,
+      "learning_rate": 0.0005978399814965716,
+      "loss": 3.9356,
+      "step": 353
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9233535528182983,
+      "learning_rate": 0.0005978177976214462,
+      "loss": 4.0232,
+      "step": 354
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1571128368377686,
+      "learning_rate": 0.0005977955008266844,
+      "loss": 3.9846,
+      "step": 355
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1271684169769287,
+      "learning_rate": 0.0005977730911207403,
+      "loss": 3.906,
+      "step": 356
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9582709074020386,
+      "learning_rate": 0.0005977505685121106,
+      "loss": 3.789,
+      "step": 357
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.4580018520355225,
+      "learning_rate": 0.000597727933009335,
+      "loss": 3.8865,
+      "step": 358
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.5286750793457031,
+      "learning_rate": 0.0005977051846209958,
+      "loss": 3.9368,
+      "step": 359
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.4999233484268188,
+      "learning_rate": 0.0005976823233557183,
+      "loss": 3.8099,
+      "step": 360
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.6052910089492798,
+      "learning_rate": 0.0005976593492221704,
+      "loss": 3.786,
+      "step": 361
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.921733021736145,
+      "learning_rate": 0.000597636262229063,
+      "loss": 3.847,
+      "step": 362
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.597245693206787,
+      "learning_rate": 0.0005976130623851497,
+      "loss": 3.5998,
+      "step": 363
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8562277555465698,
+      "learning_rate": 0.0005975897496992267,
+      "loss": 3.75,
+      "step": 364
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9612540006637573,
+      "learning_rate": 0.0005975663241801333,
+      "loss": 3.6761,
+      "step": 365
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.6576099395751953,
+      "learning_rate": 0.0005975427858367514,
+      "loss": 3.7383,
+      "step": 366
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.431196689605713,
+      "learning_rate": 0.0005975191346780058,
+      "loss": 3.7981,
+      "step": 367
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.6568377017974854,
+      "learning_rate": 0.0005974953707128639,
+      "loss": 3.6041,
+      "step": 368
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7037498950958252,
+      "learning_rate": 0.0005974714939503362,
+      "loss": 3.6462,
+      "step": 369
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7122811079025269,
+      "learning_rate": 0.0005974475043994753,
+      "loss": 3.503,
+      "step": 370
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.5176008939743042,
+      "learning_rate": 0.0005974234020693774,
+      "loss": 3.7555,
+      "step": 371
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9933843612670898,
+      "learning_rate": 0.0005973991869691808,
+      "loss": 3.5048,
+      "step": 372
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.5340226888656616,
+      "learning_rate": 0.000597374859108067,
+      "loss": 3.541,
+      "step": 373
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2983906269073486,
+      "learning_rate": 0.0005973504184952601,
+      "loss": 3.6404,
+      "step": 374
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.5586453676223755,
+      "learning_rate": 0.0005973258651400267,
+      "loss": 3.6969,
+      "step": 375
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.757725715637207,
+      "learning_rate": 0.0005973011990516766,
+      "loss": 3.599,
+      "step": 376
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.5656229257583618,
+      "learning_rate": 0.000597276420239562,
+      "loss": 3.6601,
+      "step": 377
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8797898292541504,
+      "learning_rate": 0.0005972515287130779,
+      "loss": 3.6564,
+      "step": 378
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.6309489011764526,
+      "learning_rate": 0.0005972265244816622,
+      "loss": 3.5164,
+      "step": 379
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.4178309440612793,
+      "learning_rate": 0.0005972014075547954,
+      "loss": 3.5709,
+      "step": 380
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.4988398551940918,
+      "learning_rate": 0.0005971761779420007,
+      "loss": 3.6542,
+      "step": 381
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8177722692489624,
+      "learning_rate": 0.0005971508356528442,
+      "loss": 3.5816,
+      "step": 382
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.5029029846191406,
+      "learning_rate": 0.0005971253806969344,
+      "loss": 3.2749,
+      "step": 383
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.4254779815673828,
+      "learning_rate": 0.0005970998130839229,
+      "loss": 3.2952,
+      "step": 384
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9665387868881226,
+      "learning_rate": 0.0005970741328235038,
+      "loss": 3.5455,
+      "step": 385
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8038746118545532,
+      "learning_rate": 0.0005970483399254139,
+      "loss": 3.5366,
+      "step": 386
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.6266248226165771,
+      "learning_rate": 0.0005970224343994327,
+      "loss": 3.2679,
+      "step": 387
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.4961546659469604,
+      "learning_rate": 0.0005969964162553827,
+      "loss": 3.3184,
+      "step": 388
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.736702561378479,
+      "learning_rate": 0.0005969702855031285,
+      "loss": 3.6797,
+      "step": 389
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2267932891845703,
+      "learning_rate": 0.0005969440421525781,
+      "loss": 3.4398,
+      "step": 390
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.3329787254333496,
+      "learning_rate": 0.0005969176862136814,
+      "loss": 3.3586,
+      "step": 391
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0508344173431396,
+      "learning_rate": 0.000596891217696432,
+      "loss": 3.3894,
+      "step": 392
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.559171438217163,
+      "learning_rate": 0.0005968646366108652,
+      "loss": 3.2495,
+      "step": 393
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.263432025909424,
+      "learning_rate": 0.0005968379429670595,
+      "loss": 3.516,
+      "step": 394
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.7457019090652466,
+      "learning_rate": 0.000596811136775136,
+      "loss": 3.3847,
+      "step": 395
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9085086584091187,
+      "learning_rate": 0.0005967842180452584,
+      "loss": 3.3496,
+      "step": 396
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.301761269569397,
+      "learning_rate": 0.0005967571867876334,
+      "loss": 3.3374,
+      "step": 397
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.842649221420288,
+      "learning_rate": 0.0005967300430125096,
+      "loss": 3.3492,
+      "step": 398
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.3907227516174316,
+      "learning_rate": 0.0005967027867301791,
+      "loss": 3.3157,
+      "step": 399
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.585135817527771,
+      "learning_rate": 0.0005966754179509762,
+      "loss": 3.1627,
+      "step": 400
+    },
+    {
+      "epoch": 0.76,
+      "eval_blimp_filtered_avg": 0.5182089552238806,
+      "eval_blimp_filtered_std": 0.005845949168262265,
+      "step": 400
+    },
+    {
+      "epoch": 0.76,
+      "eval_blimp_supplement_avg": 0.4051724137931034,
+      "eval_blimp_supplement_std": 0.0222271518084265,
+      "step": 400
+    },
+    {
+      "epoch": 0.76,
+      "eval_vqa_filtered_avg": 0.14,
+      "eval_vqa_filtered_std": 0.034873508801977725,
+      "step": 400
+    },
+    {
+      "epoch": 0.76,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 400
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.605951189994812,
+      "learning_rate": 0.000596647936685278,
+      "loss": 3.6585,
+      "step": 401
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.6493512392044067,
+      "learning_rate": 0.0005966203429435042,
+      "loss": 3.5596,
+      "step": 402
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.4089688062667847,
+      "learning_rate": 0.000596592636736117,
+      "loss": 3.2325,
+      "step": 403
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.5161422491073608,
+      "learning_rate": 0.0005965648180736216,
+      "loss": 3.3631,
+      "step": 404
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.6291078329086304,
+      "learning_rate": 0.0005965368869665656,
+      "loss": 3.2127,
+      "step": 405
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.8974515199661255,
+      "learning_rate": 0.0005965088434255391,
+      "loss": 3.5671,
+      "step": 406
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.5424456596374512,
+      "learning_rate": 0.0005964806874611752,
+      "loss": 3.2789,
+      "step": 407
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9937212467193604,
+      "learning_rate": 0.0005964524190841494,
+      "loss": 3.2694,
+      "step": 408
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.6998950242996216,
+      "learning_rate": 0.0005964240383051797,
+      "loss": 3.2769,
+      "step": 409
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.7817816734313965,
+      "learning_rate": 0.0005963955451350269,
+      "loss": 3.2629,
+      "step": 410
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.3802156448364258,
+      "learning_rate": 0.0005963669395844946,
+      "loss": 3.1673,
+      "step": 411
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.6681243181228638,
+      "learning_rate": 0.0005963382216644285,
+      "loss": 3.3428,
+      "step": 412
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.6750260591506958,
+      "learning_rate": 0.0005963093913857174,
+      "loss": 3.2133,
+      "step": 413
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.070629835128784,
+      "learning_rate": 0.0005962804487592923,
+      "loss": 3.1108,
+      "step": 414
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.811374306678772,
+      "learning_rate": 0.0005962513937961272,
+      "loss": 2.9411,
+      "step": 415
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3361639976501465,
+      "learning_rate": 0.0005962222265072384,
+      "loss": 3.2635,
+      "step": 416
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.5616676807403564,
+      "learning_rate": 0.0005961929469036849,
+      "loss": 3.2143,
+      "step": 417
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.8401273488998413,
+      "learning_rate": 0.0005961635549965682,
+      "loss": 3.3364,
+      "step": 418
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.572719693183899,
+      "learning_rate": 0.0005961340507970324,
+      "loss": 3.0895,
+      "step": 419
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.6161249876022339,
+      "learning_rate": 0.0005961044343162643,
+      "loss": 3.1089,
+      "step": 420
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.139598846435547,
+      "learning_rate": 0.0005960747055654933,
+      "loss": 3.1956,
+      "step": 421
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.3102551698684692,
+      "learning_rate": 0.000596044864555991,
+      "loss": 3.0896,
+      "step": 422
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.8192949295043945,
+      "learning_rate": 0.000596014911299072,
+      "loss": 3.2738,
+      "step": 423
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.365920066833496,
+      "learning_rate": 0.0005959848458060933,
+      "loss": 3.103,
+      "step": 424
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.544568657875061,
+      "learning_rate": 0.0005959546680884543,
+      "loss": 3.301,
+      "step": 425
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.6904710531234741,
+      "learning_rate": 0.0005959243781575971,
+      "loss": 3.2683,
+      "step": 426
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.982872724533081,
+      "learning_rate": 0.0005958939760250065,
+      "loss": 3.2573,
+      "step": 427
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.4018594026565552,
+      "learning_rate": 0.0005958634617022095,
+      "loss": 2.8838,
+      "step": 428
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0201377868652344,
+      "learning_rate": 0.000595832835200776,
+      "loss": 2.9191,
+      "step": 429
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.527495265007019,
+      "learning_rate": 0.000595802096532318,
+      "loss": 3.0577,
+      "step": 430
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.8245353698730469,
+      "learning_rate": 0.0005957712457084905,
+      "loss": 3.0758,
+      "step": 431
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.4750018119812012,
+      "learning_rate": 0.0005957402827409905,
+      "loss": 3.0076,
+      "step": 432
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.7355194091796875,
+      "learning_rate": 0.0005957092076415582,
+      "loss": 3.1887,
+      "step": 433
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.6584534645080566,
+      "learning_rate": 0.0005956780204219756,
+      "loss": 3.0231,
+      "step": 434
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.4617878198623657,
+      "learning_rate": 0.0005956467210940678,
+      "loss": 3.0261,
+      "step": 435
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.6631488800048828,
+      "learning_rate": 0.000595615309669702,
+      "loss": 3.0689,
+      "step": 436
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.4126659631729126,
+      "learning_rate": 0.0005955837861607882,
+      "loss": 2.9705,
+      "step": 437
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.8271287679672241,
+      "learning_rate": 0.0005955521505792785,
+      "loss": 3.1093,
+      "step": 438
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.5489999055862427,
+      "learning_rate": 0.000595520402937168,
+      "loss": 3.0143,
+      "step": 439
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.6653859615325928,
+      "learning_rate": 0.0005954885432464939,
+      "loss": 3.2807,
+      "step": 440
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.5339930057525635,
+      "learning_rate": 0.000595456571519336,
+      "loss": 2.9776,
+      "step": 441
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.7197259664535522,
+      "learning_rate": 0.0005954244877678168,
+      "loss": 3.0289,
+      "step": 442
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.7648483514785767,
+      "learning_rate": 0.0005953922920041008,
+      "loss": 2.9245,
+      "step": 443
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.765788197517395,
+      "learning_rate": 0.0005953599842403954,
+      "loss": 3.0801,
+      "step": 444
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.719543695449829,
+      "learning_rate": 0.0005953275644889503,
+      "loss": 3.0883,
+      "step": 445
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.9884003400802612,
+      "learning_rate": 0.0005952950327620575,
+      "loss": 3.1145,
+      "step": 446
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.9348846673965454,
+      "learning_rate": 0.0005952623890720519,
+      "loss": 2.9237,
+      "step": 447
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.502580165863037,
+      "learning_rate": 0.0005952296334313104,
+      "loss": 3.0478,
+      "step": 448
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.701887845993042,
+      "learning_rate": 0.0005951967658522526,
+      "loss": 3.0357,
+      "step": 449
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.0564489364624023,
+      "learning_rate": 0.0005951637863473403,
+      "loss": 3.0926,
+      "step": 450
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.4964585304260254,
+      "learning_rate": 0.0005951306949290782,
+      "loss": 2.6521,
+      "step": 451
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.904565691947937,
+      "learning_rate": 0.0005950974916100127,
+      "loss": 3.0122,
+      "step": 452
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.579287052154541,
+      "learning_rate": 0.0005950641764027336,
+      "loss": 2.9739,
+      "step": 453
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.5247546434402466,
+      "learning_rate": 0.0005950307493198722,
+      "loss": 2.8119,
+      "step": 454
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.7055646181106567,
+      "learning_rate": 0.0005949972103741028,
+      "loss": 2.921,
+      "step": 455
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.7156434059143066,
+      "learning_rate": 0.0005949635595781417,
+      "loss": 2.7714,
+      "step": 456
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.5291056632995605,
+      "learning_rate": 0.000594929796944748,
+      "loss": 3.0187,
+      "step": 457
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.314779043197632,
+      "learning_rate": 0.0005948959224867231,
+      "loss": 2.7963,
+      "step": 458
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.9716535806655884,
+      "learning_rate": 0.0005948619362169106,
+      "loss": 2.7469,
+      "step": 459
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.8394272327423096,
+      "learning_rate": 0.0005948278381481966,
+      "loss": 2.6743,
+      "step": 460
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.7328013181686401,
+      "learning_rate": 0.0005947936282935097,
+      "loss": 2.784,
+      "step": 461
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.999652624130249,
+      "learning_rate": 0.0005947593066658207,
+      "loss": 2.8759,
+      "step": 462
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.5995413064956665,
+      "learning_rate": 0.000594724873278143,
+      "loss": 2.7863,
+      "step": 463
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.7001692056655884,
+      "learning_rate": 0.0005946903281435322,
+      "loss": 2.8129,
+      "step": 464
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.5811243057250977,
+      "learning_rate": 0.0005946556712750863,
+      "loss": 2.7286,
+      "step": 465
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.8942415714263916,
+      "learning_rate": 0.0005946209026859458,
+      "loss": 2.614,
+      "step": 466
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.875490427017212,
+      "learning_rate": 0.0005945860223892931,
+      "loss": 2.8207,
+      "step": 467
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.4743330478668213,
+      "learning_rate": 0.0005945510303983538,
+      "loss": 2.6251,
+      "step": 468
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.8182228803634644,
+      "learning_rate": 0.000594515926726395,
+      "loss": 2.7773,
+      "step": 469
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.6162047386169434,
+      "learning_rate": 0.0005944807113867265,
+      "loss": 2.8374,
+      "step": 470
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5769970417022705,
+      "learning_rate": 0.0005944453843927007,
+      "loss": 3.0183,
+      "step": 471
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.665252208709717,
+      "learning_rate": 0.0005944099457577118,
+      "loss": 2.6918,
+      "step": 472
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.053450345993042,
+      "learning_rate": 0.0005943743954951968,
+      "loss": 2.5065,
+      "step": 473
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3155739307403564,
+      "learning_rate": 0.0005943387336186345,
+      "loss": 2.7534,
+      "step": 474
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.1068921089172363,
+      "learning_rate": 0.0005943029601415467,
+      "loss": 2.8305,
+      "step": 475
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.202202558517456,
+      "learning_rate": 0.0005942670750774972,
+      "loss": 2.8442,
+      "step": 476
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.5554680824279785,
+      "learning_rate": 0.0005942310784400916,
+      "loss": 2.78,
+      "step": 477
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8777050971984863,
+      "learning_rate": 0.0005941949702429787,
+      "loss": 2.5985,
+      "step": 478
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8517574071884155,
+      "learning_rate": 0.000594158750499849,
+      "loss": 2.6107,
+      "step": 479
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.126801013946533,
+      "learning_rate": 0.0005941224192244354,
+      "loss": 2.632,
+      "step": 480
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.731307029724121,
+      "learning_rate": 0.0005940859764305133,
+      "loss": 2.7326,
+      "step": 481
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.709943175315857,
+      "learning_rate": 0.0005940494221319002,
+      "loss": 2.7909,
+      "step": 482
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.4330517053604126,
+      "learning_rate": 0.0005940127563424558,
+      "loss": 2.6274,
+      "step": 483
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9239248037338257,
+      "learning_rate": 0.0005939759790760822,
+      "loss": 2.6806,
+      "step": 484
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.4965001344680786,
+      "learning_rate": 0.000593939090346724,
+      "loss": 2.9055,
+      "step": 485
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.8432910442352295,
+      "learning_rate": 0.0005939020901683676,
+      "loss": 2.7895,
+      "step": 486
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.6640048027038574,
+      "learning_rate": 0.0005938649785550416,
+      "loss": 2.7128,
+      "step": 487
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.7810298204421997,
+      "learning_rate": 0.0005938277555208177,
+      "loss": 2.6343,
+      "step": 488
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.9577538967132568,
+      "learning_rate": 0.0005937904210798089,
+      "loss": 2.6071,
+      "step": 489
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.6489572525024414,
+      "learning_rate": 0.0005937529752461708,
+      "loss": 2.5435,
+      "step": 490
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.5083832740783691,
+      "learning_rate": 0.0005937154180341015,
+      "loss": 2.6614,
+      "step": 491
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.5562772750854492,
+      "learning_rate": 0.0005936777494578407,
+      "loss": 2.473,
+      "step": 492
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.6912349462509155,
+      "learning_rate": 0.000593639969531671,
+      "loss": 2.5515,
+      "step": 493
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.424018383026123,
+      "learning_rate": 0.0005936020782699166,
+      "loss": 2.5179,
+      "step": 494
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.6390740871429443,
+      "learning_rate": 0.0005935640756869448,
+      "loss": 2.6663,
+      "step": 495
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.5983397960662842,
+      "learning_rate": 0.0005935259617971639,
+      "loss": 2.4858,
+      "step": 496
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.621964454650879,
+      "learning_rate": 0.0005934877366150253,
+      "loss": 2.4362,
+      "step": 497
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.6918779611587524,
+      "learning_rate": 0.0005934494001550224,
+      "loss": 2.4366,
+      "step": 498
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.8172147274017334,
+      "learning_rate": 0.0005934109524316906,
+      "loss": 2.4391,
+      "step": 499
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.740181803703308,
+      "learning_rate": 0.0005933723934596079,
+      "loss": 2.3854,
+      "step": 500
+    },
+    {
+      "epoch": 0.95,
+      "eval_blimp_filtered_avg": 0.5394029850746269,
+      "eval_blimp_filtered_std": 0.005708813575435007,
+      "step": 500
+    },
+    {
+      "epoch": 0.95,
+      "eval_blimp_supplement_avg": 0.4482758620689655,
+      "eval_blimp_supplement_std": 0.022955051141025275,
+      "step": 500
+    },
+    {
+      "epoch": 0.95,
+      "eval_vqa_filtered_avg": 0.12,
+      "eval_vqa_filtered_std": 0.03265986323710906,
+      "step": 500
+    },
+    {
+      "epoch": 0.95,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 500
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.6615663766860962,
+      "learning_rate": 0.0005933337232533938,
+      "loss": 2.2802,
+      "step": 501
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.676198959350586,
+      "learning_rate": 0.0005932949418277106,
+      "loss": 2.5679,
+      "step": 502
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.9866411685943604,
+      "learning_rate": 0.0005932560491972625,
+      "loss": 2.4612,
+      "step": 503
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.7591217756271362,
+      "learning_rate": 0.0005932170453767959,
+      "loss": 2.4013,
+      "step": 504
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.6951096057891846,
+      "learning_rate": 0.0005931779303810994,
+      "loss": 2.609,
+      "step": 505
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.5338921546936035,
+      "learning_rate": 0.0005931387042250036,
+      "loss": 2.6305,
+      "step": 506
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.6677510738372803,
+      "learning_rate": 0.0005930993669233815,
+      "loss": 2.5538,
+      "step": 507
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.7992523908615112,
+      "learning_rate": 0.000593059918491148,
+      "loss": 2.5584,
+      "step": 508
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.349268913269043,
+      "learning_rate": 0.0005930203589432603,
+      "loss": 2.5862,
+      "step": 509
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.917860507965088,
+      "learning_rate": 0.0005929806882947177,
+      "loss": 2.5397,
+      "step": 510
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.723149061203003,
+      "learning_rate": 0.0005929409065605615,
+      "loss": 2.2362,
+      "step": 511
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.723706603050232,
+      "learning_rate": 0.0005929010137558753,
+      "loss": 2.4037,
+      "step": 512
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.6819648742675781,
+      "learning_rate": 0.0005928610098957847,
+      "loss": 2.4483,
+      "step": 513
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.682865023612976,
+      "learning_rate": 0.0005928208949954574,
+      "loss": 2.43,
+      "step": 514
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.90224289894104,
+      "learning_rate": 0.0005927806690701032,
+      "loss": 2.1746,
+      "step": 515
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.0054173469543457,
+      "learning_rate": 0.0005927403321349742,
+      "loss": 2.4629,
+      "step": 516
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.9444178342819214,
+      "learning_rate": 0.0005926998842053642,
+      "loss": 2.5389,
+      "step": 517
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.7960498332977295,
+      "learning_rate": 0.0005926593252966096,
+      "loss": 2.5538,
+      "step": 518
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.9881337881088257,
+      "learning_rate": 0.0005926186554240883,
+      "loss": 2.3529,
+      "step": 519
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.4895299673080444,
+      "learning_rate": 0.0005925778746032206,
+      "loss": 2.1737,
+      "step": 520
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.178257703781128,
+      "learning_rate": 0.000592536982849469,
+      "loss": 2.43,
+      "step": 521
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.8235423564910889,
+      "learning_rate": 0.0005924959801783378,
+      "loss": 2.2806,
+      "step": 522
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.6286402940750122,
+      "learning_rate": 0.0005924548666053734,
+      "loss": 2.5194,
+      "step": 523
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.9562788009643555,
+      "learning_rate": 0.0005924136421461642,
+      "loss": 2.42,
+      "step": 524
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.0953783988952637,
+      "learning_rate": 0.000592372306816341,
+      "loss": 2.3843,
+      "step": 525
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.3756465911865234,
+      "learning_rate": 0.0005923308606315761,
+      "loss": 1.993,
+      "step": 526
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.5545276403427124,
+      "learning_rate": 0.0005922893036075842,
+      "loss": 2.1641,
+      "step": 527
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.271013021469116,
+      "learning_rate": 0.0005922476357601219,
+      "loss": 2.4693,
+      "step": 528
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.8858239650726318,
+      "learning_rate": 0.0005922058571049881,
+      "loss": 2.1877,
+      "step": 529
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.215648651123047,
+      "learning_rate": 0.0005921639676580231,
+      "loss": 2.3902,
+      "step": 530
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.1128978729248047,
+      "learning_rate": 0.0005921219674351096,
+      "loss": 2.1814,
+      "step": 531
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.1224586963653564,
+      "learning_rate": 0.0005920798564521727,
+      "loss": 2.249,
+      "step": 532
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 2.4209649562835693,
+      "learning_rate": 0.0005920376347251785,
+      "loss": 2.4477,
+      "step": 533
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.9727778434753418,
+      "learning_rate": 0.000591995302270136,
+      "loss": 2.2538,
+      "step": 534
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.4296822547912598,
+      "learning_rate": 0.0005919528591030957,
+      "loss": 2.2067,
+      "step": 535
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.4653446674346924,
+      "learning_rate": 0.0005919103052401502,
+      "loss": 2.4744,
+      "step": 536
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.277153730392456,
+      "learning_rate": 0.0005918676406974343,
+      "loss": 2.1974,
+      "step": 537
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.4830920696258545,
+      "learning_rate": 0.0005918248654911242,
+      "loss": 2.2215,
+      "step": 538
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 2.5799460411071777,
+      "learning_rate": 0.0005917819796374387,
+      "loss": 2.4186,
+      "step": 539
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.4727706909179688,
+      "learning_rate": 0.0005917389831526383,
+      "loss": 2.2394,
+      "step": 540
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.127298355102539,
+      "learning_rate": 0.0005916958760530251,
+      "loss": 2.185,
+      "step": 541
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.6543649435043335,
+      "learning_rate": 0.0005916526583549436,
+      "loss": 2.1256,
+      "step": 542
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.316237449645996,
+      "learning_rate": 0.0005916093300747802,
+      "loss": 2.3366,
+      "step": 543
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.8432185649871826,
+      "learning_rate": 0.0005915658912289629,
+      "loss": 2.3063,
+      "step": 544
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.4873580932617188,
+      "learning_rate": 0.000591522341833962,
+      "loss": 2.3054,
+      "step": 545
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.756935715675354,
+      "learning_rate": 0.0005914786819062895,
+      "loss": 2.1754,
+      "step": 546
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.7721258401870728,
+      "learning_rate": 0.0005914349114624994,
+      "loss": 2.1671,
+      "step": 547
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.6546987295150757,
+      "learning_rate": 0.0005913910305191875,
+      "loss": 2.3587,
+      "step": 548
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.7058765888214111,
+      "learning_rate": 0.0005913470390929915,
+      "loss": 2.2504,
+      "step": 549
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.7375134229660034,
+      "learning_rate": 0.0005913029372005913,
+      "loss": 2.2343,
+      "step": 550
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.0255112648010254,
+      "learning_rate": 0.000591258724858708,
+      "loss": 2.1163,
+      "step": 551
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.498658537864685,
+      "learning_rate": 0.0005912144020841054,
+      "loss": 2.1461,
+      "step": 552
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.3310248851776123,
+      "learning_rate": 0.0005911699688935886,
+      "loss": 2.1269,
+      "step": 553
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.549087643623352,
+      "learning_rate": 0.0005911254253040049,
+      "loss": 2.1175,
+      "step": 554
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.0794496536254883,
+      "learning_rate": 0.000591080771332243,
+      "loss": 2.0295,
+      "step": 555
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.8075915575027466,
+      "learning_rate": 0.000591036006995234,
+      "loss": 2.1697,
+      "step": 556
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.9410725831985474,
+      "learning_rate": 0.0005909911323099506,
+      "loss": 1.9723,
+      "step": 557
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.8182506561279297,
+      "learning_rate": 0.0005909461472934073,
+      "loss": 2.2143,
+      "step": 558
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.0438594818115234,
+      "learning_rate": 0.0005909010519626603,
+      "loss": 2.2072,
+      "step": 559
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 2.0529839992523193,
+      "learning_rate": 0.0005908558463348082,
+      "loss": 2.1437,
+      "step": 560
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.7490898370742798,
+      "learning_rate": 0.0005908105304269906,
+      "loss": 2.0292,
+      "step": 561
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 2.2807154655456543,
+      "learning_rate": 0.0005907651042563896,
+      "loss": 1.9722,
+      "step": 562
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.7733213901519775,
+      "learning_rate": 0.0005907195678402287,
+      "loss": 2.17,
+      "step": 563
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.7667044401168823,
+      "learning_rate": 0.0005906739211957734,
+      "loss": 2.117,
+      "step": 564
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.67091703414917,
+      "learning_rate": 0.000590628164340331,
+      "loss": 2.0029,
+      "step": 565
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.9114665985107422,
+      "learning_rate": 0.0005905822972912504,
+      "loss": 2.1066,
+      "step": 566
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.7055772542953491,
+      "learning_rate": 0.0005905363200659225,
+      "loss": 2.2294,
+      "step": 567
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.7549551725387573,
+      "learning_rate": 0.0005904902326817797,
+      "loss": 2.0305,
+      "step": 568
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.5026719570159912,
+      "learning_rate": 0.0005904440351562967,
+      "loss": 2.1197,
+      "step": 569
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.0362727642059326,
+      "learning_rate": 0.0005903977275069892,
+      "loss": 2.2633,
+      "step": 570
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.874635100364685,
+      "learning_rate": 0.0005903513097514153,
+      "loss": 2.0921,
+      "step": 571
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.491645097732544,
+      "learning_rate": 0.0005903047819071745,
+      "loss": 1.9336,
+      "step": 572
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.6624854803085327,
+      "learning_rate": 0.0005902581439919082,
+      "loss": 2.2775,
+      "step": 573
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.667988657951355,
+      "learning_rate": 0.0005902113960232995,
+      "loss": 2.0713,
+      "step": 574
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.856393575668335,
+      "learning_rate": 0.0005901645380190732,
+      "loss": 2.2668,
+      "step": 575
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.678274393081665,
+      "learning_rate": 0.0005901175699969958,
+      "loss": 2.0002,
+      "step": 576
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.8871744871139526,
+      "learning_rate": 0.0005900704919748756,
+      "loss": 1.9265,
+      "step": 577
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.5514973402023315,
+      "learning_rate": 0.0005900233039705624,
+      "loss": 2.2001,
+      "step": 578
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.9351108074188232,
+      "learning_rate": 0.000589976006001948,
+      "loss": 1.925,
+      "step": 579
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.6144050359725952,
+      "learning_rate": 0.0005899285980869658,
+      "loss": 1.9713,
+      "step": 580
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.812211275100708,
+      "learning_rate": 0.0005898810802435906,
+      "loss": 1.8412,
+      "step": 581
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.8437151908874512,
+      "learning_rate": 0.0005898334524898393,
+      "loss": 1.9369,
+      "step": 582
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.7946457862854004,
+      "learning_rate": 0.0005897857148437703,
+      "loss": 2.105,
+      "step": 583
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.8212580680847168,
+      "learning_rate": 0.0005897378673234835,
+      "loss": 1.8664,
+      "step": 584
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.7606055736541748,
+      "learning_rate": 0.0005896899099471208,
+      "loss": 2.1244,
+      "step": 585
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.8027989864349365,
+      "learning_rate": 0.0005896418427328654,
+      "loss": 1.993,
+      "step": 586
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.05260968208313,
+      "learning_rate": 0.0005895936656989424,
+      "loss": 2.1311,
+      "step": 587
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.6644285917282104,
+      "learning_rate": 0.0005895453788636185,
+      "loss": 1.8451,
+      "step": 588
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.240914821624756,
+      "learning_rate": 0.0005894969822452017,
+      "loss": 2.1367,
+      "step": 589
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.7551424503326416,
+      "learning_rate": 0.0005894484758620422,
+      "loss": 1.9738,
+      "step": 590
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.72556471824646,
+      "learning_rate": 0.0005893998597325315,
+      "loss": 1.8682,
+      "step": 591
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.601507306098938,
+      "learning_rate": 0.0005893511338751026,
+      "loss": 1.8536,
+      "step": 592
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.7317227125167847,
+      "learning_rate": 0.0005893022983082303,
+      "loss": 1.9698,
+      "step": 593
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.5000044107437134,
+      "learning_rate": 0.000589253353050431,
+      "loss": 2.1136,
+      "step": 594
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.6505439281463623,
+      "learning_rate": 0.0005892042981202625,
+      "loss": 1.7733,
+      "step": 595
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.6993253231048584,
+      "learning_rate": 0.0005891551335363244,
+      "loss": 1.7377,
+      "step": 596
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.793504238128662,
+      "learning_rate": 0.0005891058593172578,
+      "loss": 2.089,
+      "step": 597
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.7214890718460083,
+      "learning_rate": 0.0005890564754817453,
+      "loss": 1.9702,
+      "step": 598
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.0826170444488525,
+      "learning_rate": 0.0005890069820485112,
+      "loss": 1.8465,
+      "step": 599
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.6649709939956665,
+      "learning_rate": 0.0005889573790363211,
+      "loss": 1.6815,
+      "step": 600
+    },
+    {
+      "epoch": 1.14,
+      "eval_blimp_filtered_avg": 0.5056716417910447,
+      "eval_blimp_filtered_std": 0.0057643169897321255,
+      "step": 600
+    },
+    {
+      "epoch": 1.14,
+      "eval_blimp_supplement_avg": 0.45905172413793105,
+      "eval_blimp_supplement_std": 0.02270930318543937,
+      "step": 600
+    },
+    {
+      "epoch": 1.14,
+      "eval_vqa_filtered_avg": 0.13,
+      "eval_vqa_filtered_std": 0.03379976689896308,
+      "step": 600
+    },
+    {
+      "epoch": 1.14,
+      "eval_winoground_filtered_avg": 0.47,
+      "eval_winoground_filtered_std": 0.05016135580465919,
+      "step": 600
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.9447858333587646,
+      "learning_rate": 0.0005889076664639824,
+      "loss": 1.9387,
+      "step": 601
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 2.009925127029419,
+      "learning_rate": 0.0005888578443503441,
+      "loss": 1.6938,
+      "step": 602
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.8646794557571411,
+      "learning_rate": 0.0005888079127142964,
+      "loss": 1.9457,
+      "step": 603
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.7530806064605713,
+      "learning_rate": 0.0005887578715747713,
+      "loss": 1.9448,
+      "step": 604
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.7625503540039062,
+      "learning_rate": 0.0005887077209507422,
+      "loss": 2.0852,
+      "step": 605
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.042691230773926,
+      "learning_rate": 0.0005886574608612242,
+      "loss": 1.8172,
+      "step": 606
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 2.046309232711792,
+      "learning_rate": 0.0005886070913252735,
+      "loss": 2.034,
+      "step": 607
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.5638140439987183,
+      "learning_rate": 0.0005885566123619883,
+      "loss": 1.77,
+      "step": 608
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.1476330757141113,
+      "learning_rate": 0.0005885060239905079,
+      "loss": 1.951,
+      "step": 609
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.7927054166793823,
+      "learning_rate": 0.0005884553262300132,
+      "loss": 2.1277,
+      "step": 610
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.1922788619995117,
+      "learning_rate": 0.0005884045190997266,
+      "loss": 1.9493,
+      "step": 611
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.5618817806243896,
+      "learning_rate": 0.0005883536026189121,
+      "loss": 1.8649,
+      "step": 612
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.8993922472000122,
+      "learning_rate": 0.0005883025768068748,
+      "loss": 1.6418,
+      "step": 613
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.8117444515228271,
+      "learning_rate": 0.0005882514416829618,
+      "loss": 2.0332,
+      "step": 614
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.0262675285339355,
+      "learning_rate": 0.000588200197266561,
+      "loss": 1.7312,
+      "step": 615
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.7738970518112183,
+      "learning_rate": 0.0005881488435771024,
+      "loss": 2.0616,
+      "step": 616
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.8418501615524292,
+      "learning_rate": 0.0005880973806340568,
+      "loss": 1.7026,
+      "step": 617
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.7538448572158813,
+      "learning_rate": 0.000588045808456937,
+      "loss": 1.8085,
+      "step": 618
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.738259196281433,
+      "learning_rate": 0.0005879941270652965,
+      "loss": 1.7953,
+      "step": 619
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.3097543716430664,
+      "learning_rate": 0.0005879423364787312,
+      "loss": 1.9071,
+      "step": 620
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.995781421661377,
+      "learning_rate": 0.0005878904367168775,
+      "loss": 1.7424,
+      "step": 621
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 2.209105968475342,
+      "learning_rate": 0.0005878384277994135,
+      "loss": 1.7993,
+      "step": 622
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.9073684215545654,
+      "learning_rate": 0.0005877863097460591,
+      "loss": 1.7768,
+      "step": 623
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.183286666870117,
+      "learning_rate": 0.0005877340825765748,
+      "loss": 1.8931,
+      "step": 624
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.8850737810134888,
+      "learning_rate": 0.0005876817463107632,
+      "loss": 1.8865,
+      "step": 625
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.5232343673706055,
+      "learning_rate": 0.0005876293009684678,
+      "loss": 1.9584,
+      "step": 626
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.6266130208969116,
+      "learning_rate": 0.0005875767465695736,
+      "loss": 1.7413,
+      "step": 627
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 2.436488628387451,
+      "learning_rate": 0.0005875240831340069,
+      "loss": 1.923,
+      "step": 628
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.7249596118927002,
+      "learning_rate": 0.0005874713106817357,
+      "loss": 1.7461,
+      "step": 629
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.7619683742523193,
+      "learning_rate": 0.0005874184292327685,
+      "loss": 1.8768,
+      "step": 630
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.5682899951934814,
+      "learning_rate": 0.0005873654388071562,
+      "loss": 1.9814,
+      "step": 631
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.8177456855773926,
+      "learning_rate": 0.0005873123394249901,
+      "loss": 1.7303,
+      "step": 632
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.330430507659912,
+      "learning_rate": 0.0005872591311064034,
+      "loss": 1.8302,
+      "step": 633
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.0047802925109863,
+      "learning_rate": 0.0005872058138715703,
+      "loss": 1.8156,
+      "step": 634
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.597011685371399,
+      "learning_rate": 0.0005871523877407066,
+      "loss": 1.6679,
+      "step": 635
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.7770912647247314,
+      "learning_rate": 0.0005870988527340689,
+      "loss": 1.7251,
+      "step": 636
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.040426015853882,
+      "learning_rate": 0.0005870452088719554,
+      "loss": 1.8129,
+      "step": 637
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.8451988697052002,
+      "learning_rate": 0.0005869914561747056,
+      "loss": 1.9003,
+      "step": 638
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.0378787517547607,
+      "learning_rate": 0.0005869375946627003,
+      "loss": 1.8135,
+      "step": 639
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.675972580909729,
+      "learning_rate": 0.0005868836243563613,
+      "loss": 1.6665,
+      "step": 640
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.6775058507919312,
+      "learning_rate": 0.0005868295452761518,
+      "loss": 1.8969,
+      "step": 641
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.7762056589126587,
+      "learning_rate": 0.0005867753574425764,
+      "loss": 1.7204,
+      "step": 642
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.6625213623046875,
+      "learning_rate": 0.0005867210608761807,
+      "loss": 1.5716,
+      "step": 643
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.6916215419769287,
+      "learning_rate": 0.0005866666555975515,
+      "loss": 1.7451,
+      "step": 644
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.6856456995010376,
+      "learning_rate": 0.0005866121416273172,
+      "loss": 1.6905,
+      "step": 645
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.078390121459961,
+      "learning_rate": 0.000586557518986147,
+      "loss": 1.71,
+      "step": 646
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.7447023391723633,
+      "learning_rate": 0.0005865027876947513,
+      "loss": 1.7179,
+      "step": 647
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.9787840843200684,
+      "learning_rate": 0.000586447947773882,
+      "loss": 1.8664,
+      "step": 648
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.7178069353103638,
+      "learning_rate": 0.0005863929992443321,
+      "loss": 1.9042,
+      "step": 649
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.6683686971664429,
+      "learning_rate": 0.0005863379421269354,
+      "loss": 1.7847,
+      "step": 650
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.5425337553024292,
+      "learning_rate": 0.0005862827764425674,
+      "loss": 1.8211,
+      "step": 651
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.6619621515274048,
+      "learning_rate": 0.0005862275022121446,
+      "loss": 1.7334,
+      "step": 652
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.7427676916122437,
+      "learning_rate": 0.0005861721194566245,
+      "loss": 1.5734,
+      "step": 653
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.7554672956466675,
+      "learning_rate": 0.0005861166281970059,
+      "loss": 1.7707,
+      "step": 654
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.8136885166168213,
+      "learning_rate": 0.0005860610284543287,
+      "loss": 1.7581,
+      "step": 655
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.6977708339691162,
+      "learning_rate": 0.0005860053202496738,
+      "loss": 1.6395,
+      "step": 656
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.860479474067688,
+      "learning_rate": 0.0005859495036041634,
+      "loss": 1.8113,
+      "step": 657
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.5844275951385498,
+      "learning_rate": 0.0005858935785389609,
+      "loss": 1.7982,
+      "step": 658
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.8458678722381592,
+      "learning_rate": 0.0005858375450752704,
+      "loss": 1.7538,
+      "step": 659
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.6878273487091064,
+      "learning_rate": 0.0005857814032343376,
+      "loss": 1.5266,
+      "step": 660
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.7590059041976929,
+      "learning_rate": 0.000585725153037449,
+      "loss": 1.6626,
+      "step": 661
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.7350887060165405,
+      "learning_rate": 0.0005856687945059321,
+      "loss": 1.5478,
+      "step": 662
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.875821828842163,
+      "learning_rate": 0.0005856123276611558,
+      "loss": 1.7859,
+      "step": 663
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.4508082866668701,
+      "learning_rate": 0.0005855557525245298,
+      "loss": 1.5259,
+      "step": 664
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.961963415145874,
+      "learning_rate": 0.000585499069117505,
+      "loss": 1.7115,
+      "step": 665
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.7172682285308838,
+      "learning_rate": 0.0005854422774615733,
+      "loss": 1.8358,
+      "step": 666
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.5986199378967285,
+      "learning_rate": 0.0005853853775782676,
+      "loss": 1.3628,
+      "step": 667
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.8240129947662354,
+      "learning_rate": 0.0005853283694891618,
+      "loss": 1.7186,
+      "step": 668
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.4404698610305786,
+      "learning_rate": 0.000585271253215871,
+      "loss": 1.4845,
+      "step": 669
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.603102207183838,
+      "learning_rate": 0.0005852140287800512,
+      "loss": 1.3568,
+      "step": 670
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.7286044359207153,
+      "learning_rate": 0.0005851566962033996,
+      "loss": 1.6192,
+      "step": 671
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.6913315057754517,
+      "learning_rate": 0.0005850992555076539,
+      "loss": 1.5581,
+      "step": 672
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.0411903858184814,
+      "learning_rate": 0.0005850417067145934,
+      "loss": 1.8472,
+      "step": 673
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.8655729293823242,
+      "learning_rate": 0.000584984049846038,
+      "loss": 1.5206,
+      "step": 674
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.9143017530441284,
+      "learning_rate": 0.0005849262849238487,
+      "loss": 1.801,
+      "step": 675
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.4587645530700684,
+      "learning_rate": 0.0005848684119699275,
+      "loss": 1.3315,
+      "step": 676
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.7619762420654297,
+      "learning_rate": 0.0005848104310062172,
+      "loss": 1.6711,
+      "step": 677
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.7308783531188965,
+      "learning_rate": 0.0005847523420547019,
+      "loss": 1.5605,
+      "step": 678
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 2.0503058433532715,
+      "learning_rate": 0.0005846941451374062,
+      "loss": 1.481,
+      "step": 679
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.9999351501464844,
+      "learning_rate": 0.0005846358402763961,
+      "loss": 1.5323,
+      "step": 680
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.742104411125183,
+      "learning_rate": 0.000584577427493778,
+      "loss": 1.7718,
+      "step": 681
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.8699674606323242,
+      "learning_rate": 0.0005845189068116996,
+      "loss": 1.7174,
+      "step": 682
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.5845983028411865,
+      "learning_rate": 0.0005844602782523495,
+      "loss": 1.4509,
+      "step": 683
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.1740450859069824,
+      "learning_rate": 0.0005844015418379571,
+      "loss": 1.5312,
+      "step": 684
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.7380989789962769,
+      "learning_rate": 0.0005843426975907927,
+      "loss": 1.7394,
+      "step": 685
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.7831321954727173,
+      "learning_rate": 0.0005842837455331673,
+      "loss": 1.4293,
+      "step": 686
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.8860375881195068,
+      "learning_rate": 0.0005842246856874333,
+      "loss": 1.6377,
+      "step": 687
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.7887260913848877,
+      "learning_rate": 0.0005841655180759835,
+      "loss": 1.5806,
+      "step": 688
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.6463288068771362,
+      "learning_rate": 0.0005841062427212516,
+      "loss": 1.5883,
+      "step": 689
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.5919605493545532,
+      "learning_rate": 0.0005840468596457123,
+      "loss": 1.5424,
+      "step": 690
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.6784379482269287,
+      "learning_rate": 0.0005839873688718812,
+      "loss": 1.5734,
+      "step": 691
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.7006253004074097,
+      "learning_rate": 0.0005839277704223145,
+      "loss": 1.525,
+      "step": 692
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.807228922843933,
+      "learning_rate": 0.0005838680643196094,
+      "loss": 1.5677,
+      "step": 693
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.8262131214141846,
+      "learning_rate": 0.0005838082505864041,
+      "loss": 1.613,
+      "step": 694
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.025095224380493,
+      "learning_rate": 0.0005837483292453769,
+      "loss": 1.6565,
+      "step": 695
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.8095041513442993,
+      "learning_rate": 0.0005836883003192478,
+      "loss": 1.675,
+      "step": 696
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.8858921527862549,
+      "learning_rate": 0.0005836281638307771,
+      "loss": 1.4595,
+      "step": 697
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.6791009902954102,
+      "learning_rate": 0.0005835679198027656,
+      "loss": 1.7355,
+      "step": 698
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.7258856296539307,
+      "learning_rate": 0.0005835075682580558,
+      "loss": 1.5058,
+      "step": 699
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.6658928394317627,
+      "learning_rate": 0.00058344710921953,
+      "loss": 1.7049,
+      "step": 700
+    },
+    {
+      "epoch": 1.33,
+      "eval_blimp_filtered_avg": 0.5401492537313433,
+      "eval_blimp_filtered_std": 0.005663800748400189,
+      "step": 700
+    },
+    {
+      "epoch": 1.33,
+      "eval_blimp_supplement_avg": 0.4956896551724138,
+      "eval_blimp_supplement_std": 0.023105411398038053,
+      "step": 700
+    },
+    {
+      "epoch": 1.33,
+      "eval_vqa_filtered_avg": 0.11,
+      "eval_vqa_filtered_std": 0.031446603773522035,
+      "step": 700
+    },
+    {
+      "epoch": 1.33,
+      "eval_winoground_filtered_avg": 0.53,
+      "eval_winoground_filtered_std": 0.0501613558046592,
+      "step": 700
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.7652894258499146,
+      "learning_rate": 0.0005833865427101117,
+      "loss": 1.4171,
+      "step": 701
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 2.1660194396972656,
+      "learning_rate": 0.0005833258687527652,
+      "loss": 1.413,
+      "step": 702
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.6273351907730103,
+      "learning_rate": 0.0005832650873704956,
+      "loss": 1.5129,
+      "step": 703
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.9791346788406372,
+      "learning_rate": 0.0005832041985863481,
+      "loss": 1.625,
+      "step": 704
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.5898704528808594,
+      "learning_rate": 0.0005831432024234096,
+      "loss": 1.426,
+      "step": 705
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.7379738092422485,
+      "learning_rate": 0.0005830820989048069,
+      "loss": 1.3231,
+      "step": 706
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.5745335817337036,
+      "learning_rate": 0.000583020888053708,
+      "loss": 1.3626,
+      "step": 707
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.6166194677352905,
+      "learning_rate": 0.0005829595698933212,
+      "loss": 1.3367,
+      "step": 708
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.8750660419464111,
+      "learning_rate": 0.0005828981444468958,
+      "loss": 1.6249,
+      "step": 709
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.4448094367980957,
+      "learning_rate": 0.0005828366117377218,
+      "loss": 1.4226,
+      "step": 710
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.8090482950210571,
+      "learning_rate": 0.0005827749717891295,
+      "loss": 1.7526,
+      "step": 711
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.5718047618865967,
+      "learning_rate": 0.0005827132246244902,
+      "loss": 1.5566,
+      "step": 712
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.8906807899475098,
+      "learning_rate": 0.000582651370267216,
+      "loss": 1.6645,
+      "step": 713
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.5167371034622192,
+      "learning_rate": 0.000582589408740759,
+      "loss": 1.4568,
+      "step": 714
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.7676844596862793,
+      "learning_rate": 0.0005825273400686127,
+      "loss": 1.5639,
+      "step": 715
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.5925993919372559,
+      "learning_rate": 0.0005824651642743105,
+      "loss": 1.4445,
+      "step": 716
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.9019882678985596,
+      "learning_rate": 0.0005824028813814269,
+      "loss": 1.4837,
+      "step": 717
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.6891939640045166,
+      "learning_rate": 0.000582340491413577,
+      "loss": 1.5874,
+      "step": 718
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.00931715965271,
+      "learning_rate": 0.0005822779943944164,
+      "loss": 1.5913,
+      "step": 719
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.6990654468536377,
+      "learning_rate": 0.0005822153903476411,
+      "loss": 1.4077,
+      "step": 720
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.1771271228790283,
+      "learning_rate": 0.000582152679296988,
+      "loss": 1.7268,
+      "step": 721
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.6966214179992676,
+      "learning_rate": 0.0005820898612662343,
+      "loss": 1.4633,
+      "step": 722
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 2.034675359725952,
+      "learning_rate": 0.0005820269362791979,
+      "loss": 1.5385,
+      "step": 723
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.731055498123169,
+      "learning_rate": 0.0005819639043597372,
+      "loss": 1.5545,
+      "step": 724
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.0864768028259277,
+      "learning_rate": 0.0005819007655317514,
+      "loss": 1.569,
+      "step": 725
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.8034220933914185,
+      "learning_rate": 0.0005818375198191798,
+      "loss": 1.4444,
+      "step": 726
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.9619566202163696,
+      "learning_rate": 0.0005817741672460024,
+      "loss": 1.3472,
+      "step": 727
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.96491277217865,
+      "learning_rate": 0.00058171070783624,
+      "loss": 1.4942,
+      "step": 728
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.8090075254440308,
+      "learning_rate": 0.0005816471416139533,
+      "loss": 1.2992,
+      "step": 729
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.8593988418579102,
+      "learning_rate": 0.0005815834686032443,
+      "loss": 1.3746,
+      "step": 730
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.7140840291976929,
+      "learning_rate": 0.0005815196888282548,
+      "loss": 1.5369,
+      "step": 731
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.9177623987197876,
+      "learning_rate": 0.0005814558023131673,
+      "loss": 1.3386,
+      "step": 732
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.9736742973327637,
+      "learning_rate": 0.0005813918090822049,
+      "loss": 1.4465,
+      "step": 733
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.9763866662979126,
+      "learning_rate": 0.0005813277091596311,
+      "loss": 1.5244,
+      "step": 734
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 2.214632272720337,
+      "learning_rate": 0.0005812635025697498,
+      "loss": 1.4661,
+      "step": 735
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.703674077987671,
+      "learning_rate": 0.0005811991893369052,
+      "loss": 1.3252,
+      "step": 736
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.7595207691192627,
+      "learning_rate": 0.0005811347694854822,
+      "loss": 1.4824,
+      "step": 737
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.9991966485977173,
+      "learning_rate": 0.0005810702430399063,
+      "loss": 1.7967,
+      "step": 738
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.7624640464782715,
+      "learning_rate": 0.0005810056100246426,
+      "loss": 1.3985,
+      "step": 739
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.773283839225769,
+      "learning_rate": 0.0005809408704641975,
+      "loss": 1.3803,
+      "step": 740
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.8449394702911377,
+      "learning_rate": 0.0005808760243831175,
+      "loss": 1.52,
+      "step": 741
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.7400503158569336,
+      "learning_rate": 0.0005808110718059893,
+      "loss": 1.4801,
+      "step": 742
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 2.044431686401367,
+      "learning_rate": 0.00058074601275744,
+      "loss": 1.4396,
+      "step": 743
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.7667288780212402,
+      "learning_rate": 0.0005806808472621374,
+      "loss": 1.4322,
+      "step": 744
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.6776092052459717,
+      "learning_rate": 0.0005806155753447893,
+      "loss": 1.3089,
+      "step": 745
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.7932809591293335,
+      "learning_rate": 0.000580550197030144,
+      "loss": 1.4193,
+      "step": 746
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.5982379913330078,
+      "learning_rate": 0.0005804847123429902,
+      "loss": 1.2657,
+      "step": 747
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.6395643949508667,
+      "learning_rate": 0.0005804191213081569,
+      "loss": 1.3803,
+      "step": 748
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.835571050643921,
+      "learning_rate": 0.0005803534239505132,
+      "loss": 1.6177,
+      "step": 749
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.4878560304641724,
+      "learning_rate": 0.0005802876202949688,
+      "loss": 1.3887,
+      "step": 750
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.5603101253509521,
+      "learning_rate": 0.0005802217103664736,
+      "loss": 1.5145,
+      "step": 751
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.653630018234253,
+      "learning_rate": 0.0005801556941900178,
+      "loss": 1.3068,
+      "step": 752
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 2.114442825317383,
+      "learning_rate": 0.0005800895717906319,
+      "loss": 1.4265,
+      "step": 753
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.5981066226959229,
+      "learning_rate": 0.0005800233431933866,
+      "loss": 1.5053,
+      "step": 754
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.5707093477249146,
+      "learning_rate": 0.000579957008423393,
+      "loss": 1.4594,
+      "step": 755
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.5791939496994019,
+      "learning_rate": 0.0005798905675058023,
+      "loss": 1.2473,
+      "step": 756
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.5193206071853638,
+      "learning_rate": 0.0005798240204658062,
+      "loss": 1.4371,
+      "step": 757
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.8392200469970703,
+      "learning_rate": 0.0005797573673286362,
+      "loss": 1.3457,
+      "step": 758
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.6056214570999146,
+      "learning_rate": 0.0005796906081195645,
+      "loss": 1.3848,
+      "step": 759
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.6999644041061401,
+      "learning_rate": 0.0005796237428639031,
+      "loss": 1.6546,
+      "step": 760
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.6188856363296509,
+      "learning_rate": 0.0005795567715870046,
+      "loss": 1.323,
+      "step": 761
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.5726099014282227,
+      "learning_rate": 0.0005794896943142616,
+      "loss": 1.2033,
+      "step": 762
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.7603769302368164,
+      "learning_rate": 0.0005794225110711068,
+      "loss": 1.5451,
+      "step": 763
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.744202971458435,
+      "learning_rate": 0.0005793552218830132,
+      "loss": 1.3945,
+      "step": 764
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.6027835607528687,
+      "learning_rate": 0.0005792878267754939,
+      "loss": 1.3056,
+      "step": 765
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.9673374891281128,
+      "learning_rate": 0.0005792203257741024,
+      "loss": 1.281,
+      "step": 766
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.631151556968689,
+      "learning_rate": 0.0005791527189044322,
+      "loss": 1.3386,
+      "step": 767
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.7949496507644653,
+      "learning_rate": 0.0005790850061921166,
+      "loss": 1.2982,
+      "step": 768
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.772820234298706,
+      "learning_rate": 0.0005790171876628294,
+      "loss": 1.3876,
+      "step": 769
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.8199372291564941,
+      "learning_rate": 0.0005789492633422847,
+      "loss": 1.3823,
+      "step": 770
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.7258069515228271,
+      "learning_rate": 0.0005788812332562363,
+      "loss": 1.3164,
+      "step": 771
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.8272298574447632,
+      "learning_rate": 0.0005788130974304783,
+      "loss": 1.6272,
+      "step": 772
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.6267541646957397,
+      "learning_rate": 0.0005787448558908448,
+      "loss": 1.2603,
+      "step": 773
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.518512487411499,
+      "learning_rate": 0.00057867650866321,
+      "loss": 1.2181,
+      "step": 774
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.876643419265747,
+      "learning_rate": 0.0005786080557734885,
+      "loss": 1.4285,
+      "step": 775
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.7908297777175903,
+      "learning_rate": 0.0005785394972476343,
+      "loss": 1.5632,
+      "step": 776
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.791141390800476,
+      "learning_rate": 0.000578470833111642,
+      "loss": 1.3773,
+      "step": 777
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.8712741136550903,
+      "learning_rate": 0.0005784020633915462,
+      "loss": 1.4546,
+      "step": 778
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.4962559938430786,
+      "learning_rate": 0.0005783331881134213,
+      "loss": 1.3301,
+      "step": 779
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.8008034229278564,
+      "learning_rate": 0.0005782642073033818,
+      "loss": 1.3758,
+      "step": 780
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.5973652601242065,
+      "learning_rate": 0.0005781951209875821,
+      "loss": 1.2234,
+      "step": 781
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.1171815395355225,
+      "learning_rate": 0.0005781259291922171,
+      "loss": 1.349,
+      "step": 782
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.7217167615890503,
+      "learning_rate": 0.000578056631943521,
+      "loss": 1.2581,
+      "step": 783
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.6320884227752686,
+      "learning_rate": 0.0005779872292677685,
+      "loss": 1.2135,
+      "step": 784
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.931865930557251,
+      "learning_rate": 0.0005779177211912741,
+      "loss": 1.3518,
+      "step": 785
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.7073534727096558,
+      "learning_rate": 0.0005778481077403921,
+      "loss": 1.5228,
+      "step": 786
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.7011784315109253,
+      "learning_rate": 0.0005777783889415172,
+      "loss": 1.5694,
+      "step": 787
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.6956641674041748,
+      "learning_rate": 0.0005777085648210835,
+      "loss": 1.3392,
+      "step": 788
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.0096213817596436,
+      "learning_rate": 0.0005776386354055652,
+      "loss": 1.3177,
+      "step": 789
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.13425350189209,
+      "learning_rate": 0.0005775686007214769,
+      "loss": 1.4432,
+      "step": 790
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.9607853889465332,
+      "learning_rate": 0.0005774984607953724,
+      "loss": 1.3755,
+      "step": 791
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.7971360683441162,
+      "learning_rate": 0.000577428215653846,
+      "loss": 1.2222,
+      "step": 792
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.6023298501968384,
+      "learning_rate": 0.0005773578653235312,
+      "loss": 1.2206,
+      "step": 793
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.7726649045944214,
+      "learning_rate": 0.0005772874098311022,
+      "loss": 1.3942,
+      "step": 794
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.7128273248672485,
+      "learning_rate": 0.0005772168492032726,
+      "loss": 1.2646,
+      "step": 795
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.8491873741149902,
+      "learning_rate": 0.0005771461834667957,
+      "loss": 1.7108,
+      "step": 796
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.67250394821167,
+      "learning_rate": 0.0005770754126484651,
+      "loss": 1.4476,
+      "step": 797
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.4105380773544312,
+      "learning_rate": 0.0005770045367751142,
+      "loss": 1.323,
+      "step": 798
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.444622278213501,
+      "learning_rate": 0.0005769335558736158,
+      "loss": 1.4916,
+      "step": 799
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.4651328325271606,
+      "learning_rate": 0.0005768624699708827,
+      "loss": 1.22,
+      "step": 800
+    },
+    {
+      "epoch": 1.52,
+      "eval_blimp_filtered_avg": 0.5302985074626866,
+      "eval_blimp_filtered_std": 0.005664833607271249,
+      "step": 800
+    },
+    {
+      "epoch": 1.52,
+      "eval_blimp_supplement_avg": 0.5129310344827587,
+      "eval_blimp_supplement_std": 0.023031946714375563,
+      "step": 800
+    },
+    {
+      "epoch": 1.52,
+      "eval_vqa_filtered_avg": 0.11,
+      "eval_vqa_filtered_std": 0.031446603773522035,
+      "step": 800
+    },
+    {
+      "epoch": 1.52,
+      "eval_winoground_filtered_avg": 0.49,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 800
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.6334147453308105,
+      "learning_rate": 0.0005767912790938678,
+      "loss": 1.2612,
+      "step": 801
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.834556221961975,
+      "learning_rate": 0.0005767199832695637,
+      "loss": 1.5583,
+      "step": 802
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.513660192489624,
+      "learning_rate": 0.0005766485825250024,
+      "loss": 1.2651,
+      "step": 803
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.7345637083053589,
+      "learning_rate": 0.0005765770768872563,
+      "loss": 1.4996,
+      "step": 804
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.7554194927215576,
+      "learning_rate": 0.000576505466383437,
+      "loss": 1.5352,
+      "step": 805
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.523672103881836,
+      "learning_rate": 0.0005764337510406962,
+      "loss": 1.2552,
+      "step": 806
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.5473676919937134,
+      "learning_rate": 0.0005763619308862254,
+      "loss": 1.4369,
+      "step": 807
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.6362680196762085,
+      "learning_rate": 0.0005762900059472553,
+      "loss": 1.1811,
+      "step": 808
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.5492461919784546,
+      "learning_rate": 0.0005762179762510572,
+      "loss": 1.1939,
+      "step": 809
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.6109896898269653,
+      "learning_rate": 0.0005761458418249413,
+      "loss": 1.2243,
+      "step": 810
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.8737928867340088,
+      "learning_rate": 0.000576073602696258,
+      "loss": 1.4348,
+      "step": 811
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.6194127798080444,
+      "learning_rate": 0.0005760012588923972,
+      "loss": 1.2056,
+      "step": 812
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.5402929782867432,
+      "learning_rate": 0.0005759288104407887,
+      "loss": 1.2532,
+      "step": 813
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.8927984237670898,
+      "learning_rate": 0.0005758562573689016,
+      "loss": 1.0521,
+      "step": 814
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.4728201627731323,
+      "learning_rate": 0.000575783599704245,
+      "loss": 1.1562,
+      "step": 815
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.9503824710845947,
+      "learning_rate": 0.0005757108374743676,
+      "loss": 1.1742,
+      "step": 816
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.7147475481033325,
+      "learning_rate": 0.0005756379707068576,
+      "loss": 1.0718,
+      "step": 817
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.9215115308761597,
+      "learning_rate": 0.000575564999429343,
+      "loss": 1.0854,
+      "step": 818
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.8600759506225586,
+      "learning_rate": 0.0005754919236694912,
+      "loss": 1.2866,
+      "step": 819
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.5456554889678955,
+      "learning_rate": 0.0005754187434550095,
+      "loss": 1.301,
+      "step": 820
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.5389868021011353,
+      "learning_rate": 0.0005753454588136447,
+      "loss": 1.2732,
+      "step": 821
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.6426384449005127,
+      "learning_rate": 0.0005752720697731832,
+      "loss": 1.0669,
+      "step": 822
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.7796432971954346,
+      "learning_rate": 0.0005751985763614508,
+      "loss": 1.4752,
+      "step": 823
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.7174537181854248,
+      "learning_rate": 0.0005751249786063131,
+      "loss": 1.3495,
+      "step": 824
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.7588887214660645,
+      "learning_rate": 0.0005750512765356753,
+      "loss": 1.3295,
+      "step": 825
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.8343795537948608,
+      "learning_rate": 0.0005749774701774818,
+      "loss": 1.3519,
+      "step": 826
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.4600590467453003,
+      "learning_rate": 0.0005749035595597169,
+      "loss": 1.187,
+      "step": 827
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.5805381536483765,
+      "learning_rate": 0.0005748295447104044,
+      "loss": 1.2419,
+      "step": 828
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.6340856552124023,
+      "learning_rate": 0.0005747554256576075,
+      "loss": 1.2576,
+      "step": 829
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.7864017486572266,
+      "learning_rate": 0.0005746812024294288,
+      "loss": 1.393,
+      "step": 830
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.5507687330245972,
+      "learning_rate": 0.0005746068750540106,
+      "loss": 1.293,
+      "step": 831
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.6027079820632935,
+      "learning_rate": 0.0005745324435595347,
+      "loss": 1.1603,
+      "step": 832
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.576684832572937,
+      "learning_rate": 0.0005744579079742223,
+      "loss": 1.1116,
+      "step": 833
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.6921007633209229,
+      "learning_rate": 0.0005743832683263341,
+      "loss": 1.2363,
+      "step": 834
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.5260730981826782,
+      "learning_rate": 0.0005743085246441701,
+      "loss": 1.0766,
+      "step": 835
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.5014102458953857,
+      "learning_rate": 0.0005742336769560699,
+      "loss": 1.1269,
+      "step": 836
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.6353541612625122,
+      "learning_rate": 0.0005741587252904126,
+      "loss": 1.1373,
+      "step": 837
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.5109964609146118,
+      "learning_rate": 0.0005740836696756166,
+      "loss": 1.2221,
+      "step": 838
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.71192467212677,
+      "learning_rate": 0.0005740085101401396,
+      "loss": 1.3422,
+      "step": 839
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.6052314043045044,
+      "learning_rate": 0.0005739332467124792,
+      "loss": 1.3454,
+      "step": 840
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.7959022521972656,
+      "learning_rate": 0.0005738578794211717,
+      "loss": 1.4163,
+      "step": 841
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.7739613056182861,
+      "learning_rate": 0.0005737824082947933,
+      "loss": 1.2641,
+      "step": 842
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.476391315460205,
+      "learning_rate": 0.0005737068333619592,
+      "loss": 1.1886,
+      "step": 843
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.8103336095809937,
+      "learning_rate": 0.0005736311546513243,
+      "loss": 1.2944,
+      "step": 844
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.690895438194275,
+      "learning_rate": 0.0005735553721915827,
+      "loss": 1.2053,
+      "step": 845
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.4636420011520386,
+      "learning_rate": 0.0005734794860114678,
+      "loss": 1.0772,
+      "step": 846
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.543880581855774,
+      "learning_rate": 0.0005734034961397524,
+      "loss": 1.123,
+      "step": 847
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.8123525381088257,
+      "learning_rate": 0.0005733274026052485,
+      "loss": 1.3029,
+      "step": 848
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.8699501752853394,
+      "learning_rate": 0.0005732512054368074,
+      "loss": 1.0521,
+      "step": 849
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.437688946723938,
+      "learning_rate": 0.00057317490466332,
+      "loss": 1.0255,
+      "step": 850
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.643018126487732,
+      "learning_rate": 0.0005730985003137162,
+      "loss": 1.0833,
+      "step": 851
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.7865830659866333,
+      "learning_rate": 0.0005730219924169652,
+      "loss": 1.1934,
+      "step": 852
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.7060679197311401,
+      "learning_rate": 0.0005729453810020752,
+      "loss": 1.2234,
+      "step": 853
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.6875605583190918,
+      "learning_rate": 0.0005728686660980945,
+      "loss": 1.2598,
+      "step": 854
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.4949389696121216,
+      "learning_rate": 0.0005727918477341096,
+      "loss": 1.3143,
+      "step": 855
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.7253321409225464,
+      "learning_rate": 0.0005727149259392468,
+      "loss": 1.1714,
+      "step": 856
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.5096007585525513,
+      "learning_rate": 0.0005726379007426718,
+      "loss": 1.0687,
+      "step": 857
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.6282732486724854,
+      "learning_rate": 0.0005725607721735888,
+      "loss": 1.1636,
+      "step": 858
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.038818359375,
+      "learning_rate": 0.000572483540261242,
+      "loss": 1.0442,
+      "step": 859
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.9408632516860962,
+      "learning_rate": 0.0005724062050349142,
+      "loss": 1.3383,
+      "step": 860
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.6709492206573486,
+      "learning_rate": 0.0005723287665239277,
+      "loss": 1.3228,
+      "step": 861
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.3585251569747925,
+      "learning_rate": 0.0005722512247576436,
+      "loss": 1.1164,
+      "step": 862
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.5750173330307007,
+      "learning_rate": 0.0005721735797654626,
+      "loss": 1.2084,
+      "step": 863
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.6760964393615723,
+      "learning_rate": 0.0005720958315768242,
+      "loss": 1.4033,
+      "step": 864
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.475570797920227,
+      "learning_rate": 0.0005720179802212073,
+      "loss": 1.0582,
+      "step": 865
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.569371223449707,
+      "learning_rate": 0.0005719400257281298,
+      "loss": 1.2487,
+      "step": 866
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.7434478998184204,
+      "learning_rate": 0.0005718619681271483,
+      "loss": 1.1071,
+      "step": 867
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.576335072517395,
+      "learning_rate": 0.0005717838074478593,
+      "loss": 1.0778,
+      "step": 868
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.954398274421692,
+      "learning_rate": 0.0005717055437198977,
+      "loss": 1.1213,
+      "step": 869
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.624519944190979,
+      "learning_rate": 0.0005716271769729378,
+      "loss": 0.9506,
+      "step": 870
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.509380578994751,
+      "learning_rate": 0.000571548707236693,
+      "loss": 1.0241,
+      "step": 871
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.6489779949188232,
+      "learning_rate": 0.0005714701345409154,
+      "loss": 1.1662,
+      "step": 872
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.6839085817337036,
+      "learning_rate": 0.0005713914589153965,
+      "loss": 1.3767,
+      "step": 873
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.7281668186187744,
+      "learning_rate": 0.0005713126803899667,
+      "loss": 1.1997,
+      "step": 874
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.7581475973129272,
+      "learning_rate": 0.0005712337989944954,
+      "loss": 1.3711,
+      "step": 875
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.644707441329956,
+      "learning_rate": 0.000571154814758891,
+      "loss": 1.2084,
+      "step": 876
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.511725664138794,
+      "learning_rate": 0.000571075727713101,
+      "loss": 1.1737,
+      "step": 877
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.5169894695281982,
+      "learning_rate": 0.0005709965378871116,
+      "loss": 1.1134,
+      "step": 878
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.6067129373550415,
+      "learning_rate": 0.0005709172453109483,
+      "loss": 1.1707,
+      "step": 879
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.6253957748413086,
+      "learning_rate": 0.0005708378500146754,
+      "loss": 1.2326,
+      "step": 880
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.6408096551895142,
+      "learning_rate": 0.0005707583520283962,
+      "loss": 1.1086,
+      "step": 881
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.743558645248413,
+      "learning_rate": 0.0005706787513822528,
+      "loss": 0.9372,
+      "step": 882
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.5785974264144897,
+      "learning_rate": 0.0005705990481064264,
+      "loss": 1.0961,
+      "step": 883
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.618737816810608,
+      "learning_rate": 0.000570519242231137,
+      "loss": 1.0021,
+      "step": 884
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.6257400512695312,
+      "learning_rate": 0.0005704393337866436,
+      "loss": 1.0793,
+      "step": 885
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.7120611667633057,
+      "learning_rate": 0.0005703593228032438,
+      "loss": 1.27,
+      "step": 886
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.4682568311691284,
+      "learning_rate": 0.0005702792093112746,
+      "loss": 0.9779,
+      "step": 887
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.511576771736145,
+      "learning_rate": 0.0005701989933411116,
+      "loss": 1.0601,
+      "step": 888
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.454260230064392,
+      "learning_rate": 0.0005701186749231689,
+      "loss": 1.1716,
+      "step": 889
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.7777795791625977,
+      "learning_rate": 0.0005700382540879,
+      "loss": 1.3095,
+      "step": 890
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.2625861167907715,
+      "learning_rate": 0.000569957730865797,
+      "loss": 0.823,
+      "step": 891
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.5426040887832642,
+      "learning_rate": 0.0005698771052873907,
+      "loss": 1.0472,
+      "step": 892
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.6888166666030884,
+      "learning_rate": 0.0005697963773832511,
+      "loss": 1.1833,
+      "step": 893
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.6085785627365112,
+      "learning_rate": 0.0005697155471839864,
+      "loss": 1.0789,
+      "step": 894
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.505272388458252,
+      "learning_rate": 0.0005696346147202441,
+      "loss": 0.9728,
+      "step": 895
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.7196341753005981,
+      "learning_rate": 0.0005695535800227103,
+      "loss": 1.1674,
+      "step": 896
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.472322940826416,
+      "learning_rate": 0.0005694724431221099,
+      "loss": 0.8621,
+      "step": 897
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.89956796169281,
+      "learning_rate": 0.0005693912040492063,
+      "loss": 1.0774,
+      "step": 898
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.68740975856781,
+      "learning_rate": 0.000569309862834802,
+      "loss": 1.0683,
+      "step": 899
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.5891575813293457,
+      "learning_rate": 0.0005692284195097381,
+      "loss": 1.0154,
+      "step": 900
+    },
+    {
+      "epoch": 1.71,
+      "eval_blimp_filtered_avg": 0.5397014925373135,
+      "eval_blimp_filtered_std": 0.005754275945312061,
+      "step": 900
+    },
+    {
+      "epoch": 1.71,
+      "eval_blimp_supplement_avg": 0.4978448275862069,
+      "eval_blimp_supplement_std": 0.02318452976375872,
+      "step": 900
+    },
+    {
+      "epoch": 1.71,
+      "eval_vqa_filtered_avg": 0.18,
+      "eval_vqa_filtered_std": 0.03861229196653694,
+      "step": 900
+    },
+    {
+      "epoch": 1.71,
+      "eval_winoground_filtered_avg": 0.51,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 900
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.4794012308120728,
+      "learning_rate": 0.0005691468741048942,
+      "loss": 0.9472,
+      "step": 901
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.5077193975448608,
+      "learning_rate": 0.000569065226651189,
+      "loss": 1.0151,
+      "step": 902
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.6121363639831543,
+      "learning_rate": 0.0005689834771795798,
+      "loss": 1.0029,
+      "step": 903
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.5602844953536987,
+      "learning_rate": 0.0005689016257210621,
+      "loss": 1.1585,
+      "step": 904
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.6923860311508179,
+      "learning_rate": 0.0005688196723066707,
+      "loss": 1.0825,
+      "step": 905
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.3946092128753662,
+      "learning_rate": 0.0005687376169674786,
+      "loss": 1.0626,
+      "step": 906
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.5385193824768066,
+      "learning_rate": 0.0005686554597345978,
+      "loss": 1.0719,
+      "step": 907
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.7652556896209717,
+      "learning_rate": 0.0005685732006391786,
+      "loss": 1.0831,
+      "step": 908
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.6178914308547974,
+      "learning_rate": 0.0005684908397124103,
+      "loss": 0.9795,
+      "step": 909
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.4182031154632568,
+      "learning_rate": 0.0005684083769855206,
+      "loss": 0.9397,
+      "step": 910
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.6043729782104492,
+      "learning_rate": 0.0005683258124897756,
+      "loss": 0.9526,
+      "step": 911
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.0052108764648438,
+      "learning_rate": 0.0005682431462564804,
+      "loss": 1.2447,
+      "step": 912
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.8096027374267578,
+      "learning_rate": 0.0005681603783169783,
+      "loss": 1.1329,
+      "step": 913
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.7398737668991089,
+      "learning_rate": 0.0005680775087026513,
+      "loss": 1.2691,
+      "step": 914
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.6498949527740479,
+      "learning_rate": 0.0005679945374449203,
+      "loss": 0.9353,
+      "step": 915
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.3814806938171387,
+      "learning_rate": 0.000567911464575244,
+      "loss": 1.083,
+      "step": 916
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.8439457416534424,
+      "learning_rate": 0.0005678282901251203,
+      "loss": 1.3112,
+      "step": 917
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.5944379568099976,
+      "learning_rate": 0.0005677450141260852,
+      "loss": 1.2795,
+      "step": 918
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.418800711631775,
+      "learning_rate": 0.0005676616366097135,
+      "loss": 0.9465,
+      "step": 919
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.5997589826583862,
+      "learning_rate": 0.0005675781576076183,
+      "loss": 1.0259,
+      "step": 920
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.6024202108383179,
+      "learning_rate": 0.0005674945771514512,
+      "loss": 1.0898,
+      "step": 921
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.5967144966125488,
+      "learning_rate": 0.0005674108952729023,
+      "loss": 1.0492,
+      "step": 922
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.5208961963653564,
+      "learning_rate": 0.0005673271120037002,
+      "loss": 1.0278,
+      "step": 923
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.56473970413208,
+      "learning_rate": 0.0005672432273756118,
+      "loss": 0.8919,
+      "step": 924
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.6778044700622559,
+      "learning_rate": 0.0005671592414204427,
+      "loss": 1.072,
+      "step": 925
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.6013200283050537,
+      "learning_rate": 0.0005670751541700364,
+      "loss": 0.9394,
+      "step": 926
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.8382151126861572,
+      "learning_rate": 0.0005669909656562757,
+      "loss": 1.1387,
+      "step": 927
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.4691205024719238,
+      "learning_rate": 0.0005669066759110807,
+      "loss": 1.0754,
+      "step": 928
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.585595726966858,
+      "learning_rate": 0.0005668222849664107,
+      "loss": 0.8552,
+      "step": 929
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.654751181602478,
+      "learning_rate": 0.000566737792854263,
+      "loss": 1.0109,
+      "step": 930
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.5488654375076294,
+      "learning_rate": 0.0005666531996066736,
+      "loss": 0.9499,
+      "step": 931
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.5710396766662598,
+      "learning_rate": 0.0005665685052557163,
+      "loss": 1.0608,
+      "step": 932
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.5682061910629272,
+      "learning_rate": 0.0005664837098335036,
+      "loss": 0.9284,
+      "step": 933
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.5311015844345093,
+      "learning_rate": 0.0005663988133721863,
+      "loss": 0.8948,
+      "step": 934
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.7017039060592651,
+      "learning_rate": 0.0005663138159039536,
+      "loss": 1.0145,
+      "step": 935
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.5077508687973022,
+      "learning_rate": 0.0005662287174610328,
+      "loss": 0.876,
+      "step": 936
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.6767970323562622,
+      "learning_rate": 0.0005661435180756893,
+      "loss": 1.022,
+      "step": 937
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.676486611366272,
+      "learning_rate": 0.0005660582177802273,
+      "loss": 1.1649,
+      "step": 938
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.5758018493652344,
+      "learning_rate": 0.0005659728166069888,
+      "loss": 0.9846,
+      "step": 939
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.5328469276428223,
+      "learning_rate": 0.0005658873145883544,
+      "loss": 1.1249,
+      "step": 940
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.525166630744934,
+      "learning_rate": 0.0005658017117567428,
+      "loss": 0.8935,
+      "step": 941
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.6155019998550415,
+      "learning_rate": 0.0005657160081446107,
+      "loss": 1.1018,
+      "step": 942
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.8829741477966309,
+      "learning_rate": 0.0005656302037844534,
+      "loss": 0.9688,
+      "step": 943
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.756117820739746,
+      "learning_rate": 0.0005655442987088041,
+      "loss": 0.9702,
+      "step": 944
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.5683172941207886,
+      "learning_rate": 0.0005654582929502342,
+      "loss": 0.9608,
+      "step": 945
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.5200135707855225,
+      "learning_rate": 0.0005653721865413535,
+      "loss": 0.9275,
+      "step": 946
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.6280202865600586,
+      "learning_rate": 0.00056528597951481,
+      "loss": 0.9782,
+      "step": 947
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.404720664024353,
+      "learning_rate": 0.0005651996719032891,
+      "loss": 0.8799,
+      "step": 948
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.5600579977035522,
+      "learning_rate": 0.0005651132637395157,
+      "loss": 0.9371,
+      "step": 949
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.63922119140625,
+      "learning_rate": 0.0005650267550562514,
+      "loss": 0.8885,
+      "step": 950
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.6731245517730713,
+      "learning_rate": 0.0005649401458862969,
+      "loss": 1.0702,
+      "step": 951
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.745483160018921,
+      "learning_rate": 0.0005648534362624905,
+      "loss": 1.0497,
+      "step": 952
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.5604596138000488,
+      "learning_rate": 0.0005647666262177086,
+      "loss": 0.9735,
+      "step": 953
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.7214564085006714,
+      "learning_rate": 0.0005646797157848662,
+      "loss": 1.0737,
+      "step": 954
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.5365569591522217,
+      "learning_rate": 0.0005645927049969157,
+      "loss": 1.0574,
+      "step": 955
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.4952247142791748,
+      "learning_rate": 0.000564505593886848,
+      "loss": 1.0386,
+      "step": 956
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.6169289350509644,
+      "learning_rate": 0.0005644183824876916,
+      "loss": 0.9106,
+      "step": 957
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.4425609111785889,
+      "learning_rate": 0.0005643310708325136,
+      "loss": 0.9421,
+      "step": 958
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.6006109714508057,
+      "learning_rate": 0.0005642436589544188,
+      "loss": 0.9155,
+      "step": 959
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.6475728750228882,
+      "learning_rate": 0.0005641561468865497,
+      "loss": 0.9962,
+      "step": 960
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.4960026741027832,
+      "learning_rate": 0.0005640685346620873,
+      "loss": 0.8887,
+      "step": 961
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.5954712629318237,
+      "learning_rate": 0.0005639808223142504,
+      "loss": 0.9304,
+      "step": 962
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.5579990148544312,
+      "learning_rate": 0.0005638930098762956,
+      "loss": 0.8939,
+      "step": 963
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.550196647644043,
+      "learning_rate": 0.0005638050973815177,
+      "loss": 0.8396,
+      "step": 964
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.7078702449798584,
+      "learning_rate": 0.0005637170848632493,
+      "loss": 0.9452,
+      "step": 965
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.6379247903823853,
+      "learning_rate": 0.0005636289723548608,
+      "loss": 0.9941,
+      "step": 966
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.3526487350463867,
+      "learning_rate": 0.000563540759889761,
+      "loss": 0.9164,
+      "step": 967
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.2389808893203735,
+      "learning_rate": 0.0005634524475013958,
+      "loss": 0.8941,
+      "step": 968
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.538102626800537,
+      "learning_rate": 0.0005633640352232498,
+      "loss": 0.9009,
+      "step": 969
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.5016802549362183,
+      "learning_rate": 0.0005632755230888448,
+      "loss": 0.9711,
+      "step": 970
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.4295623302459717,
+      "learning_rate": 0.0005631869111317412,
+      "loss": 0.8499,
+      "step": 971
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.6166443824768066,
+      "learning_rate": 0.0005630981993855365,
+      "loss": 0.8938,
+      "step": 972
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.6802626848220825,
+      "learning_rate": 0.0005630093878838664,
+      "loss": 1.1007,
+      "step": 973
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.552329659461975,
+      "learning_rate": 0.0005629204766604044,
+      "loss": 0.875,
+      "step": 974
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.5048490762710571,
+      "learning_rate": 0.0005628314657488618,
+      "loss": 0.8505,
+      "step": 975
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.5711601972579956,
+      "learning_rate": 0.0005627423551829878,
+      "loss": 0.9214,
+      "step": 976
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.5058889389038086,
+      "learning_rate": 0.000562653144996569,
+      "loss": 0.8427,
+      "step": 977
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.517235517501831,
+      "learning_rate": 0.0005625638352234304,
+      "loss": 0.8733,
+      "step": 978
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.2079651355743408,
+      "learning_rate": 0.0005624744258974343,
+      "loss": 0.8473,
+      "step": 979
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.4688547849655151,
+      "learning_rate": 0.0005623849170524806,
+      "loss": 1.1064,
+      "step": 980
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.413800835609436,
+      "learning_rate": 0.0005622953087225075,
+      "loss": 0.9456,
+      "step": 981
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.4036400318145752,
+      "learning_rate": 0.0005622056009414903,
+      "loss": 0.8638,
+      "step": 982
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.4577522277832031,
+      "learning_rate": 0.0005621157937434427,
+      "loss": 0.8708,
+      "step": 983
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.5061801671981812,
+      "learning_rate": 0.0005620258871624154,
+      "loss": 0.9946,
+      "step": 984
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.5701953172683716,
+      "learning_rate": 0.0005619358812324972,
+      "loss": 0.913,
+      "step": 985
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.4197837114334106,
+      "learning_rate": 0.0005618457759878144,
+      "loss": 0.9055,
+      "step": 986
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.3408355712890625,
+      "learning_rate": 0.0005617555714625312,
+      "loss": 0.9147,
+      "step": 987
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.628377079963684,
+      "learning_rate": 0.000561665267690849,
+      "loss": 0.9921,
+      "step": 988
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.429315209388733,
+      "learning_rate": 0.0005615748647070072,
+      "loss": 0.9009,
+      "step": 989
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.5553059577941895,
+      "learning_rate": 0.0005614843625452826,
+      "loss": 1.0176,
+      "step": 990
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.3065555095672607,
+      "learning_rate": 0.0005613937612399901,
+      "loss": 1.0126,
+      "step": 991
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.503562092781067,
+      "learning_rate": 0.0005613030608254812,
+      "loss": 0.8647,
+      "step": 992
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.4253453016281128,
+      "learning_rate": 0.0005612122613361461,
+      "loss": 0.9449,
+      "step": 993
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.5055451393127441,
+      "learning_rate": 0.0005611213628064118,
+      "loss": 1.1252,
+      "step": 994
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.343181848526001,
+      "learning_rate": 0.0005610303652707431,
+      "loss": 0.8552,
+      "step": 995
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.4615721702575684,
+      "learning_rate": 0.0005609392687636424,
+      "loss": 0.8936,
+      "step": 996
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.5606213808059692,
+      "learning_rate": 0.0005608480733196495,
+      "loss": 1.0143,
+      "step": 997
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.523314356803894,
+      "learning_rate": 0.0005607567789733418,
+      "loss": 0.8836,
+      "step": 998
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.6402840614318848,
+      "learning_rate": 0.0005606653857593341,
+      "loss": 0.9353,
+      "step": 999
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.6590620279312134,
+      "learning_rate": 0.0005605738937122789,
+      "loss": 1.0266,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9,
+      "eval_blimp_filtered_avg": 0.5425373134328358,
+      "eval_blimp_filtered_std": 0.005753396031287384,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9,
+      "eval_blimp_supplement_avg": 0.5193965517241379,
+      "eval_blimp_supplement_std": 0.022986733912307433,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9,
+      "eval_vqa_filtered_avg": 0.09,
+      "eval_vqa_filtered_std": 0.028762349126466146,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.500929594039917,
+      "learning_rate": 0.0005604823028668659,
+      "loss": 0.9482,
+      "step": 1001
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.5340286493301392,
+      "learning_rate": 0.0005603906132578224,
+      "loss": 1.0413,
+      "step": 1002
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.506769061088562,
+      "learning_rate": 0.0005602988249199132,
+      "loss": 1.0844,
+      "step": 1003
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.534415602684021,
+      "learning_rate": 0.0005602069378879403,
+      "loss": 1.0715,
+      "step": 1004
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.3757232427597046,
+      "learning_rate": 0.0005601149521967437,
+      "loss": 0.8773,
+      "step": 1005
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.4228408336639404,
+      "learning_rate": 0.0005600228678811999,
+      "loss": 0.9313,
+      "step": 1006
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.5590691566467285,
+      "learning_rate": 0.0005599306849762237,
+      "loss": 0.9522,
+      "step": 1007
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.4194802045822144,
+      "learning_rate": 0.0005598384035167666,
+      "loss": 0.8994,
+      "step": 1008
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.5922983884811401,
+      "learning_rate": 0.0005597460235378178,
+      "loss": 0.9922,
+      "step": 1009
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.76673424243927,
+      "learning_rate": 0.0005596535450744038,
+      "loss": 0.9047,
+      "step": 1010
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.2803122997283936,
+      "learning_rate": 0.0005595609681615886,
+      "loss": 0.7374,
+      "step": 1011
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.3341882228851318,
+      "learning_rate": 0.0005594682928344732,
+      "loss": 0.8108,
+      "step": 1012
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.4129725694656372,
+      "learning_rate": 0.000559375519128196,
+      "loss": 0.8621,
+      "step": 1013
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.4827826023101807,
+      "learning_rate": 0.000559282647077933,
+      "loss": 0.723,
+      "step": 1014
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.5681564807891846,
+      "learning_rate": 0.0005591896767188971,
+      "loss": 1.0241,
+      "step": 1015
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.2975112199783325,
+      "learning_rate": 0.0005590966080863388,
+      "loss": 0.8556,
+      "step": 1016
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.5117037296295166,
+      "learning_rate": 0.0005590034412155455,
+      "loss": 0.9118,
+      "step": 1017
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.390901803970337,
+      "learning_rate": 0.0005589101761418422,
+      "loss": 0.8827,
+      "step": 1018
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.5488499402999878,
+      "learning_rate": 0.0005588168129005909,
+      "loss": 0.9712,
+      "step": 1019
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.2405641078948975,
+      "learning_rate": 0.0005587233515271909,
+      "loss": 0.7285,
+      "step": 1020
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.4206037521362305,
+      "learning_rate": 0.0005586297920570789,
+      "loss": 0.9507,
+      "step": 1021
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.5782568454742432,
+      "learning_rate": 0.0005585361345257283,
+      "loss": 0.9026,
+      "step": 1022
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.6711467504501343,
+      "learning_rate": 0.0005584423789686502,
+      "loss": 1.0249,
+      "step": 1023
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.232337474822998,
+      "learning_rate": 0.0005583485254213928,
+      "loss": 0.7286,
+      "step": 1024
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.3306467533111572,
+      "learning_rate": 0.0005582545739195408,
+      "loss": 0.843,
+      "step": 1025
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.3668138980865479,
+      "learning_rate": 0.0005581605244987171,
+      "loss": 0.7302,
+      "step": 1026
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.2047479152679443,
+      "learning_rate": 0.0005580663771945808,
+      "loss": 0.647,
+      "step": 1027
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.6679412126541138,
+      "learning_rate": 0.0005579721320428286,
+      "loss": 1.0161,
+      "step": 1028
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.242721438407898,
+      "learning_rate": 0.0005578777890791943,
+      "loss": 0.7146,
+      "step": 1029
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.425229787826538,
+      "learning_rate": 0.0005577833483394486,
+      "loss": 0.6608,
+      "step": 1030
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.4252395629882812,
+      "learning_rate": 0.0005576888098593992,
+      "loss": 0.7291,
+      "step": 1031
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.3466156721115112,
+      "learning_rate": 0.0005575941736748913,
+      "loss": 0.7975,
+      "step": 1032
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.4100033044815063,
+      "learning_rate": 0.0005574994398218066,
+      "loss": 0.7431,
+      "step": 1033
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.5105019807815552,
+      "learning_rate": 0.0005574046083360642,
+      "loss": 0.791,
+      "step": 1034
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.237671971321106,
+      "learning_rate": 0.0005573096792536201,
+      "loss": 0.6136,
+      "step": 1035
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.4818798303604126,
+      "learning_rate": 0.0005572146526104672,
+      "loss": 0.872,
+      "step": 1036
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.6817600727081299,
+      "learning_rate": 0.0005571195284426357,
+      "loss": 0.9281,
+      "step": 1037
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.6444350481033325,
+      "learning_rate": 0.0005570243067861924,
+      "loss": 0.8239,
+      "step": 1038
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.4234269857406616,
+      "learning_rate": 0.0005569289876772411,
+      "loss": 0.8174,
+      "step": 1039
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.2344282865524292,
+      "learning_rate": 0.0005568335711519231,
+      "loss": 0.6556,
+      "step": 1040
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.4072575569152832,
+      "learning_rate": 0.0005567380572464159,
+      "loss": 0.8416,
+      "step": 1041
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.3910527229309082,
+      "learning_rate": 0.0005566424459969344,
+      "loss": 0.8927,
+      "step": 1042
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.30292809009552,
+      "learning_rate": 0.00055654673743973,
+      "loss": 0.7214,
+      "step": 1043
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.724034309387207,
+      "learning_rate": 0.0005564509316110917,
+      "loss": 0.8434,
+      "step": 1044
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.4499229192733765,
+      "learning_rate": 0.0005563550285473445,
+      "loss": 0.7724,
+      "step": 1045
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.5575714111328125,
+      "learning_rate": 0.0005562590282848509,
+      "loss": 0.8581,
+      "step": 1046
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.6649889945983887,
+      "learning_rate": 0.00055616293086001,
+      "loss": 1.0032,
+      "step": 1047
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.2967489957809448,
+      "learning_rate": 0.0005560667363092578,
+      "loss": 0.6556,
+      "step": 1048
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.3620800971984863,
+      "learning_rate": 0.0005559704446690672,
+      "loss": 0.7714,
+      "step": 1049
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.3426313400268555,
+      "learning_rate": 0.0005558740559759477,
+      "loss": 0.8912,
+      "step": 1050
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.3938499689102173,
+      "learning_rate": 0.0005557775702664456,
+      "loss": 0.8593,
+      "step": 1051
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.2807201147079468,
+      "learning_rate": 0.0005556809875771445,
+      "loss": 0.7515,
+      "step": 1052
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.1795872449874878,
+      "learning_rate": 0.0005555843079446642,
+      "loss": 0.5456,
+      "step": 1053
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.5675437450408936,
+      "learning_rate": 0.0005554875314056614,
+      "loss": 0.7592,
+      "step": 1054
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.3221852779388428,
+      "learning_rate": 0.0005553906579968293,
+      "loss": 0.6605,
+      "step": 1055
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.3823745250701904,
+      "learning_rate": 0.0005552936877548987,
+      "loss": 0.6545,
+      "step": 1056
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.521744728088379,
+      "learning_rate": 0.0005551966207166361,
+      "loss": 0.7787,
+      "step": 1057
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.457723617553711,
+      "learning_rate": 0.0005550994569188452,
+      "loss": 0.7148,
+      "step": 1058
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.454422116279602,
+      "learning_rate": 0.0005550021963983664,
+      "loss": 0.7398,
+      "step": 1059
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.3853087425231934,
+      "learning_rate": 0.0005549048391920767,
+      "loss": 0.6723,
+      "step": 1060
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.5117624998092651,
+      "learning_rate": 0.0005548073853368896,
+      "loss": 0.6725,
+      "step": 1061
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.3345850706100464,
+      "learning_rate": 0.0005547098348697555,
+      "loss": 0.6908,
+      "step": 1062
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.3250091075897217,
+      "learning_rate": 0.0005546121878276613,
+      "loss": 0.6771,
+      "step": 1063
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.3916566371917725,
+      "learning_rate": 0.0005545144442476305,
+      "loss": 0.7199,
+      "step": 1064
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.4373723268508911,
+      "learning_rate": 0.0005544166041667233,
+      "loss": 0.6217,
+      "step": 1065
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.6065548658370972,
+      "learning_rate": 0.0005543186676220364,
+      "loss": 0.7363,
+      "step": 1066
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.3329182863235474,
+      "learning_rate": 0.0005542206346507031,
+      "loss": 0.7273,
+      "step": 1067
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.3518093824386597,
+      "learning_rate": 0.0005541225052898934,
+      "loss": 0.6177,
+      "step": 1068
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.6493172645568848,
+      "learning_rate": 0.0005540242795768134,
+      "loss": 0.7593,
+      "step": 1069
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.2327256202697754,
+      "learning_rate": 0.0005539259575487064,
+      "loss": 0.6004,
+      "step": 1070
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.5543930530548096,
+      "learning_rate": 0.0005538275392428514,
+      "loss": 0.7964,
+      "step": 1071
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.256238341331482,
+      "learning_rate": 0.000553729024696565,
+      "loss": 0.6312,
+      "step": 1072
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.3741647005081177,
+      "learning_rate": 0.0005536304139471992,
+      "loss": 0.7245,
+      "step": 1073
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.381074070930481,
+      "learning_rate": 0.000553531707032143,
+      "loss": 0.73,
+      "step": 1074
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.520234227180481,
+      "learning_rate": 0.0005534329039888219,
+      "loss": 0.7833,
+      "step": 1075
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.6729586124420166,
+      "learning_rate": 0.0005533340048546976,
+      "loss": 0.7881,
+      "step": 1076
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.3037949800491333,
+      "learning_rate": 0.0005532350096672686,
+      "loss": 0.6571,
+      "step": 1077
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.6993945837020874,
+      "learning_rate": 0.0005531359184640694,
+      "loss": 0.7428,
+      "step": 1078
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.5856446027755737,
+      "learning_rate": 0.0005530367312826713,
+      "loss": 0.7501,
+      "step": 1079
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.2828586101531982,
+      "learning_rate": 0.0005529374481606814,
+      "loss": 0.5932,
+      "step": 1080
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.3649215698242188,
+      "learning_rate": 0.000552838069135744,
+      "loss": 0.5958,
+      "step": 1081
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.4890849590301514,
+      "learning_rate": 0.0005527385942455392,
+      "loss": 0.7323,
+      "step": 1082
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.43606698513031,
+      "learning_rate": 0.0005526390235277834,
+      "loss": 0.6429,
+      "step": 1083
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.46323823928833,
+      "learning_rate": 0.0005525393570202298,
+      "loss": 0.6701,
+      "step": 1084
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.5526909828186035,
+      "learning_rate": 0.0005524395947606674,
+      "loss": 0.6929,
+      "step": 1085
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.706939935684204,
+      "learning_rate": 0.0005523397367869217,
+      "loss": 0.675,
+      "step": 1086
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.45332932472229,
+      "learning_rate": 0.0005522397831368549,
+      "loss": 0.6462,
+      "step": 1087
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.2495710849761963,
+      "learning_rate": 0.0005521397338483647,
+      "loss": 0.5538,
+      "step": 1088
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.563302755355835,
+      "learning_rate": 0.0005520395889593855,
+      "loss": 0.6906,
+      "step": 1089
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.3946380615234375,
+      "learning_rate": 0.0005519393485078882,
+      "loss": 0.6928,
+      "step": 1090
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.300888180732727,
+      "learning_rate": 0.0005518390125318794,
+      "loss": 0.5715,
+      "step": 1091
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.5751183032989502,
+      "learning_rate": 0.0005517385810694024,
+      "loss": 0.7182,
+      "step": 1092
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.3841338157653809,
+      "learning_rate": 0.0005516380541585361,
+      "loss": 0.7047,
+      "step": 1093
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.3402959108352661,
+      "learning_rate": 0.0005515374318373963,
+      "loss": 0.6613,
+      "step": 1094
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.3172447681427002,
+      "learning_rate": 0.0005514367141441344,
+      "loss": 0.6001,
+      "step": 1095
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.2702518701553345,
+      "learning_rate": 0.0005513359011169385,
+      "loss": 0.6698,
+      "step": 1096
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.5927789211273193,
+      "learning_rate": 0.0005512349927940322,
+      "loss": 0.8893,
+      "step": 1097
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.4210339784622192,
+      "learning_rate": 0.0005511339892136757,
+      "loss": 0.6202,
+      "step": 1098
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.3698623180389404,
+      "learning_rate": 0.0005510328904141652,
+      "loss": 0.6774,
+      "step": 1099
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.2860921621322632,
+      "learning_rate": 0.0005509316964338331,
+      "loss": 0.6292,
+      "step": 1100
+    },
+    {
+      "epoch": 2.09,
+      "eval_blimp_filtered_avg": 0.5305970149253731,
+      "eval_blimp_filtered_std": 0.005755703075287266,
+      "step": 1100
+    },
+    {
+      "epoch": 2.09,
+      "eval_blimp_supplement_avg": 0.49353448275862066,
+      "eval_blimp_supplement_std": 0.02327750321498575,
+      "step": 1100
+    },
+    {
+      "epoch": 2.09,
+      "eval_vqa_filtered_avg": 0.1,
+      "eval_vqa_filtered_std": 0.030151134457776334,
+      "step": 1100
+    },
+    {
+      "epoch": 2.09,
+      "eval_winoground_filtered_avg": 0.52,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 1100
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.4746211767196655,
+      "learning_rate": 0.0005508304073110476,
+      "loss": 0.7684,
+      "step": 1101
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.3619978427886963,
+      "learning_rate": 0.0005507290230842133,
+      "loss": 0.6025,
+      "step": 1102
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.1793198585510254,
+      "learning_rate": 0.0005506275437917705,
+      "loss": 0.6508,
+      "step": 1103
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.2876158952713013,
+      "learning_rate": 0.0005505259694721962,
+      "loss": 0.6406,
+      "step": 1104
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.2953238487243652,
+      "learning_rate": 0.0005504243001640026,
+      "loss": 0.5844,
+      "step": 1105
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.505699872970581,
+      "learning_rate": 0.0005503225359057384,
+      "loss": 0.6653,
+      "step": 1106
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.4688055515289307,
+      "learning_rate": 0.0005502206767359881,
+      "loss": 0.6914,
+      "step": 1107
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.4894388914108276,
+      "learning_rate": 0.0005501187226933726,
+      "loss": 0.6149,
+      "step": 1108
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.4663147926330566,
+      "learning_rate": 0.0005500166738165483,
+      "loss": 0.6914,
+      "step": 1109
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.1087398529052734,
+      "learning_rate": 0.0005499145301442077,
+      "loss": 0.4279,
+      "step": 1110
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.3234204053878784,
+      "learning_rate": 0.0005498122917150793,
+      "loss": 0.6059,
+      "step": 1111
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.168749451637268,
+      "learning_rate": 0.0005497099585679274,
+      "loss": 0.5741,
+      "step": 1112
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.4145094156265259,
+      "learning_rate": 0.0005496075307415523,
+      "loss": 0.615,
+      "step": 1113
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.6092028617858887,
+      "learning_rate": 0.0005495050082747904,
+      "loss": 0.8013,
+      "step": 1114
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.4376575946807861,
+      "learning_rate": 0.0005494023912065136,
+      "loss": 0.7397,
+      "step": 1115
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.4166585206985474,
+      "learning_rate": 0.0005492996795756301,
+      "loss": 0.6127,
+      "step": 1116
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.1342171430587769,
+      "learning_rate": 0.0005491968734210833,
+      "loss": 0.5774,
+      "step": 1117
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.2208878993988037,
+      "learning_rate": 0.0005490939727818534,
+      "loss": 0.5874,
+      "step": 1118
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.4748659133911133,
+      "learning_rate": 0.0005489909776969553,
+      "loss": 0.765,
+      "step": 1119
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.6184762716293335,
+      "learning_rate": 0.0005488878882054407,
+      "loss": 0.7211,
+      "step": 1120
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.4727346897125244,
+      "learning_rate": 0.0005487847043463964,
+      "loss": 0.6731,
+      "step": 1121
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.5640757083892822,
+      "learning_rate": 0.0005486814261589455,
+      "loss": 0.8533,
+      "step": 1122
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.2640109062194824,
+      "learning_rate": 0.0005485780536822466,
+      "loss": 0.6348,
+      "step": 1123
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.2774038314819336,
+      "learning_rate": 0.0005484745869554937,
+      "loss": 0.5775,
+      "step": 1124
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.301966905593872,
+      "learning_rate": 0.0005483710260179174,
+      "loss": 0.6552,
+      "step": 1125
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.2398481369018555,
+      "learning_rate": 0.0005482673709087832,
+      "loss": 0.6289,
+      "step": 1126
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.198415756225586,
+      "learning_rate": 0.000548163621667393,
+      "loss": 0.562,
+      "step": 1127
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.1996324062347412,
+      "learning_rate": 0.0005480597783330837,
+      "loss": 0.616,
+      "step": 1128
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.2851948738098145,
+      "learning_rate": 0.0005479558409452282,
+      "loss": 0.782,
+      "step": 1129
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.3797651529312134,
+      "learning_rate": 0.0005478518095432352,
+      "loss": 0.7006,
+      "step": 1130
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.279357671737671,
+      "learning_rate": 0.000547747684166549,
+      "loss": 0.6222,
+      "step": 1131
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.416731357574463,
+      "learning_rate": 0.0005476434648546491,
+      "loss": 0.7442,
+      "step": 1132
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.2348054647445679,
+      "learning_rate": 0.0005475391516470514,
+      "loss": 0.5086,
+      "step": 1133
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.2984541654586792,
+      "learning_rate": 0.0005474347445833065,
+      "loss": 0.6019,
+      "step": 1134
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.3380697965621948,
+      "learning_rate": 0.0005473302437030014,
+      "loss": 0.5803,
+      "step": 1135
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.4335356950759888,
+      "learning_rate": 0.0005472256490457583,
+      "loss": 0.5916,
+      "step": 1136
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.5321451425552368,
+      "learning_rate": 0.0005471209606512348,
+      "loss": 0.6852,
+      "step": 1137
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.513339638710022,
+      "learning_rate": 0.0005470161785591242,
+      "loss": 0.6375,
+      "step": 1138
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.4876329898834229,
+      "learning_rate": 0.0005469113028091555,
+      "loss": 0.7227,
+      "step": 1139
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.3559253215789795,
+      "learning_rate": 0.000546806333441093,
+      "loss": 0.6196,
+      "step": 1140
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.2288427352905273,
+      "learning_rate": 0.0005467012704947364,
+      "loss": 0.6596,
+      "step": 1141
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.338293433189392,
+      "learning_rate": 0.0005465961140099213,
+      "loss": 0.704,
+      "step": 1142
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.4050347805023193,
+      "learning_rate": 0.0005464908640265184,
+      "loss": 0.6174,
+      "step": 1143
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.411130428314209,
+      "learning_rate": 0.0005463855205844337,
+      "loss": 0.6876,
+      "step": 1144
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.359480619430542,
+      "learning_rate": 0.0005462800837236092,
+      "loss": 0.7553,
+      "step": 1145
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.5972455739974976,
+      "learning_rate": 0.0005461745534840218,
+      "loss": 0.7495,
+      "step": 1146
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.4160215854644775,
+      "learning_rate": 0.0005460689299056842,
+      "loss": 0.6824,
+      "step": 1147
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.5010898113250732,
+      "learning_rate": 0.0005459632130286441,
+      "loss": 0.5707,
+      "step": 1148
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.5028258562088013,
+      "learning_rate": 0.0005458574028929848,
+      "loss": 0.6581,
+      "step": 1149
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.352087140083313,
+      "learning_rate": 0.0005457514995388249,
+      "loss": 0.8044,
+      "step": 1150
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.4838099479675293,
+      "learning_rate": 0.0005456455030063187,
+      "loss": 0.7756,
+      "step": 1151
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.3315023183822632,
+      "learning_rate": 0.0005455394133356551,
+      "loss": 0.7222,
+      "step": 1152
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.239300012588501,
+      "learning_rate": 0.0005454332305670588,
+      "loss": 0.6984,
+      "step": 1153
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.5473740100860596,
+      "learning_rate": 0.00054532695474079,
+      "loss": 0.7346,
+      "step": 1154
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.1469244956970215,
+      "learning_rate": 0.0005452205858971434,
+      "loss": 0.6031,
+      "step": 1155
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.3548284769058228,
+      "learning_rate": 0.00054511412407645,
+      "loss": 0.8476,
+      "step": 1156
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.3716113567352295,
+      "learning_rate": 0.0005450075693190752,
+      "loss": 0.6699,
+      "step": 1157
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.1049784421920776,
+      "learning_rate": 0.00054490092166542,
+      "loss": 0.5415,
+      "step": 1158
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.2669706344604492,
+      "learning_rate": 0.0005447941811559206,
+      "loss": 0.6661,
+      "step": 1159
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.39290452003479,
+      "learning_rate": 0.0005446873478310484,
+      "loss": 0.5746,
+      "step": 1160
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.3264832496643066,
+      "learning_rate": 0.00054458042173131,
+      "loss": 0.6296,
+      "step": 1161
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.2292811870574951,
+      "learning_rate": 0.0005444734028972472,
+      "loss": 0.6147,
+      "step": 1162
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.3869529962539673,
+      "learning_rate": 0.0005443662913694368,
+      "loss": 0.6414,
+      "step": 1163
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.3488372564315796,
+      "learning_rate": 0.0005442590871884909,
+      "loss": 0.6009,
+      "step": 1164
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.2840557098388672,
+      "learning_rate": 0.0005441517903950567,
+      "loss": 0.6194,
+      "step": 1165
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.1416616439819336,
+      "learning_rate": 0.0005440444010298166,
+      "loss": 0.5808,
+      "step": 1166
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.3573343753814697,
+      "learning_rate": 0.0005439369191334878,
+      "loss": 0.7605,
+      "step": 1167
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.2414463758468628,
+      "learning_rate": 0.000543829344746823,
+      "loss": 0.5501,
+      "step": 1168
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.3555327653884888,
+      "learning_rate": 0.0005437216779106095,
+      "loss": 0.7367,
+      "step": 1169
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.2232248783111572,
+      "learning_rate": 0.0005436139186656702,
+      "loss": 0.6133,
+      "step": 1170
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.0723257064819336,
+      "learning_rate": 0.0005435060670528626,
+      "loss": 0.5076,
+      "step": 1171
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.3257569074630737,
+      "learning_rate": 0.0005433981231130795,
+      "loss": 0.6507,
+      "step": 1172
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.2080302238464355,
+      "learning_rate": 0.0005432900868872484,
+      "loss": 0.591,
+      "step": 1173
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.4123554229736328,
+      "learning_rate": 0.000543181958416332,
+      "loss": 0.6517,
+      "step": 1174
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.2955987453460693,
+      "learning_rate": 0.0005430737377413279,
+      "loss": 0.6278,
+      "step": 1175
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.5045647621154785,
+      "learning_rate": 0.000542965424903269,
+      "loss": 0.6194,
+      "step": 1176
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.2886570692062378,
+      "learning_rate": 0.0005428570199432225,
+      "loss": 0.5613,
+      "step": 1177
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.3222930431365967,
+      "learning_rate": 0.000542748522902291,
+      "loss": 0.4981,
+      "step": 1178
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.1491953134536743,
+      "learning_rate": 0.0005426399338216119,
+      "loss": 0.6021,
+      "step": 1179
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.2761077880859375,
+      "learning_rate": 0.0005425312527423575,
+      "loss": 0.5323,
+      "step": 1180
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.122342586517334,
+      "learning_rate": 0.000542422479705735,
+      "loss": 0.4959,
+      "step": 1181
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.573423981666565,
+      "learning_rate": 0.0005423136147529861,
+      "loss": 0.7718,
+      "step": 1182
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.560260534286499,
+      "learning_rate": 0.0005422046579253882,
+      "loss": 0.6549,
+      "step": 1183
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.2193498611450195,
+      "learning_rate": 0.0005420956092642528,
+      "loss": 0.6112,
+      "step": 1184
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.5002398490905762,
+      "learning_rate": 0.0005419864688109264,
+      "loss": 0.7841,
+      "step": 1185
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.2262381315231323,
+      "learning_rate": 0.0005418772366067903,
+      "loss": 0.5966,
+      "step": 1186
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.3525958061218262,
+      "learning_rate": 0.0005417679126932607,
+      "loss": 0.6687,
+      "step": 1187
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.1568193435668945,
+      "learning_rate": 0.0005416584971117887,
+      "loss": 0.5309,
+      "step": 1188
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.394896149635315,
+      "learning_rate": 0.0005415489899038595,
+      "loss": 0.7551,
+      "step": 1189
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.072039246559143,
+      "learning_rate": 0.0005414393911109939,
+      "loss": 0.6203,
+      "step": 1190
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.4850800037384033,
+      "learning_rate": 0.0005413297007747467,
+      "loss": 0.7457,
+      "step": 1191
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.2290942668914795,
+      "learning_rate": 0.0005412199189367082,
+      "loss": 0.5923,
+      "step": 1192
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.259556770324707,
+      "learning_rate": 0.0005411100456385024,
+      "loss": 0.7688,
+      "step": 1193
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.2056968212127686,
+      "learning_rate": 0.0005410000809217889,
+      "loss": 0.6455,
+      "step": 1194
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.3752925395965576,
+      "learning_rate": 0.0005408900248282613,
+      "loss": 0.716,
+      "step": 1195
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.1796507835388184,
+      "learning_rate": 0.0005407798773996483,
+      "loss": 0.5048,
+      "step": 1196
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.121598720550537,
+      "learning_rate": 0.000540669638677713,
+      "loss": 0.4981,
+      "step": 1197
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.1450914144515991,
+      "learning_rate": 0.0005405593087042532,
+      "loss": 0.592,
+      "step": 1198
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.517426609992981,
+      "learning_rate": 0.000540448887521101,
+      "loss": 0.7208,
+      "step": 1199
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.2737383842468262,
+      "learning_rate": 0.0005403383751701236,
+      "loss": 0.5975,
+      "step": 1200
+    },
+    {
+      "epoch": 2.28,
+      "eval_blimp_filtered_avg": 0.5452238805970149,
+      "eval_blimp_filtered_std": 0.0056909500186820006,
+      "step": 1200
+    },
+    {
+      "epoch": 2.28,
+      "eval_blimp_supplement_avg": 0.5474137931034483,
+      "eval_blimp_supplement_std": 0.02310439530100771,
+      "step": 1200
+    },
+    {
+      "epoch": 2.28,
+      "eval_vqa_filtered_avg": 0.24,
+      "eval_vqa_filtered_std": 0.04292346959909283,
+      "step": 1200
+    },
+    {
+      "epoch": 2.28,
+      "eval_winoground_filtered_avg": 0.47,
+      "eval_winoground_filtered_std": 0.05016135580465919,
+      "step": 1200
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.385967493057251,
+      "learning_rate": 0.0005402277716932226,
+      "loss": 0.7901,
+      "step": 1201
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.0919837951660156,
+      "learning_rate": 0.0005401170771323337,
+      "loss": 0.4993,
+      "step": 1202
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.084858775138855,
+      "learning_rate": 0.0005400062915294278,
+      "loss": 0.5182,
+      "step": 1203
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.5750778913497925,
+      "learning_rate": 0.0005398954149265098,
+      "loss": 0.6832,
+      "step": 1204
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.2105236053466797,
+      "learning_rate": 0.0005397844473656195,
+      "loss": 0.5709,
+      "step": 1205
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.2092056274414062,
+      "learning_rate": 0.0005396733888888309,
+      "loss": 0.6309,
+      "step": 1206
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.1940475702285767,
+      "learning_rate": 0.0005395622395382526,
+      "loss": 0.603,
+      "step": 1207
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.4054979085922241,
+      "learning_rate": 0.0005394509993560275,
+      "loss": 0.5921,
+      "step": 1208
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.1107707023620605,
+      "learning_rate": 0.0005393396683843333,
+      "loss": 0.5075,
+      "step": 1209
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.2244164943695068,
+      "learning_rate": 0.0005392282466653819,
+      "loss": 0.5834,
+      "step": 1210
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.1315029859542847,
+      "learning_rate": 0.0005391167342414194,
+      "loss": 0.4992,
+      "step": 1211
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.4358667135238647,
+      "learning_rate": 0.0005390051311547265,
+      "loss": 0.6411,
+      "step": 1212
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.3313822746276855,
+      "learning_rate": 0.0005388934374476184,
+      "loss": 0.5949,
+      "step": 1213
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.5017824172973633,
+      "learning_rate": 0.0005387816531624445,
+      "loss": 0.7075,
+      "step": 1214
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.2543458938598633,
+      "learning_rate": 0.0005386697783415885,
+      "loss": 0.5962,
+      "step": 1215
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.2398961782455444,
+      "learning_rate": 0.0005385578130274684,
+      "loss": 0.6533,
+      "step": 1216
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.1184738874435425,
+      "learning_rate": 0.000538445757262537,
+      "loss": 0.4338,
+      "step": 1217
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.0197902917861938,
+      "learning_rate": 0.0005383336110892805,
+      "loss": 0.4322,
+      "step": 1218
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.1600701808929443,
+      "learning_rate": 0.0005382213745502203,
+      "loss": 0.6026,
+      "step": 1219
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.23685622215271,
+      "learning_rate": 0.0005381090476879115,
+      "loss": 0.4949,
+      "step": 1220
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.1563807725906372,
+      "learning_rate": 0.0005379966305449434,
+      "loss": 0.6078,
+      "step": 1221
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.2964361906051636,
+      "learning_rate": 0.00053788412316394,
+      "loss": 0.615,
+      "step": 1222
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.3767518997192383,
+      "learning_rate": 0.0005377715255875591,
+      "loss": 0.7617,
+      "step": 1223
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.1916979551315308,
+      "learning_rate": 0.0005376588378584928,
+      "loss": 0.5387,
+      "step": 1224
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.082127571105957,
+      "learning_rate": 0.0005375460600194674,
+      "loss": 0.5346,
+      "step": 1225
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.1842464208602905,
+      "learning_rate": 0.0005374331921132437,
+      "loss": 0.6061,
+      "step": 1226
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.1735645532608032,
+      "learning_rate": 0.0005373202341826159,
+      "loss": 0.556,
+      "step": 1227
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.1136012077331543,
+      "learning_rate": 0.000537207186270413,
+      "loss": 0.5391,
+      "step": 1228
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.0984896421432495,
+      "learning_rate": 0.0005370940484194978,
+      "loss": 0.5005,
+      "step": 1229
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.280604362487793,
+      "learning_rate": 0.0005369808206727674,
+      "loss": 0.6407,
+      "step": 1230
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.123395562171936,
+      "learning_rate": 0.0005368675030731527,
+      "loss": 0.5589,
+      "step": 1231
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.294639229774475,
+      "learning_rate": 0.000536754095663619,
+      "loss": 0.7263,
+      "step": 1232
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.0396147966384888,
+      "learning_rate": 0.0005366405984871653,
+      "loss": 0.4384,
+      "step": 1233
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.0730780363082886,
+      "learning_rate": 0.0005365270115868249,
+      "loss": 0.469,
+      "step": 1234
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.256927490234375,
+      "learning_rate": 0.0005364133350056652,
+      "loss": 0.5712,
+      "step": 1235
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.2224918603897095,
+      "learning_rate": 0.0005362995687867873,
+      "loss": 0.4925,
+      "step": 1236
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.2360129356384277,
+      "learning_rate": 0.0005361857129733263,
+      "loss": 0.623,
+      "step": 1237
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.3014785051345825,
+      "learning_rate": 0.0005360717676084518,
+      "loss": 0.6722,
+      "step": 1238
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.2032703161239624,
+      "learning_rate": 0.0005359577327353664,
+      "loss": 0.6512,
+      "step": 1239
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.3740237951278687,
+      "learning_rate": 0.0005358436083973077,
+      "loss": 0.7723,
+      "step": 1240
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.1594096422195435,
+      "learning_rate": 0.0005357293946375466,
+      "loss": 0.6383,
+      "step": 1241
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.0508567094802856,
+      "learning_rate": 0.0005356150914993878,
+      "loss": 0.4349,
+      "step": 1242
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.1471633911132812,
+      "learning_rate": 0.0005355006990261702,
+      "loss": 0.6055,
+      "step": 1243
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.1722520589828491,
+      "learning_rate": 0.0005353862172612668,
+      "loss": 0.5508,
+      "step": 1244
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.5032809972763062,
+      "learning_rate": 0.0005352716462480837,
+      "loss": 0.5829,
+      "step": 1245
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.3677853345870972,
+      "learning_rate": 0.0005351569860300614,
+      "loss": 0.6974,
+      "step": 1246
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.2743384838104248,
+      "learning_rate": 0.0005350422366506742,
+      "loss": 0.5558,
+      "step": 1247
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.2992634773254395,
+      "learning_rate": 0.0005349273981534301,
+      "loss": 0.5722,
+      "step": 1248
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.1169432401657104,
+      "learning_rate": 0.0005348124705818707,
+      "loss": 0.4629,
+      "step": 1249
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.2705353498458862,
+      "learning_rate": 0.0005346974539795719,
+      "loss": 0.6241,
+      "step": 1250
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.1465766429901123,
+      "learning_rate": 0.0005345823483901427,
+      "loss": 0.5414,
+      "step": 1251
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.0817683935165405,
+      "learning_rate": 0.0005344671538572263,
+      "loss": 0.5596,
+      "step": 1252
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.040462851524353,
+      "learning_rate": 0.0005343518704244996,
+      "loss": 0.5548,
+      "step": 1253
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.1111599206924438,
+      "learning_rate": 0.0005342364981356729,
+      "loss": 0.5445,
+      "step": 1254
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.1635833978652954,
+      "learning_rate": 0.0005341210370344904,
+      "loss": 0.5508,
+      "step": 1255
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.2373753786087036,
+      "learning_rate": 0.0005340054871647303,
+      "loss": 0.5881,
+      "step": 1256
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.1255908012390137,
+      "learning_rate": 0.0005338898485702035,
+      "loss": 0.517,
+      "step": 1257
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.094839096069336,
+      "learning_rate": 0.0005337741212947557,
+      "loss": 0.5842,
+      "step": 1258
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.0670535564422607,
+      "learning_rate": 0.0005336583053822655,
+      "loss": 0.4771,
+      "step": 1259
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.982903003692627,
+      "learning_rate": 0.0005335424008766452,
+      "loss": 0.426,
+      "step": 1260
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.0715211629867554,
+      "learning_rate": 0.0005334264078218408,
+      "loss": 0.6076,
+      "step": 1261
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.1426498889923096,
+      "learning_rate": 0.000533310326261832,
+      "loss": 0.5692,
+      "step": 1262
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.2722489833831787,
+      "learning_rate": 0.0005331941562406318,
+      "loss": 0.6415,
+      "step": 1263
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.3547619581222534,
+      "learning_rate": 0.0005330778978022869,
+      "loss": 0.6101,
+      "step": 1264
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.7743272185325623,
+      "learning_rate": 0.0005329615509908774,
+      "loss": 0.3048,
+      "step": 1265
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.146203875541687,
+      "learning_rate": 0.000532845115850517,
+      "loss": 0.4807,
+      "step": 1266
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.9322463274002075,
+      "learning_rate": 0.0005327285924253529,
+      "loss": 0.4298,
+      "step": 1267
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.202066421508789,
+      "learning_rate": 0.0005326119807595657,
+      "loss": 0.5997,
+      "step": 1268
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.0615713596343994,
+      "learning_rate": 0.0005324952808973695,
+      "loss": 0.509,
+      "step": 1269
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.096771001815796,
+      "learning_rate": 0.0005323784928830121,
+      "loss": 0.5546,
+      "step": 1270
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.1760188341140747,
+      "learning_rate": 0.000532261616760774,
+      "loss": 0.5071,
+      "step": 1271
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.163087248802185,
+      "learning_rate": 0.0005321446525749701,
+      "loss": 0.4641,
+      "step": 1272
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.2960073947906494,
+      "learning_rate": 0.0005320276003699477,
+      "loss": 0.6512,
+      "step": 1273
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.170253872871399,
+      "learning_rate": 0.0005319104601900881,
+      "loss": 0.5024,
+      "step": 1274
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.1258649826049805,
+      "learning_rate": 0.0005317932320798059,
+      "loss": 0.482,
+      "step": 1275
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.1639217138290405,
+      "learning_rate": 0.0005316759160835489,
+      "loss": 0.5342,
+      "step": 1276
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.2060658931732178,
+      "learning_rate": 0.0005315585122457981,
+      "loss": 0.5537,
+      "step": 1277
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.2621166706085205,
+      "learning_rate": 0.0005314410206110681,
+      "loss": 0.6804,
+      "step": 1278
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.1161220073699951,
+      "learning_rate": 0.0005313234412239067,
+      "loss": 0.4184,
+      "step": 1279
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.0188971757888794,
+      "learning_rate": 0.0005312057741288946,
+      "loss": 0.4438,
+      "step": 1280
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.032352328300476,
+      "learning_rate": 0.0005310880193706466,
+      "loss": 0.4266,
+      "step": 1281
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.2344611883163452,
+      "learning_rate": 0.0005309701769938098,
+      "loss": 0.4777,
+      "step": 1282
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.0544072389602661,
+      "learning_rate": 0.0005308522470430651,
+      "loss": 0.4551,
+      "step": 1283
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.2582460641860962,
+      "learning_rate": 0.0005307342295631265,
+      "loss": 0.6192,
+      "step": 1284
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.2455390691757202,
+      "learning_rate": 0.0005306161245987409,
+      "loss": 0.5295,
+      "step": 1285
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.4048810005187988,
+      "learning_rate": 0.0005304979321946887,
+      "loss": 0.6194,
+      "step": 1286
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.1630373001098633,
+      "learning_rate": 0.0005303796523957834,
+      "loss": 0.4879,
+      "step": 1287
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.1699961423873901,
+      "learning_rate": 0.0005302612852468717,
+      "loss": 0.5436,
+      "step": 1288
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.153743028640747,
+      "learning_rate": 0.000530142830792833,
+      "loss": 0.502,
+      "step": 1289
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.2872878313064575,
+      "learning_rate": 0.0005300242890785806,
+      "loss": 0.5633,
+      "step": 1290
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.2654225826263428,
+      "learning_rate": 0.0005299056601490597,
+      "loss": 0.6516,
+      "step": 1291
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.9791042804718018,
+      "learning_rate": 0.0005297869440492499,
+      "loss": 0.4741,
+      "step": 1292
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.0517257452011108,
+      "learning_rate": 0.0005296681408241627,
+      "loss": 0.398,
+      "step": 1293
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.2565898895263672,
+      "learning_rate": 0.0005295492505188437,
+      "loss": 0.5291,
+      "step": 1294
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.226732850074768,
+      "learning_rate": 0.0005294302731783706,
+      "loss": 0.6588,
+      "step": 1295
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.1778227090835571,
+      "learning_rate": 0.0005293112088478543,
+      "loss": 0.503,
+      "step": 1296
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.1466021537780762,
+      "learning_rate": 0.0005291920575724393,
+      "loss": 0.5928,
+      "step": 1297
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.230020523071289,
+      "learning_rate": 0.0005290728193973024,
+      "loss": 0.553,
+      "step": 1298
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.0105340480804443,
+      "learning_rate": 0.0005289534943676535,
+      "loss": 0.4662,
+      "step": 1299
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.1131094694137573,
+      "learning_rate": 0.0005288340825287355,
+      "loss": 0.6029,
+      "step": 1300
+    },
+    {
+      "epoch": 2.47,
+      "eval_blimp_filtered_avg": 0.5376119402985075,
+      "eval_blimp_filtered_std": 0.005809306287326862,
+      "step": 1300
+    },
+    {
+      "epoch": 2.47,
+      "eval_blimp_supplement_avg": 0.5452586206896551,
+      "eval_blimp_supplement_std": 0.02308717836425835,
+      "step": 1300
+    },
+    {
+      "epoch": 2.47,
+      "eval_vqa_filtered_avg": 0.19,
+      "eval_vqa_filtered_std": 0.03942772444036623,
+      "step": 1300
+    },
+    {
+      "epoch": 2.47,
+      "eval_winoground_filtered_avg": 0.49,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 1300
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.9951886534690857,
+      "learning_rate": 0.0005287145839258244,
+      "loss": 0.4088,
+      "step": 1301
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.1819701194763184,
+      "learning_rate": 0.0005285949986042287,
+      "loss": 0.5843,
+      "step": 1302
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.4075337648391724,
+      "learning_rate": 0.00052847532660929,
+      "loss": 0.5526,
+      "step": 1303
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.2643067836761475,
+      "learning_rate": 0.0005283555679863829,
+      "loss": 0.652,
+      "step": 1304
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.2077029943466187,
+      "learning_rate": 0.0005282357227809144,
+      "loss": 0.5723,
+      "step": 1305
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.0924301147460938,
+      "learning_rate": 0.0005281157910383249,
+      "loss": 0.4757,
+      "step": 1306
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.0669763088226318,
+      "learning_rate": 0.0005279957728040874,
+      "loss": 0.3988,
+      "step": 1307
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.2569003105163574,
+      "learning_rate": 0.0005278756681237073,
+      "loss": 0.5513,
+      "step": 1308
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.3680627346038818,
+      "learning_rate": 0.0005277554770427233,
+      "loss": 0.5424,
+      "step": 1309
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.375889778137207,
+      "learning_rate": 0.0005276351996067067,
+      "loss": 0.5572,
+      "step": 1310
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.2472538948059082,
+      "learning_rate": 0.0005275148358612615,
+      "loss": 0.4775,
+      "step": 1311
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.1869252920150757,
+      "learning_rate": 0.0005273943858520242,
+      "loss": 0.5649,
+      "step": 1312
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.256609320640564,
+      "learning_rate": 0.0005272738496246647,
+      "loss": 0.6225,
+      "step": 1313
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.076629877090454,
+      "learning_rate": 0.0005271532272248846,
+      "loss": 0.4852,
+      "step": 1314
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.2317068576812744,
+      "learning_rate": 0.0005270325186984189,
+      "loss": 0.5821,
+      "step": 1315
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.0975897312164307,
+      "learning_rate": 0.0005269117240910352,
+      "loss": 0.58,
+      "step": 1316
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.1005042791366577,
+      "learning_rate": 0.0005267908434485335,
+      "loss": 0.419,
+      "step": 1317
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.0512101650238037,
+      "learning_rate": 0.0005266698768167465,
+      "loss": 0.4869,
+      "step": 1318
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.2176756858825684,
+      "learning_rate": 0.0005265488242415396,
+      "loss": 0.6186,
+      "step": 1319
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.0498640537261963,
+      "learning_rate": 0.0005264276857688104,
+      "loss": 0.4301,
+      "step": 1320
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.9733733534812927,
+      "learning_rate": 0.0005263064614444898,
+      "loss": 0.4079,
+      "step": 1321
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.2924946546554565,
+      "learning_rate": 0.0005261851513145405,
+      "loss": 0.59,
+      "step": 1322
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.1251529455184937,
+      "learning_rate": 0.0005260637554249583,
+      "loss": 0.4381,
+      "step": 1323
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.2370188236236572,
+      "learning_rate": 0.0005259422738217711,
+      "loss": 0.5833,
+      "step": 1324
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.075130581855774,
+      "learning_rate": 0.0005258207065510395,
+      "loss": 0.44,
+      "step": 1325
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.3652069568634033,
+      "learning_rate": 0.0005256990536588567,
+      "loss": 0.539,
+      "step": 1326
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.143554449081421,
+      "learning_rate": 0.000525577315191348,
+      "loss": 0.5505,
+      "step": 1327
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.0044890642166138,
+      "learning_rate": 0.0005254554911946714,
+      "loss": 0.452,
+      "step": 1328
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.132042646408081,
+      "learning_rate": 0.0005253335817150175,
+      "loss": 0.4799,
+      "step": 1329
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.9550961852073669,
+      "learning_rate": 0.0005252115867986089,
+      "loss": 0.3689,
+      "step": 1330
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.3057632446289062,
+      "learning_rate": 0.0005250895064917009,
+      "loss": 0.5586,
+      "step": 1331
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.0486844778060913,
+      "learning_rate": 0.000524967340840581,
+      "loss": 0.4478,
+      "step": 1332
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.0674594640731812,
+      "learning_rate": 0.0005248450898915691,
+      "loss": 0.4853,
+      "step": 1333
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.3522800207138062,
+      "learning_rate": 0.0005247227536910177,
+      "loss": 0.749,
+      "step": 1334
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.070720911026001,
+      "learning_rate": 0.0005246003322853113,
+      "loss": 0.4943,
+      "step": 1335
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.0830868482589722,
+      "learning_rate": 0.0005244778257208667,
+      "loss": 0.4382,
+      "step": 1336
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.070736289024353,
+      "learning_rate": 0.000524355234044133,
+      "loss": 0.5654,
+      "step": 1337
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.121034860610962,
+      "learning_rate": 0.0005242325573015919,
+      "loss": 0.4612,
+      "step": 1338
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.1283791065216064,
+      "learning_rate": 0.0005241097955397572,
+      "loss": 0.5251,
+      "step": 1339
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.2688888311386108,
+      "learning_rate": 0.0005239869488051744,
+      "loss": 0.5584,
+      "step": 1340
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.1030299663543701,
+      "learning_rate": 0.0005238640171444221,
+      "loss": 0.4761,
+      "step": 1341
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.007702112197876,
+      "learning_rate": 0.0005237410006041106,
+      "loss": 0.4721,
+      "step": 1342
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.0394130945205688,
+      "learning_rate": 0.0005236178992308823,
+      "loss": 0.5347,
+      "step": 1343
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.2185136079788208,
+      "learning_rate": 0.0005234947130714121,
+      "loss": 0.4791,
+      "step": 1344
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.0789039134979248,
+      "learning_rate": 0.0005233714421724068,
+      "loss": 0.5747,
+      "step": 1345
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.1003581285476685,
+      "learning_rate": 0.0005232480865806052,
+      "loss": 0.419,
+      "step": 1346
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.9265324473381042,
+      "learning_rate": 0.0005231246463427787,
+      "loss": 0.3143,
+      "step": 1347
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.367985486984253,
+      "learning_rate": 0.0005230011215057303,
+      "loss": 0.637,
+      "step": 1348
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.1648881435394287,
+      "learning_rate": 0.0005228775121162955,
+      "loss": 0.4975,
+      "step": 1349
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.0849450826644897,
+      "learning_rate": 0.0005227538182213414,
+      "loss": 0.4613,
+      "step": 1350
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.2113218307495117,
+      "learning_rate": 0.0005226300398677675,
+      "loss": 0.5404,
+      "step": 1351
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.8685306906700134,
+      "learning_rate": 0.0005225061771025052,
+      "loss": 0.3558,
+      "step": 1352
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.130334496498108,
+      "learning_rate": 0.0005223822299725179,
+      "loss": 0.4739,
+      "step": 1353
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.1690311431884766,
+      "learning_rate": 0.0005222581985248009,
+      "loss": 0.613,
+      "step": 1354
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.0737041234970093,
+      "learning_rate": 0.0005221340828063817,
+      "loss": 0.5115,
+      "step": 1355
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.076979160308838,
+      "learning_rate": 0.0005220098828643196,
+      "loss": 0.5072,
+      "step": 1356
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.988743245601654,
+      "learning_rate": 0.0005218855987457058,
+      "loss": 0.4458,
+      "step": 1357
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.05391526222229,
+      "learning_rate": 0.0005217612304976634,
+      "loss": 0.4059,
+      "step": 1358
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.2042262554168701,
+      "learning_rate": 0.0005216367781673476,
+      "loss": 0.5343,
+      "step": 1359
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.4106839895248413,
+      "learning_rate": 0.0005215122418019452,
+      "loss": 0.5981,
+      "step": 1360
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.2683615684509277,
+      "learning_rate": 0.0005213876214486753,
+      "loss": 0.6263,
+      "step": 1361
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.1339666843414307,
+      "learning_rate": 0.0005212629171547882,
+      "loss": 0.5104,
+      "step": 1362
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.1311287879943848,
+      "learning_rate": 0.0005211381289675666,
+      "loss": 0.6074,
+      "step": 1363
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.9865415096282959,
+      "learning_rate": 0.0005210132569343246,
+      "loss": 0.4276,
+      "step": 1364
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.1249173879623413,
+      "learning_rate": 0.0005208883011024084,
+      "loss": 0.4533,
+      "step": 1365
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.1481037139892578,
+      "learning_rate": 0.000520763261519196,
+      "loss": 0.5756,
+      "step": 1366
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.9990204572677612,
+      "learning_rate": 0.0005206381382320967,
+      "loss": 0.443,
+      "step": 1367
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.186910629272461,
+      "learning_rate": 0.000520512931288552,
+      "loss": 0.5159,
+      "step": 1368
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.1607739925384521,
+      "learning_rate": 0.0005203876407360348,
+      "loss": 0.481,
+      "step": 1369
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.1858322620391846,
+      "learning_rate": 0.0005202622666220502,
+      "loss": 0.5735,
+      "step": 1370
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.0905629396438599,
+      "learning_rate": 0.0005201368089941345,
+      "loss": 0.5768,
+      "step": 1371
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.2370423078536987,
+      "learning_rate": 0.0005200112678998556,
+      "loss": 0.5982,
+      "step": 1372
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.2354854345321655,
+      "learning_rate": 0.0005198856433868135,
+      "loss": 0.5594,
+      "step": 1373
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.974384605884552,
+      "learning_rate": 0.0005197599355026396,
+      "loss": 0.3763,
+      "step": 1374
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.1345148086547852,
+      "learning_rate": 0.0005196341442949967,
+      "loss": 0.6,
+      "step": 1375
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.088945746421814,
+      "learning_rate": 0.0005195082698115795,
+      "loss": 0.405,
+      "step": 1376
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.94474858045578,
+      "learning_rate": 0.0005193823121001142,
+      "loss": 0.4508,
+      "step": 1377
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.9893271327018738,
+      "learning_rate": 0.0005192562712083583,
+      "loss": 0.5154,
+      "step": 1378
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.125919222831726,
+      "learning_rate": 0.0005191301471841014,
+      "loss": 0.5537,
+      "step": 1379
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.077056884765625,
+      "learning_rate": 0.0005190039400751641,
+      "loss": 0.4763,
+      "step": 1380
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.1028717756271362,
+      "learning_rate": 0.0005188776499293987,
+      "loss": 0.551,
+      "step": 1381
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.0959690809249878,
+      "learning_rate": 0.000518751276794689,
+      "loss": 0.4484,
+      "step": 1382
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.0888437032699585,
+      "learning_rate": 0.00051862482071895,
+      "loss": 0.4346,
+      "step": 1383
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.9902840852737427,
+      "learning_rate": 0.0005184982817501287,
+      "loss": 0.4169,
+      "step": 1384
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.9585584402084351,
+      "learning_rate": 0.000518371659936203,
+      "loss": 0.3324,
+      "step": 1385
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.1085971593856812,
+      "learning_rate": 0.0005182449553251826,
+      "loss": 0.5078,
+      "step": 1386
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.0215016603469849,
+      "learning_rate": 0.0005181181679651082,
+      "loss": 0.4014,
+      "step": 1387
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.0104665756225586,
+      "learning_rate": 0.0005179912979040522,
+      "loss": 0.4525,
+      "step": 1388
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.149693489074707,
+      "learning_rate": 0.0005178643451901183,
+      "loss": 0.3899,
+      "step": 1389
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.1196115016937256,
+      "learning_rate": 0.0005177373098714413,
+      "loss": 0.4563,
+      "step": 1390
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.0359539985656738,
+      "learning_rate": 0.0005176101919961878,
+      "loss": 0.4153,
+      "step": 1391
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.0674057006835938,
+      "learning_rate": 0.0005174829916125551,
+      "loss": 0.4806,
+      "step": 1392
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.9393556714057922,
+      "learning_rate": 0.0005173557087687723,
+      "loss": 0.3946,
+      "step": 1393
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.2735085487365723,
+      "learning_rate": 0.0005172283435130994,
+      "loss": 0.4868,
+      "step": 1394
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.124184012413025,
+      "learning_rate": 0.0005171008958938278,
+      "loss": 0.3791,
+      "step": 1395
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.1126409769058228,
+      "learning_rate": 0.0005169733659592802,
+      "loss": 0.4796,
+      "step": 1396
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.178626537322998,
+      "learning_rate": 0.0005168457537578106,
+      "loss": 0.49,
+      "step": 1397
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.9160767793655396,
+      "learning_rate": 0.0005167180593378037,
+      "loss": 0.3083,
+      "step": 1398
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.0219823122024536,
+      "learning_rate": 0.0005165902827476759,
+      "loss": 0.3737,
+      "step": 1399
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.2896318435668945,
+      "learning_rate": 0.0005164624240358745,
+      "loss": 0.5244,
+      "step": 1400
+    },
+    {
+      "epoch": 2.66,
+      "eval_blimp_filtered_avg": 0.5405970149253732,
+      "eval_blimp_filtered_std": 0.005795345317455896,
+      "step": 1400
+    },
+    {
+      "epoch": 2.66,
+      "eval_blimp_supplement_avg": 0.5064655172413793,
+      "eval_blimp_supplement_std": 0.023232073557745602,
+      "step": 1400
+    },
+    {
+      "epoch": 2.66,
+      "eval_vqa_filtered_avg": 0.13,
+      "eval_vqa_filtered_std": 0.03379976689896309,
+      "step": 1400
+    },
+    {
+      "epoch": 2.66,
+      "eval_winoground_filtered_avg": 0.53,
+      "eval_winoground_filtered_std": 0.05016135580465919,
+      "step": 1400
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.0715683698654175,
+      "learning_rate": 0.000516334483250878,
+      "loss": 0.508,
+      "step": 1401
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.0437368154525757,
+      "learning_rate": 0.0005162064604411961,
+      "loss": 0.5164,
+      "step": 1402
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.0643526315689087,
+      "learning_rate": 0.0005160783556553695,
+      "loss": 0.3766,
+      "step": 1403
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.2057632207870483,
+      "learning_rate": 0.0005159501689419699,
+      "loss": 0.509,
+      "step": 1404
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.0900951623916626,
+      "learning_rate": 0.0005158219003496003,
+      "loss": 0.4274,
+      "step": 1405
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.0064342021942139,
+      "learning_rate": 0.0005156935499268944,
+      "loss": 0.4209,
+      "step": 1406
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.2086085081100464,
+      "learning_rate": 0.0005155651177225172,
+      "loss": 0.4529,
+      "step": 1407
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.3940248489379883,
+      "learning_rate": 0.0005154366037851649,
+      "loss": 0.6298,
+      "step": 1408
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.2188093662261963,
+      "learning_rate": 0.000515308008163564,
+      "loss": 0.5654,
+      "step": 1409
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.1258479356765747,
+      "learning_rate": 0.0005151793309064727,
+      "loss": 0.4146,
+      "step": 1410
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.163074254989624,
+      "learning_rate": 0.0005150505720626799,
+      "loss": 0.429,
+      "step": 1411
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.072563648223877,
+      "learning_rate": 0.0005149217316810052,
+      "loss": 0.493,
+      "step": 1412
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.1860970258712769,
+      "learning_rate": 0.0005147928098102992,
+      "loss": 0.5811,
+      "step": 1413
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.0305029153823853,
+      "learning_rate": 0.0005146638064994439,
+      "loss": 0.3879,
+      "step": 1414
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.9886533617973328,
+      "learning_rate": 0.0005145347217973515,
+      "loss": 0.5024,
+      "step": 1415
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.1188766956329346,
+      "learning_rate": 0.0005144055557529654,
+      "loss": 0.4806,
+      "step": 1416
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.9852072596549988,
+      "learning_rate": 0.0005142763084152596,
+      "loss": 0.4616,
+      "step": 1417
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.139087438583374,
+      "learning_rate": 0.0005141469798332394,
+      "loss": 0.5785,
+      "step": 1418
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.1665657758712769,
+      "learning_rate": 0.0005140175700559406,
+      "loss": 0.5023,
+      "step": 1419
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.0724141597747803,
+      "learning_rate": 0.0005138880791324294,
+      "loss": 0.4665,
+      "step": 1420
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.0219131708145142,
+      "learning_rate": 0.0005137585071118036,
+      "loss": 0.4089,
+      "step": 1421
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.1368484497070312,
+      "learning_rate": 0.0005136288540431912,
+      "loss": 0.5184,
+      "step": 1422
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.9532016515731812,
+      "learning_rate": 0.000513499119975751,
+      "loss": 0.3785,
+      "step": 1423
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.1011239290237427,
+      "learning_rate": 0.0005133693049586724,
+      "loss": 0.3792,
+      "step": 1424
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.004177451133728,
+      "learning_rate": 0.0005132394090411759,
+      "loss": 0.4119,
+      "step": 1425
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.067179799079895,
+      "learning_rate": 0.0005131094322725124,
+      "loss": 0.4499,
+      "step": 1426
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.237831950187683,
+      "learning_rate": 0.0005129793747019631,
+      "loss": 0.4918,
+      "step": 1427
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.0732711553573608,
+      "learning_rate": 0.0005128492363788407,
+      "loss": 0.529,
+      "step": 1428
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.0348151922225952,
+      "learning_rate": 0.0005127190173524876,
+      "loss": 0.4116,
+      "step": 1429
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.0614460706710815,
+      "learning_rate": 0.0005125887176722776,
+      "loss": 0.3724,
+      "step": 1430
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.1131497621536255,
+      "learning_rate": 0.0005124583373876145,
+      "loss": 0.3792,
+      "step": 1431
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.24030339717865,
+      "learning_rate": 0.000512327876547933,
+      "loss": 0.5806,
+      "step": 1432
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.0023142099380493,
+      "learning_rate": 0.000512197335202698,
+      "loss": 0.4133,
+      "step": 1433
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.9921138286590576,
+      "learning_rate": 0.0005120667134014053,
+      "loss": 0.3284,
+      "step": 1434
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.885846734046936,
+      "learning_rate": 0.0005119360111935811,
+      "loss": 0.3422,
+      "step": 1435
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.2196145057678223,
+      "learning_rate": 0.0005118052286287818,
+      "loss": 0.4763,
+      "step": 1436
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.9756330251693726,
+      "learning_rate": 0.0005116743657565947,
+      "loss": 0.4124,
+      "step": 1437
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.0042308568954468,
+      "learning_rate": 0.0005115434226266372,
+      "loss": 0.3852,
+      "step": 1438
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.178466796875,
+      "learning_rate": 0.0005114123992885573,
+      "loss": 0.6433,
+      "step": 1439
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.9289754629135132,
+      "learning_rate": 0.0005112812957920336,
+      "loss": 0.3858,
+      "step": 1440
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.1588973999023438,
+      "learning_rate": 0.0005111501121867744,
+      "loss": 0.5331,
+      "step": 1441
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.0232298374176025,
+      "learning_rate": 0.0005110188485225194,
+      "loss": 0.5275,
+      "step": 1442
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.9754258990287781,
+      "learning_rate": 0.000510887504849038,
+      "loss": 0.3746,
+      "step": 1443
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.0055285692214966,
+      "learning_rate": 0.0005107560812161298,
+      "loss": 0.5236,
+      "step": 1444
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.0718809366226196,
+      "learning_rate": 0.000510624577673625,
+      "loss": 0.4578,
+      "step": 1445
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.9894866347312927,
+      "learning_rate": 0.0005104929942713842,
+      "loss": 0.4362,
+      "step": 1446
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.0897440910339355,
+      "learning_rate": 0.0005103613310592979,
+      "loss": 0.4694,
+      "step": 1447
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.0016191005706787,
+      "learning_rate": 0.0005102295880872875,
+      "loss": 0.442,
+      "step": 1448
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.2285269498825073,
+      "learning_rate": 0.000510097765405304,
+      "loss": 0.49,
+      "step": 1449
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.2317713499069214,
+      "learning_rate": 0.0005099658630633287,
+      "loss": 0.5532,
+      "step": 1450
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.8522248268127441,
+      "learning_rate": 0.0005098338811113735,
+      "loss": 0.2958,
+      "step": 1451
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.16655695438385,
+      "learning_rate": 0.0005097018195994803,
+      "loss": 0.604,
+      "step": 1452
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.9830549955368042,
+      "learning_rate": 0.000509569678577721,
+      "loss": 0.4265,
+      "step": 1453
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.016083836555481,
+      "learning_rate": 0.0005094374580961976,
+      "loss": 0.451,
+      "step": 1454
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.9253076314926147,
+      "learning_rate": 0.0005093051582050429,
+      "loss": 0.3722,
+      "step": 1455
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.9553015828132629,
+      "learning_rate": 0.0005091727789544188,
+      "loss": 0.4125,
+      "step": 1456
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.9619435667991638,
+      "learning_rate": 0.0005090403203945179,
+      "loss": 0.3693,
+      "step": 1457
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.0178430080413818,
+      "learning_rate": 0.0005089077825755628,
+      "loss": 0.396,
+      "step": 1458
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.1068609952926636,
+      "learning_rate": 0.0005087751655478062,
+      "loss": 0.4162,
+      "step": 1459
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.0717236995697021,
+      "learning_rate": 0.0005086424693615307,
+      "loss": 0.453,
+      "step": 1460
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.0057470798492432,
+      "learning_rate": 0.0005085096940670488,
+      "loss": 0.3607,
+      "step": 1461
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.2476129531860352,
+      "learning_rate": 0.0005083768397147034,
+      "loss": 0.4377,
+      "step": 1462
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.0154598951339722,
+      "learning_rate": 0.0005082439063548669,
+      "loss": 0.3065,
+      "step": 1463
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.1665611267089844,
+      "learning_rate": 0.0005081108940379422,
+      "loss": 0.4433,
+      "step": 1464
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.0628063678741455,
+      "learning_rate": 0.0005079778028143615,
+      "loss": 0.4291,
+      "step": 1465
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.9196938872337341,
+      "learning_rate": 0.0005078446327345874,
+      "loss": 0.3762,
+      "step": 1466
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.8761244416236877,
+      "learning_rate": 0.0005077113838491122,
+      "loss": 0.3194,
+      "step": 1467
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.0546696186065674,
+      "learning_rate": 0.000507578056208458,
+      "loss": 0.5058,
+      "step": 1468
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.2443108558654785,
+      "learning_rate": 0.0005074446498631772,
+      "loss": 0.5795,
+      "step": 1469
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.8759688138961792,
+      "learning_rate": 0.0005073111648638514,
+      "loss": 0.4088,
+      "step": 1470
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.9856649041175842,
+      "learning_rate": 0.0005071776012610924,
+      "loss": 0.5128,
+      "step": 1471
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.0559920072555542,
+      "learning_rate": 0.0005070439591055418,
+      "loss": 0.4478,
+      "step": 1472
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.155787706375122,
+      "learning_rate": 0.0005069102384478711,
+      "loss": 0.344,
+      "step": 1473
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.0607080459594727,
+      "learning_rate": 0.0005067764393387811,
+      "loss": 0.3984,
+      "step": 1474
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.9042722582817078,
+      "learning_rate": 0.0005066425618290028,
+      "loss": 0.3494,
+      "step": 1475
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.1479640007019043,
+      "learning_rate": 0.0005065086059692967,
+      "loss": 0.5397,
+      "step": 1476
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.8420533537864685,
+      "learning_rate": 0.0005063745718104531,
+      "loss": 0.2975,
+      "step": 1477
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.1782907247543335,
+      "learning_rate": 0.0005062404594032919,
+      "loss": 0.5916,
+      "step": 1478
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.1028553247451782,
+      "learning_rate": 0.000506106268798663,
+      "loss": 0.4819,
+      "step": 1479
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.0908656120300293,
+      "learning_rate": 0.0005059720000474453,
+      "loss": 0.4708,
+      "step": 1480
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.0836502313613892,
+      "learning_rate": 0.0005058376532005479,
+      "loss": 0.475,
+      "step": 1481
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.951991856098175,
+      "learning_rate": 0.0005057032283089094,
+      "loss": 0.3262,
+      "step": 1482
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.9690190553665161,
+      "learning_rate": 0.0005055687254234976,
+      "loss": 0.3064,
+      "step": 1483
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.9460926651954651,
+      "learning_rate": 0.0005054341445953105,
+      "loss": 0.3726,
+      "step": 1484
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.9183267951011658,
+      "learning_rate": 0.0005052994858753752,
+      "loss": 0.3139,
+      "step": 1485
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.0804426670074463,
+      "learning_rate": 0.0005051647493147484,
+      "loss": 0.4552,
+      "step": 1486
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.1017197370529175,
+      "learning_rate": 0.0005050299349645166,
+      "loss": 0.5211,
+      "step": 1487
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.075170636177063,
+      "learning_rate": 0.0005048950428757954,
+      "loss": 0.5257,
+      "step": 1488
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.9420687556266785,
+      "learning_rate": 0.00050476007309973,
+      "loss": 0.4461,
+      "step": 1489
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.0037490129470825,
+      "learning_rate": 0.0005046250256874952,
+      "loss": 0.4306,
+      "step": 1490
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.0218480825424194,
+      "learning_rate": 0.0005044899006902951,
+      "loss": 0.4008,
+      "step": 1491
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.9720267653465271,
+      "learning_rate": 0.0005043546981593635,
+      "loss": 0.4205,
+      "step": 1492
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.05379319190979,
+      "learning_rate": 0.0005042194181459629,
+      "loss": 0.5932,
+      "step": 1493
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.0265417098999023,
+      "learning_rate": 0.0005040840607013861,
+      "loss": 0.3909,
+      "step": 1494
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.8825443387031555,
+      "learning_rate": 0.0005039486258769544,
+      "loss": 0.3901,
+      "step": 1495
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.8760673999786377,
+      "learning_rate": 0.000503813113724019,
+      "loss": 0.3102,
+      "step": 1496
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.1905553340911865,
+      "learning_rate": 0.0005036775242939604,
+      "loss": 0.4153,
+      "step": 1497
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.9693546295166016,
+      "learning_rate": 0.000503541857638188,
+      "loss": 0.4161,
+      "step": 1498
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.172996997833252,
+      "learning_rate": 0.0005034061138081407,
+      "loss": 0.3903,
+      "step": 1499
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.1781742572784424,
+      "learning_rate": 0.0005032702928552871,
+      "loss": 0.4641,
+      "step": 1500
+    },
+    {
+      "epoch": 2.85,
+      "eval_blimp_filtered_avg": 0.5392537313432836,
+      "eval_blimp_filtered_std": 0.005798780505038517,
+      "step": 1500
+    },
+    {
+      "epoch": 2.85,
+      "eval_blimp_supplement_avg": 0.5172413793103449,
+      "eval_blimp_supplement_std": 0.023293130692496913,
+      "step": 1500
+    },
+    {
+      "epoch": 2.85,
+      "eval_vqa_filtered_avg": 0.2,
+      "eval_vqa_filtered_std": 0.04020151261036845,
+      "step": 1500
+    },
+    {
+      "epoch": 2.85,
+      "eval_winoground_filtered_avg": 0.51,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 1500
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.0768730640411377,
+      "learning_rate": 0.0005031343948311243,
+      "loss": 0.3917,
+      "step": 1501
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.0366015434265137,
+      "learning_rate": 0.0005029984197871789,
+      "loss": 0.3636,
+      "step": 1502
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.0964019298553467,
+      "learning_rate": 0.000502862367775007,
+      "loss": 0.4847,
+      "step": 1503
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.1319012641906738,
+      "learning_rate": 0.0005027262388461934,
+      "loss": 0.4651,
+      "step": 1504
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.0217337608337402,
+      "learning_rate": 0.0005025900330523525,
+      "loss": 0.4535,
+      "step": 1505
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.0546666383743286,
+      "learning_rate": 0.0005024537504451276,
+      "loss": 0.5137,
+      "step": 1506
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.8207849860191345,
+      "learning_rate": 0.0005023173910761912,
+      "loss": 0.3315,
+      "step": 1507
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.96149080991745,
+      "learning_rate": 0.0005021809549972448,
+      "loss": 0.3469,
+      "step": 1508
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.8568252325057983,
+      "learning_rate": 0.000502044442260019,
+      "loss": 0.3674,
+      "step": 1509
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.9629746079444885,
+      "learning_rate": 0.0005019078529162736,
+      "loss": 0.4451,
+      "step": 1510
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.9299527406692505,
+      "learning_rate": 0.0005017711870177975,
+      "loss": 0.3066,
+      "step": 1511
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.1167079210281372,
+      "learning_rate": 0.0005016344446164084,
+      "loss": 0.3823,
+      "step": 1512
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.0323776006698608,
+      "learning_rate": 0.0005014976257639532,
+      "loss": 0.356,
+      "step": 1513
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.110533356666565,
+      "learning_rate": 0.0005013607305123075,
+      "loss": 0.4155,
+      "step": 1514
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.0767037868499756,
+      "learning_rate": 0.0005012237589133762,
+      "loss": 0.4301,
+      "step": 1515
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.0523003339767456,
+      "learning_rate": 0.000501086711019093,
+      "loss": 0.4551,
+      "step": 1516
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.3244515657424927,
+      "learning_rate": 0.0005009495868814205,
+      "loss": 0.4623,
+      "step": 1517
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.9882186651229858,
+      "learning_rate": 0.0005008123865523502,
+      "loss": 0.4582,
+      "step": 1518
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.9676164388656616,
+      "learning_rate": 0.0005006751100839026,
+      "loss": 0.4132,
+      "step": 1519
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.9180042743682861,
+      "learning_rate": 0.0005005377575281271,
+      "loss": 0.4386,
+      "step": 1520
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.9433060884475708,
+      "learning_rate": 0.0005004003289371016,
+      "loss": 0.3941,
+      "step": 1521
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.8789255023002625,
+      "learning_rate": 0.0005002628243629334,
+      "loss": 0.2975,
+      "step": 1522
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.9058659076690674,
+      "learning_rate": 0.000500125243857758,
+      "loss": 0.3668,
+      "step": 1523
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.021515965461731,
+      "learning_rate": 0.0004999875874737402,
+      "loss": 0.3962,
+      "step": 1524
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.9307779669761658,
+      "learning_rate": 0.0004998498552630733,
+      "loss": 0.382,
+      "step": 1525
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.952597975730896,
+      "learning_rate": 0.0004997120472779793,
+      "loss": 0.3728,
+      "step": 1526
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.9721535444259644,
+      "learning_rate": 0.000499574163570709,
+      "loss": 0.4308,
+      "step": 1527
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.9406330585479736,
+      "learning_rate": 0.0004994362041935422,
+      "loss": 0.4022,
+      "step": 1528
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.2060368061065674,
+      "learning_rate": 0.0004992981691987868,
+      "loss": 0.5009,
+      "step": 1529
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.064517855644226,
+      "learning_rate": 0.0004991600586387799,
+      "loss": 0.4996,
+      "step": 1530
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.8824906945228577,
+      "learning_rate": 0.0004990218725658872,
+      "loss": 0.2969,
+      "step": 1531
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.1503268480300903,
+      "learning_rate": 0.0004988836110325025,
+      "loss": 0.4476,
+      "step": 1532
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.0337167978286743,
+      "learning_rate": 0.0004987452740910489,
+      "loss": 0.4636,
+      "step": 1533
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.9304941892623901,
+      "learning_rate": 0.0004986068617939776,
+      "loss": 0.4012,
+      "step": 1534
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.9859055876731873,
+      "learning_rate": 0.0004984683741937688,
+      "loss": 0.4213,
+      "step": 1535
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.8300942182540894,
+      "learning_rate": 0.0004983298113429308,
+      "loss": 0.3333,
+      "step": 1536
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.0503365993499756,
+      "learning_rate": 0.0004981911732940007,
+      "loss": 0.5608,
+      "step": 1537
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.001155972480774,
+      "learning_rate": 0.0004980524600995441,
+      "loss": 0.3856,
+      "step": 1538
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.083827257156372,
+      "learning_rate": 0.0004979136718121551,
+      "loss": 0.4436,
+      "step": 1539
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.0124115943908691,
+      "learning_rate": 0.000497774808484456,
+      "loss": 0.3436,
+      "step": 1540
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.9383167028427124,
+      "learning_rate": 0.000497635870169098,
+      "loss": 0.5156,
+      "step": 1541
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.9876752495765686,
+      "learning_rate": 0.0004974968569187603,
+      "loss": 0.4921,
+      "step": 1542
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.9480469226837158,
+      "learning_rate": 0.0004973577687861509,
+      "loss": 0.3729,
+      "step": 1543
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.97214275598526,
+      "learning_rate": 0.0004972186058240059,
+      "loss": 0.4199,
+      "step": 1544
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.953370988368988,
+      "learning_rate": 0.00049707936808509,
+      "loss": 0.4074,
+      "step": 1545
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.9177734851837158,
+      "learning_rate": 0.0004969400556221959,
+      "loss": 0.4098,
+      "step": 1546
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.8884267807006836,
+      "learning_rate": 0.0004968006684881451,
+      "loss": 0.371,
+      "step": 1547
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.2168620824813843,
+      "learning_rate": 0.0004966612067357871,
+      "loss": 0.4813,
+      "step": 1548
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.9950072169303894,
+      "learning_rate": 0.0004965216704179999,
+      "loss": 0.4525,
+      "step": 1549
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.151914358139038,
+      "learning_rate": 0.0004963820595876892,
+      "loss": 0.6082,
+      "step": 1550
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.9278011918067932,
+      "learning_rate": 0.0004962423742977898,
+      "loss": 0.392,
+      "step": 1551
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.0625840425491333,
+      "learning_rate": 0.0004961026146012644,
+      "loss": 0.4929,
+      "step": 1552
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.8793632984161377,
+      "learning_rate": 0.0004959627805511034,
+      "loss": 0.308,
+      "step": 1553
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.9269048571586609,
+      "learning_rate": 0.0004958228722003262,
+      "loss": 0.417,
+      "step": 1554
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.8447689414024353,
+      "learning_rate": 0.0004956828896019799,
+      "loss": 0.3498,
+      "step": 1555
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.0223203897476196,
+      "learning_rate": 0.0004955428328091398,
+      "loss": 0.3679,
+      "step": 1556
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.8199823498725891,
+      "learning_rate": 0.0004954027018749095,
+      "loss": 0.3089,
+      "step": 1557
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.9902119636535645,
+      "learning_rate": 0.0004952624968524203,
+      "loss": 0.4621,
+      "step": 1558
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.9957500100135803,
+      "learning_rate": 0.0004951222177948323,
+      "loss": 0.3915,
+      "step": 1559
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.0612666606903076,
+      "learning_rate": 0.0004949818647553329,
+      "loss": 0.4443,
+      "step": 1560
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.0264724493026733,
+      "learning_rate": 0.0004948414377871381,
+      "loss": 0.4711,
+      "step": 1561
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.8346537351608276,
+      "learning_rate": 0.0004947009369434917,
+      "loss": 0.2996,
+      "step": 1562
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.9985209107398987,
+      "learning_rate": 0.0004945603622776656,
+      "loss": 0.4312,
+      "step": 1563
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.8846251368522644,
+      "learning_rate": 0.0004944197138429594,
+      "loss": 0.3876,
+      "step": 1564
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.0236539840698242,
+      "learning_rate": 0.0004942789916927012,
+      "loss": 0.4073,
+      "step": 1565
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.8909901976585388,
+      "learning_rate": 0.0004941381958802467,
+      "loss": 0.3029,
+      "step": 1566
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.9777463674545288,
+      "learning_rate": 0.0004939973264589793,
+      "loss": 0.4429,
+      "step": 1567
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.1261863708496094,
+      "learning_rate": 0.000493856383482311,
+      "loss": 0.4831,
+      "step": 1568
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.8505493998527527,
+      "learning_rate": 0.0004937153670036811,
+      "loss": 0.3465,
+      "step": 1569
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.948000431060791,
+      "learning_rate": 0.000493574277076557,
+      "loss": 0.442,
+      "step": 1570
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.9175281524658203,
+      "learning_rate": 0.0004934331137544337,
+      "loss": 0.3285,
+      "step": 1571
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.927971601486206,
+      "learning_rate": 0.0004932918770908344,
+      "loss": 0.342,
+      "step": 1572
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.9144307971000671,
+      "learning_rate": 0.0004931505671393102,
+      "loss": 0.3803,
+      "step": 1573
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.096097707748413,
+      "learning_rate": 0.000493009183953439,
+      "loss": 0.4751,
+      "step": 1574
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.1528116464614868,
+      "learning_rate": 0.000492867727586828,
+      "loss": 0.4342,
+      "step": 1575
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.9170631170272827,
+      "learning_rate": 0.0004927261980931109,
+      "loss": 0.4487,
+      "step": 1576
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.875718891620636,
+      "learning_rate": 0.0004925845955259496,
+      "loss": 0.3518,
+      "step": 1577
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.7244861125946045,
+      "learning_rate": 0.0004924429199390337,
+      "loss": 0.2681,
+      "step": 1578
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.7114342451095581,
+      "learning_rate": 0.0004923011713860807,
+      "loss": 0.2456,
+      "step": 1579
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.7306610941886902,
+      "learning_rate": 0.0004921593499208351,
+      "loss": 0.2102,
+      "step": 1580
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.8104738593101501,
+      "learning_rate": 0.0004920174555970698,
+      "loss": 0.1924,
+      "step": 1581
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.8847179412841797,
+      "learning_rate": 0.0004918754884685847,
+      "loss": 0.2801,
+      "step": 1582
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.9546067714691162,
+      "learning_rate": 0.0004917334485892079,
+      "loss": 0.2852,
+      "step": 1583
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.8702731132507324,
+      "learning_rate": 0.0004915913360127946,
+      "loss": 0.2339,
+      "step": 1584
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.7458964586257935,
+      "learning_rate": 0.000491449150793228,
+      "loss": 0.18,
+      "step": 1585
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.824317455291748,
+      "learning_rate": 0.0004913068929844182,
+      "loss": 0.187,
+      "step": 1586
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.942261815071106,
+      "learning_rate": 0.0004911645626403035,
+      "loss": 0.2803,
+      "step": 1587
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 1.0565816164016724,
+      "learning_rate": 0.0004910221598148492,
+      "loss": 0.3942,
+      "step": 1588
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 1.0009928941726685,
+      "learning_rate": 0.0004908796845620485,
+      "loss": 0.2592,
+      "step": 1589
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 1.0129635334014893,
+      "learning_rate": 0.0004907371369359217,
+      "loss": 0.2576,
+      "step": 1590
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 1.0173312425613403,
+      "learning_rate": 0.0004905945169905168,
+      "loss": 0.2972,
+      "step": 1591
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 1.0006557703018188,
+      "learning_rate": 0.0004904518247799091,
+      "loss": 0.3014,
+      "step": 1592
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 1.0242513418197632,
+      "learning_rate": 0.0004903090603582014,
+      "loss": 0.3023,
+      "step": 1593
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.7629026770591736,
+      "learning_rate": 0.0004901662237795236,
+      "loss": 0.1984,
+      "step": 1594
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.8585073351860046,
+      "learning_rate": 0.0004900233150980332,
+      "loss": 0.2499,
+      "step": 1595
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 1.0251237154006958,
+      "learning_rate": 0.0004898803343679152,
+      "loss": 0.2847,
+      "step": 1596
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.9117810130119324,
+      "learning_rate": 0.0004897372816433815,
+      "loss": 0.302,
+      "step": 1597
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.9581530690193176,
+      "learning_rate": 0.0004895941569786716,
+      "loss": 0.2922,
+      "step": 1598
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.9487888813018799,
+      "learning_rate": 0.0004894509604280522,
+      "loss": 0.2975,
+      "step": 1599
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.8367571234703064,
+      "learning_rate": 0.0004893076920458173,
+      "loss": 0.238,
+      "step": 1600
+    },
+    {
+      "epoch": 3.04,
+      "eval_blimp_filtered_avg": 0.5202985074626866,
+      "eval_blimp_filtered_std": 0.005817009221165117,
+      "step": 1600
+    },
+    {
+      "epoch": 3.04,
+      "eval_blimp_supplement_avg": 0.5237068965517241,
+      "eval_blimp_supplement_std": 0.023237946550153234,
+      "step": 1600
+    },
+    {
+      "epoch": 3.04,
+      "eval_vqa_filtered_avg": 0.15,
+      "eval_vqa_filtered_std": 0.0358870281282637,
+      "step": 1600
+    },
+    {
+      "epoch": 3.04,
+      "eval_winoground_filtered_avg": 0.57,
+      "eval_winoground_filtered_std": 0.04975698519562428,
+      "step": 1600
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.9771935939788818,
+      "learning_rate": 0.0004891643518862877,
+      "loss": 0.2613,
+      "step": 1601
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.9161196947097778,
+      "learning_rate": 0.0004890209400038123,
+      "loss": 0.2512,
+      "step": 1602
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.7642145752906799,
+      "learning_rate": 0.0004888774564527664,
+      "loss": 0.2255,
+      "step": 1603
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.8867236971855164,
+      "learning_rate": 0.0004887339012875529,
+      "loss": 0.2644,
+      "step": 1604
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.9194115400314331,
+      "learning_rate": 0.0004885902745626016,
+      "loss": 0.2249,
+      "step": 1605
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.805370569229126,
+      "learning_rate": 0.0004884465763323695,
+      "loss": 0.2241,
+      "step": 1606
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.7856239676475525,
+      "learning_rate": 0.000488302806651341,
+      "loss": 0.1871,
+      "step": 1607
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.8552716970443726,
+      "learning_rate": 0.00048815896557402713,
+      "loss": 0.2219,
+      "step": 1608
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 1.089538335800171,
+      "learning_rate": 0.0004880150531549663,
+      "loss": 0.3042,
+      "step": 1609
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.9803922772407532,
+      "learning_rate": 0.0004878710694487238,
+      "loss": 0.26,
+      "step": 1610
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 1.0482945442199707,
+      "learning_rate": 0.0004877270145098921,
+      "loss": 0.2646,
+      "step": 1611
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.8392722606658936,
+      "learning_rate": 0.00048758288839309067,
+      "loss": 0.2429,
+      "step": 1612
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 1.0645718574523926,
+      "learning_rate": 0.00048743869115296573,
+      "loss": 0.2926,
+      "step": 1613
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.8703613877296448,
+      "learning_rate": 0.0004872944228441908,
+      "loss": 0.2641,
+      "step": 1614
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.8940256237983704,
+      "learning_rate": 0.00048715008352146605,
+      "loss": 0.2414,
+      "step": 1615
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 1.0254648923873901,
+      "learning_rate": 0.00048700567323951887,
+      "loss": 0.3174,
+      "step": 1616
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.6667085289955139,
+      "learning_rate": 0.0004868611920531032,
+      "loss": 0.1908,
+      "step": 1617
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.7777180075645447,
+      "learning_rate": 0.0004867166400170002,
+      "loss": 0.2251,
+      "step": 1618
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.8020041584968567,
+      "learning_rate": 0.0004865720171860178,
+      "loss": 0.2551,
+      "step": 1619
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.7695719003677368,
+      "learning_rate": 0.00048642732361499065,
+      "loss": 0.207,
+      "step": 1620
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.8519812226295471,
+      "learning_rate": 0.0004862825593587802,
+      "loss": 0.2598,
+      "step": 1621
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.9150854349136353,
+      "learning_rate": 0.000486137724472275,
+      "loss": 0.2721,
+      "step": 1622
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.991997480392456,
+      "learning_rate": 0.00048599281901039,
+      "loss": 0.3367,
+      "step": 1623
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.8089627623558044,
+      "learning_rate": 0.0004858478430280672,
+      "loss": 0.2434,
+      "step": 1624
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.7579286098480225,
+      "learning_rate": 0.00048570279658027523,
+      "loss": 0.2049,
+      "step": 1625
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.7090297341346741,
+      "learning_rate": 0.0004855576797220094,
+      "loss": 0.2161,
+      "step": 1626
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.8292402029037476,
+      "learning_rate": 0.00048541249250829164,
+      "loss": 0.1878,
+      "step": 1627
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 1.1044442653656006,
+      "learning_rate": 0.0004852672349941709,
+      "loss": 0.3018,
+      "step": 1628
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.963199257850647,
+      "learning_rate": 0.0004851219072347224,
+      "loss": 0.2974,
+      "step": 1629
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.9070634245872498,
+      "learning_rate": 0.0004849765092850481,
+      "loss": 0.2475,
+      "step": 1630
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 1.0448908805847168,
+      "learning_rate": 0.0004848310412002767,
+      "loss": 0.3021,
+      "step": 1631
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.8086839318275452,
+      "learning_rate": 0.0004846855030355634,
+      "loss": 0.2431,
+      "step": 1632
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.9447100162506104,
+      "learning_rate": 0.00048453989484608994,
+      "loss": 0.2645,
+      "step": 1633
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.8199095129966736,
+      "learning_rate": 0.0004843942166870648,
+      "loss": 0.1855,
+      "step": 1634
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.9078908562660217,
+      "learning_rate": 0.00048424846861372264,
+      "loss": 0.3371,
+      "step": 1635
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.8610200881958008,
+      "learning_rate": 0.000484102650681325,
+      "loss": 0.312,
+      "step": 1636
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.9628205895423889,
+      "learning_rate": 0.0004839567629451596,
+      "loss": 0.2714,
+      "step": 1637
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.8173263072967529,
+      "learning_rate": 0.0004838108054605408,
+      "loss": 0.2099,
+      "step": 1638
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.9361103177070618,
+      "learning_rate": 0.0004836647782828094,
+      "loss": 0.2804,
+      "step": 1639
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.8456994891166687,
+      "learning_rate": 0.0004835186814673326,
+      "loss": 0.2566,
+      "step": 1640
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.8395893573760986,
+      "learning_rate": 0.00048337251506950406,
+      "loss": 0.2464,
+      "step": 1641
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.9126836657524109,
+      "learning_rate": 0.0004832262791447436,
+      "loss": 0.2467,
+      "step": 1642
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.8928797841072083,
+      "learning_rate": 0.0004830799737484976,
+      "loss": 0.2464,
+      "step": 1643
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.7720723152160645,
+      "learning_rate": 0.0004829335989362389,
+      "loss": 0.154,
+      "step": 1644
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.8591120839118958,
+      "learning_rate": 0.00048278715476346635,
+      "loss": 0.2096,
+      "step": 1645
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 1.0270227193832397,
+      "learning_rate": 0.00048264064128570526,
+      "loss": 0.3114,
+      "step": 1646
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.9896075129508972,
+      "learning_rate": 0.00048249405855850725,
+      "loss": 0.2357,
+      "step": 1647
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.9220013618469238,
+      "learning_rate": 0.00048234740663745016,
+      "loss": 0.2877,
+      "step": 1648
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.8865196108818054,
+      "learning_rate": 0.000482200685578138,
+      "loss": 0.2664,
+      "step": 1649
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.8167412281036377,
+      "learning_rate": 0.0004820538954362012,
+      "loss": 0.2206,
+      "step": 1650
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.9956057667732239,
+      "learning_rate": 0.00048190703626729603,
+      "loss": 0.2951,
+      "step": 1651
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.9930279850959778,
+      "learning_rate": 0.00048176010812710533,
+      "loss": 0.3201,
+      "step": 1652
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.8051810264587402,
+      "learning_rate": 0.00048161311107133777,
+      "loss": 0.2088,
+      "step": 1653
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 1.0922728776931763,
+      "learning_rate": 0.0004814660451557283,
+      "loss": 0.3215,
+      "step": 1654
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.7643908262252808,
+      "learning_rate": 0.000481318910436038,
+      "loss": 0.1749,
+      "step": 1655
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.8333352208137512,
+      "learning_rate": 0.0004811717069680539,
+      "loss": 0.2597,
+      "step": 1656
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.7828906774520874,
+      "learning_rate": 0.0004810244348075894,
+      "loss": 0.1908,
+      "step": 1657
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.9457138776779175,
+      "learning_rate": 0.00048087709401048353,
+      "loss": 0.2548,
+      "step": 1658
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.8293535709381104,
+      "learning_rate": 0.00048072968463260164,
+      "loss": 0.2306,
+      "step": 1659
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 1.1084262132644653,
+      "learning_rate": 0.00048058220672983503,
+      "loss": 0.3661,
+      "step": 1660
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.8997864127159119,
+      "learning_rate": 0.0004804346603581009,
+      "loss": 0.2582,
+      "step": 1661
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.8688557147979736,
+      "learning_rate": 0.00048028704557334244,
+      "loss": 0.2248,
+      "step": 1662
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.9156268835067749,
+      "learning_rate": 0.0004801393624315288,
+      "loss": 0.2288,
+      "step": 1663
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.9201476573944092,
+      "learning_rate": 0.000479991610988655,
+      "loss": 0.2762,
+      "step": 1664
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.8761970400810242,
+      "learning_rate": 0.00047984379130074214,
+      "loss": 0.2706,
+      "step": 1665
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 1.1126759052276611,
+      "learning_rate": 0.0004796959034238369,
+      "loss": 0.3351,
+      "step": 1666
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 1.0252509117126465,
+      "learning_rate": 0.00047954794741401213,
+      "loss": 0.3302,
+      "step": 1667
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.8313443660736084,
+      "learning_rate": 0.0004793999233273662,
+      "loss": 0.2253,
+      "step": 1668
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.9196571707725525,
+      "learning_rate": 0.00047925183122002357,
+      "loss": 0.2559,
+      "step": 1669
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.7829031944274902,
+      "learning_rate": 0.00047910367114813423,
+      "loss": 0.2247,
+      "step": 1670
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.8847941160202026,
+      "learning_rate": 0.00047895544316787434,
+      "loss": 0.2527,
+      "step": 1671
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.8432680368423462,
+      "learning_rate": 0.0004788071473354452,
+      "loss": 0.2496,
+      "step": 1672
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.6904547810554504,
+      "learning_rate": 0.00047865878370707436,
+      "loss": 0.1623,
+      "step": 1673
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 1.064517617225647,
+      "learning_rate": 0.0004785103523390149,
+      "loss": 0.2544,
+      "step": 1674
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.8830809593200684,
+      "learning_rate": 0.0004783618532875456,
+      "loss": 0.2834,
+      "step": 1675
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.9792788028717041,
+      "learning_rate": 0.0004782132866089708,
+      "loss": 0.2786,
+      "step": 1676
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.9855789542198181,
+      "learning_rate": 0.00047806465235962053,
+      "loss": 0.2973,
+      "step": 1677
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.8373962640762329,
+      "learning_rate": 0.0004779159505958506,
+      "loss": 0.2422,
+      "step": 1678
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.8975450992584229,
+      "learning_rate": 0.0004777671813740422,
+      "loss": 0.2308,
+      "step": 1679
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.8988074064254761,
+      "learning_rate": 0.0004776183447506023,
+      "loss": 0.25,
+      "step": 1680
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 1.0293368101119995,
+      "learning_rate": 0.0004774694407819632,
+      "loss": 0.3471,
+      "step": 1681
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.9481083750724792,
+      "learning_rate": 0.00047732046952458283,
+      "loss": 0.2985,
+      "step": 1682
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.8688005208969116,
+      "learning_rate": 0.0004771714310349448,
+      "loss": 0.2739,
+      "step": 1683
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.9371510744094849,
+      "learning_rate": 0.00047702232536955786,
+      "loss": 0.2691,
+      "step": 1684
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.8336614370346069,
+      "learning_rate": 0.00047687315258495666,
+      "loss": 0.2465,
+      "step": 1685
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.8456622362136841,
+      "learning_rate": 0.00047672391273770087,
+      "loss": 0.245,
+      "step": 1686
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.8711954355239868,
+      "learning_rate": 0.00047657460588437594,
+      "loss": 0.2663,
+      "step": 1687
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.8434468507766724,
+      "learning_rate": 0.0004764252320815925,
+      "loss": 0.2241,
+      "step": 1688
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.7671254873275757,
+      "learning_rate": 0.00047627579138598665,
+      "loss": 0.2744,
+      "step": 1689
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.8651210069656372,
+      "learning_rate": 0.0004761262838542198,
+      "loss": 0.3014,
+      "step": 1690
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.9088365435600281,
+      "learning_rate": 0.00047597670954297887,
+      "loss": 0.25,
+      "step": 1691
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.813748300075531,
+      "learning_rate": 0.0004758270685089758,
+      "loss": 0.2712,
+      "step": 1692
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.9270053505897522,
+      "learning_rate": 0.0004756773608089481,
+      "loss": 0.3285,
+      "step": 1693
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.8532443046569824,
+      "learning_rate": 0.0004755275864996586,
+      "loss": 0.3229,
+      "step": 1694
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.9715031385421753,
+      "learning_rate": 0.0004753777456378949,
+      "loss": 0.3055,
+      "step": 1695
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.7146458029747009,
+      "learning_rate": 0.0004752278382804705,
+      "loss": 0.1863,
+      "step": 1696
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.7760129570960999,
+      "learning_rate": 0.00047507786448422365,
+      "loss": 0.2768,
+      "step": 1697
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.9156978130340576,
+      "learning_rate": 0.0004749278243060179,
+      "loss": 0.2411,
+      "step": 1698
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.8896867632865906,
+      "learning_rate": 0.00047477771780274206,
+      "loss": 0.2664,
+      "step": 1699
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.7771943807601929,
+      "learning_rate": 0.00047462754503131005,
+      "loss": 0.2041,
+      "step": 1700
+    },
+    {
+      "epoch": 3.23,
+      "eval_blimp_filtered_avg": 0.5425373134328358,
+      "eval_blimp_filtered_std": 0.005742042848322313,
+      "step": 1700
+    },
+    {
+      "epoch": 3.23,
+      "eval_blimp_supplement_avg": 0.5323275862068966,
+      "eval_blimp_supplement_std": 0.023122487843413718,
+      "step": 1700
+    },
+    {
+      "epoch": 3.23,
+      "eval_vqa_filtered_avg": 0.16,
+      "eval_vqa_filtered_std": 0.03684529491774708,
+      "step": 1700
+    },
+    {
+      "epoch": 3.23,
+      "eval_winoground_filtered_avg": 0.51,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 1700
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.8913343548774719,
+      "learning_rate": 0.00047447730604866086,
+      "loss": 0.1968,
+      "step": 1701
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 1.0623739957809448,
+      "learning_rate": 0.00047432700091175854,
+      "loss": 0.3004,
+      "step": 1702
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.8643488883972168,
+      "learning_rate": 0.0004741766296775924,
+      "loss": 0.2435,
+      "step": 1703
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.9903916716575623,
+      "learning_rate": 0.00047402619240317673,
+      "loss": 0.2767,
+      "step": 1704
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.6774560213088989,
+      "learning_rate": 0.00047387568914555086,
+      "loss": 0.2007,
+      "step": 1705
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.761810839176178,
+      "learning_rate": 0.00047372511996177894,
+      "loss": 0.2127,
+      "step": 1706
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.7326056957244873,
+      "learning_rate": 0.0004735744849089505,
+      "loss": 0.1929,
+      "step": 1707
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.7604557871818542,
+      "learning_rate": 0.00047342378404417973,
+      "loss": 0.2144,
+      "step": 1708
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 1.0077298879623413,
+      "learning_rate": 0.0004732730174246059,
+      "loss": 0.3815,
+      "step": 1709
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.8862660527229309,
+      "learning_rate": 0.0004731221851073933,
+      "loss": 0.3322,
+      "step": 1710
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.843356728553772,
+      "learning_rate": 0.0004729712871497309,
+      "loss": 0.2134,
+      "step": 1711
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.8757851719856262,
+      "learning_rate": 0.00047282032360883265,
+      "loss": 0.2809,
+      "step": 1712
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.8975480198860168,
+      "learning_rate": 0.00047266929454193755,
+      "loss": 0.2336,
+      "step": 1713
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.8711280822753906,
+      "learning_rate": 0.00047251820000630926,
+      "loss": 0.2385,
+      "step": 1714
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.9138095378875732,
+      "learning_rate": 0.00047236704005923605,
+      "loss": 0.2243,
+      "step": 1715
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.9561814665794373,
+      "learning_rate": 0.0004722158147580316,
+      "loss": 0.3045,
+      "step": 1716
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.8951904773712158,
+      "learning_rate": 0.00047206452416003386,
+      "loss": 0.255,
+      "step": 1717
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.9862191677093506,
+      "learning_rate": 0.0004719131683226056,
+      "loss": 0.2459,
+      "step": 1718
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.9088127613067627,
+      "learning_rate": 0.00047176174730313453,
+      "loss": 0.2329,
+      "step": 1719
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.8778693675994873,
+      "learning_rate": 0.00047161026115903284,
+      "loss": 0.2387,
+      "step": 1720
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 1.0340452194213867,
+      "learning_rate": 0.00047145870994773765,
+      "loss": 0.3626,
+      "step": 1721
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 0.8745329976081848,
+      "learning_rate": 0.0004713070937267106,
+      "loss": 0.3095,
+      "step": 1722
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.7564894556999207,
+      "learning_rate": 0.00047115541255343793,
+      "loss": 0.2533,
+      "step": 1723
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.8412438035011292,
+      "learning_rate": 0.00047100366648543076,
+      "loss": 0.2605,
+      "step": 1724
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.8909465074539185,
+      "learning_rate": 0.0004708518555802244,
+      "loss": 0.2999,
+      "step": 1725
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.8907904624938965,
+      "learning_rate": 0.0004706999798953791,
+      "loss": 0.3716,
+      "step": 1726
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.6813023686408997,
+      "learning_rate": 0.0004705480394884796,
+      "loss": 0.22,
+      "step": 1727
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.8728662133216858,
+      "learning_rate": 0.0004703960344171351,
+      "loss": 0.2824,
+      "step": 1728
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.7727020978927612,
+      "learning_rate": 0.0004702439647389793,
+      "loss": 0.2277,
+      "step": 1729
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.7586476802825928,
+      "learning_rate": 0.00047009183051167057,
+      "loss": 0.1823,
+      "step": 1730
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 1.0866812467575073,
+      "learning_rate": 0.0004699396317928915,
+      "loss": 0.3798,
+      "step": 1731
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.9190901517868042,
+      "learning_rate": 0.00046978736864034936,
+      "loss": 0.2826,
+      "step": 1732
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.8600939512252808,
+      "learning_rate": 0.0004696350411117757,
+      "loss": 0.2338,
+      "step": 1733
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.8057361245155334,
+      "learning_rate": 0.00046948264926492656,
+      "loss": 0.2325,
+      "step": 1734
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.8117899894714355,
+      "learning_rate": 0.00046933019315758233,
+      "loss": 0.2314,
+      "step": 1735
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.9652373194694519,
+      "learning_rate": 0.0004691776728475478,
+      "loss": 0.3237,
+      "step": 1736
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.7493473887443542,
+      "learning_rate": 0.000469025088392652,
+      "loss": 0.2295,
+      "step": 1737
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.7187075614929199,
+      "learning_rate": 0.0004688724398507485,
+      "loss": 0.2272,
+      "step": 1738
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.9114627838134766,
+      "learning_rate": 0.00046871972727971495,
+      "loss": 0.2942,
+      "step": 1739
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.834536612033844,
+      "learning_rate": 0.0004685669507374534,
+      "loss": 0.2942,
+      "step": 1740
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.8227934837341309,
+      "learning_rate": 0.0004684141102818899,
+      "loss": 0.3085,
+      "step": 1741
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.724820077419281,
+      "learning_rate": 0.0004682612059709753,
+      "loss": 0.1968,
+      "step": 1742
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.7963129281997681,
+      "learning_rate": 0.00046810823786268404,
+      "loss": 0.2209,
+      "step": 1743
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.7247688174247742,
+      "learning_rate": 0.00046795520601501504,
+      "loss": 0.227,
+      "step": 1744
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.8857767581939697,
+      "learning_rate": 0.0004678021104859916,
+      "loss": 0.2477,
+      "step": 1745
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.8686899542808533,
+      "learning_rate": 0.0004676489513336606,
+      "loss": 0.3011,
+      "step": 1746
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.9105591773986816,
+      "learning_rate": 0.00046749572861609347,
+      "loss": 0.2806,
+      "step": 1747
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.7681505084037781,
+      "learning_rate": 0.0004673424423913857,
+      "loss": 0.2302,
+      "step": 1748
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.8475282192230225,
+      "learning_rate": 0.00046718909271765686,
+      "loss": 0.2629,
+      "step": 1749
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.8257328867912292,
+      "learning_rate": 0.0004670356796530505,
+      "loss": 0.2503,
+      "step": 1750
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.8875759243965149,
+      "learning_rate": 0.0004668822032557341,
+      "loss": 0.2718,
+      "step": 1751
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.9054738283157349,
+      "learning_rate": 0.00046672866358389933,
+      "loss": 0.2996,
+      "step": 1752
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.8658378720283508,
+      "learning_rate": 0.00046657506069576184,
+      "loss": 0.247,
+      "step": 1753
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.6770904660224915,
+      "learning_rate": 0.00046642139464956104,
+      "loss": 0.1955,
+      "step": 1754
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.8122730851173401,
+      "learning_rate": 0.00046626766550356076,
+      "loss": 0.2242,
+      "step": 1755
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.8624380230903625,
+      "learning_rate": 0.00046611387331604826,
+      "loss": 0.2658,
+      "step": 1756
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.8542689085006714,
+      "learning_rate": 0.0004659600181453349,
+      "loss": 0.2612,
+      "step": 1757
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.810631275177002,
+      "learning_rate": 0.00046580610004975595,
+      "loss": 0.2638,
+      "step": 1758
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 0.7271432280540466,
+      "learning_rate": 0.0004656521190876704,
+      "loss": 0.255,
+      "step": 1759
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.7823249697685242,
+      "learning_rate": 0.00046549807531746126,
+      "loss": 0.293,
+      "step": 1760
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.8456376194953918,
+      "learning_rate": 0.0004653439687975352,
+      "loss": 0.246,
+      "step": 1761
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.8167811632156372,
+      "learning_rate": 0.0004651897995863229,
+      "loss": 0.2327,
+      "step": 1762
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.8075510859489441,
+      "learning_rate": 0.0004650355677422785,
+      "loss": 0.262,
+      "step": 1763
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.7998697757720947,
+      "learning_rate": 0.0004648812733238801,
+      "loss": 0.2668,
+      "step": 1764
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 1.1107600927352905,
+      "learning_rate": 0.00046472691638962965,
+      "loss": 0.3145,
+      "step": 1765
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.8066977262496948,
+      "learning_rate": 0.00046457249699805237,
+      "loss": 0.2308,
+      "step": 1766
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.7746822834014893,
+      "learning_rate": 0.0004644180152076974,
+      "loss": 0.1933,
+      "step": 1767
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.8920192122459412,
+      "learning_rate": 0.0004642634710771378,
+      "loss": 0.3042,
+      "step": 1768
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.6649077534675598,
+      "learning_rate": 0.00046410886466496996,
+      "loss": 0.1752,
+      "step": 1769
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.903344452381134,
+      "learning_rate": 0.0004639541960298137,
+      "loss": 0.2659,
+      "step": 1770
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.849105954170227,
+      "learning_rate": 0.00046379946523031297,
+      "loss": 0.2347,
+      "step": 1771
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.8402355313301086,
+      "learning_rate": 0.0004636446723251349,
+      "loss": 0.2351,
+      "step": 1772
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 1.0331000089645386,
+      "learning_rate": 0.0004634898173729703,
+      "loss": 0.3863,
+      "step": 1773
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.7553647756576538,
+      "learning_rate": 0.00046333490043253336,
+      "loss": 0.237,
+      "step": 1774
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.8808202743530273,
+      "learning_rate": 0.0004631799215625621,
+      "loss": 0.2603,
+      "step": 1775
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 1.098543405532837,
+      "learning_rate": 0.00046302488082181764,
+      "loss": 0.3896,
+      "step": 1776
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.7932633757591248,
+      "learning_rate": 0.0004628697782690846,
+      "loss": 0.2709,
+      "step": 1777
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.7767360806465149,
+      "learning_rate": 0.0004627146139631715,
+      "loss": 0.2198,
+      "step": 1778
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.6740002632141113,
+      "learning_rate": 0.00046255938796290976,
+      "loss": 0.1995,
+      "step": 1779
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.8023741841316223,
+      "learning_rate": 0.0004624041003271543,
+      "loss": 0.3174,
+      "step": 1780
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.8884091377258301,
+      "learning_rate": 0.0004622487511147837,
+      "loss": 0.254,
+      "step": 1781
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.8564543724060059,
+      "learning_rate": 0.0004620933403846994,
+      "loss": 0.2512,
+      "step": 1782
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.777367889881134,
+      "learning_rate": 0.00046193786819582653,
+      "loss": 0.2205,
+      "step": 1783
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.8416761159896851,
+      "learning_rate": 0.00046178233460711355,
+      "loss": 0.2645,
+      "step": 1784
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.9022892117500305,
+      "learning_rate": 0.0004616267396775319,
+      "loss": 0.3175,
+      "step": 1785
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 1.0481648445129395,
+      "learning_rate": 0.0004614710834660767,
+      "loss": 0.3015,
+      "step": 1786
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.9611367583274841,
+      "learning_rate": 0.0004613153660317657,
+      "loss": 0.2711,
+      "step": 1787
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.9286614656448364,
+      "learning_rate": 0.00046115958743364046,
+      "loss": 0.3263,
+      "step": 1788
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.7599608898162842,
+      "learning_rate": 0.0004610037477307655,
+      "loss": 0.1968,
+      "step": 1789
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.8382947444915771,
+      "learning_rate": 0.00046084784698222835,
+      "loss": 0.2154,
+      "step": 1790
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.798367440700531,
+      "learning_rate": 0.0004606918852471399,
+      "loss": 0.2452,
+      "step": 1791
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.8631271719932556,
+      "learning_rate": 0.00046053586258463423,
+      "loss": 0.3115,
+      "step": 1792
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.6910966038703918,
+      "learning_rate": 0.0004603797790538682,
+      "loss": 0.1715,
+      "step": 1793
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.9161866307258606,
+      "learning_rate": 0.00046022363471402206,
+      "loss": 0.3118,
+      "step": 1794
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.7160239219665527,
+      "learning_rate": 0.0004600674296242989,
+      "loss": 0.1823,
+      "step": 1795
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.8758367300033569,
+      "learning_rate": 0.0004599111638439251,
+      "loss": 0.2707,
+      "step": 1796
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.9270631670951843,
+      "learning_rate": 0.00045975483743214966,
+      "loss": 0.353,
+      "step": 1797
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.8585798144340515,
+      "learning_rate": 0.000459598450448245,
+      "loss": 0.2622,
+      "step": 1798
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.5725598335266113,
+      "learning_rate": 0.00045944200295150624,
+      "loss": 0.1474,
+      "step": 1799
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.7205086946487427,
+      "learning_rate": 0.00045928549500125145,
+      "loss": 0.1911,
+      "step": 1800
+    },
+    {
+      "epoch": 3.42,
+      "eval_blimp_filtered_avg": 0.5437313432835821,
+      "eval_blimp_filtered_std": 0.005768979927594455,
+      "step": 1800
+    },
+    {
+      "epoch": 3.42,
+      "eval_blimp_supplement_avg": 0.4892241379310345,
+      "eval_blimp_supplement_std": 0.023095756671360063,
+      "step": 1800
+    },
+    {
+      "epoch": 3.42,
+      "eval_vqa_filtered_avg": 0.19,
+      "eval_vqa_filtered_std": 0.03942772444036623,
+      "step": 1800
+    },
+    {
+      "epoch": 3.42,
+      "eval_winoground_filtered_avg": 0.46,
+      "eval_winoground_filtered_std": 0.05009082659620333,
+      "step": 1800
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.7342203855514526,
+      "learning_rate": 0.00045912892665682175,
+      "loss": 0.2372,
+      "step": 1801
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.8348943591117859,
+      "learning_rate": 0.000458972297977581,
+      "loss": 0.2305,
+      "step": 1802
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.9850256443023682,
+      "learning_rate": 0.00045881560902291614,
+      "loss": 0.2486,
+      "step": 1803
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.920192301273346,
+      "learning_rate": 0.0004586588598522368,
+      "loss": 0.2902,
+      "step": 1804
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.8440070748329163,
+      "learning_rate": 0.0004585020505249755,
+      "loss": 0.24,
+      "step": 1805
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.7701641321182251,
+      "learning_rate": 0.00045834518110058746,
+      "loss": 0.2463,
+      "step": 1806
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.8310728073120117,
+      "learning_rate": 0.000458188251638551,
+      "loss": 0.239,
+      "step": 1807
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.8051304221153259,
+      "learning_rate": 0.0004580312621983667,
+      "loss": 0.2488,
+      "step": 1808
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.6570320129394531,
+      "learning_rate": 0.0004578742128395585,
+      "loss": 0.2099,
+      "step": 1809
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.8045438528060913,
+      "learning_rate": 0.00045771710362167245,
+      "loss": 0.2638,
+      "step": 1810
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.8699944019317627,
+      "learning_rate": 0.0004575599346042778,
+      "loss": 0.2405,
+      "step": 1811
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.7820262908935547,
+      "learning_rate": 0.00045740270584696604,
+      "loss": 0.2273,
+      "step": 1812
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.7144664525985718,
+      "learning_rate": 0.0004572454174093516,
+      "loss": 0.2164,
+      "step": 1813
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.7218126058578491,
+      "learning_rate": 0.0004570880693510715,
+      "loss": 0.2098,
+      "step": 1814
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.8810663819313049,
+      "learning_rate": 0.0004569306617317853,
+      "loss": 0.3042,
+      "step": 1815
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.7971299290657043,
+      "learning_rate": 0.00045677319461117516,
+      "loss": 0.2398,
+      "step": 1816
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 1.0783491134643555,
+      "learning_rate": 0.00045661566804894576,
+      "loss": 0.2976,
+      "step": 1817
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.8880136013031006,
+      "learning_rate": 0.00045645808210482443,
+      "loss": 0.3119,
+      "step": 1818
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.8670411705970764,
+      "learning_rate": 0.0004563004368385609,
+      "loss": 0.3044,
+      "step": 1819
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.8575292825698853,
+      "learning_rate": 0.0004561427323099275,
+      "loss": 0.2428,
+      "step": 1820
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.8761627078056335,
+      "learning_rate": 0.00045598496857871896,
+      "loss": 0.2696,
+      "step": 1821
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.6587812304496765,
+      "learning_rate": 0.0004558271457047524,
+      "loss": 0.1862,
+      "step": 1822
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.8381554484367371,
+      "learning_rate": 0.0004556692637478676,
+      "loss": 0.225,
+      "step": 1823
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 1.038540244102478,
+      "learning_rate": 0.00045551132276792634,
+      "loss": 0.3244,
+      "step": 1824
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.9997128248214722,
+      "learning_rate": 0.00045535332282481313,
+      "loss": 0.3337,
+      "step": 1825
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.7354126572608948,
+      "learning_rate": 0.00045519526397843474,
+      "loss": 0.2168,
+      "step": 1826
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 0.7276039123535156,
+      "learning_rate": 0.0004550371462887203,
+      "loss": 0.1678,
+      "step": 1827
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.8754634857177734,
+      "learning_rate": 0.00045487896981562096,
+      "loss": 0.2239,
+      "step": 1828
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.804906964302063,
+      "learning_rate": 0.0004547207346191106,
+      "loss": 0.2005,
+      "step": 1829
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.8780803680419922,
+      "learning_rate": 0.0004545624407591852,
+      "loss": 0.3509,
+      "step": 1830
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.8129563331604004,
+      "learning_rate": 0.0004544040882958629,
+      "loss": 0.2643,
+      "step": 1831
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.6855679750442505,
+      "learning_rate": 0.0004542456772891841,
+      "loss": 0.2035,
+      "step": 1832
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.7677226066589355,
+      "learning_rate": 0.00045408720779921137,
+      "loss": 0.295,
+      "step": 1833
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.8141078352928162,
+      "learning_rate": 0.00045392867988602956,
+      "loss": 0.2316,
+      "step": 1834
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.7825101613998413,
+      "learning_rate": 0.0004537700936097456,
+      "loss": 0.2665,
+      "step": 1835
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.8865115642547607,
+      "learning_rate": 0.00045361144903048844,
+      "loss": 0.3082,
+      "step": 1836
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.6459495425224304,
+      "learning_rate": 0.0004534527462084094,
+      "loss": 0.1765,
+      "step": 1837
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.7147268652915955,
+      "learning_rate": 0.00045329398520368167,
+      "loss": 0.2186,
+      "step": 1838
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.7477104067802429,
+      "learning_rate": 0.0004531351660765004,
+      "loss": 0.2405,
+      "step": 1839
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.6847060918807983,
+      "learning_rate": 0.0004529762888870832,
+      "loss": 0.1835,
+      "step": 1840
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.9766390323638916,
+      "learning_rate": 0.0004528173536956693,
+      "loss": 0.278,
+      "step": 1841
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.7346838712692261,
+      "learning_rate": 0.00045265836056252004,
+      "loss": 0.1647,
+      "step": 1842
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.8242331147193909,
+      "learning_rate": 0.00045249930954791885,
+      "loss": 0.2185,
+      "step": 1843
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.7885558605194092,
+      "learning_rate": 0.0004523402007121709,
+      "loss": 0.2294,
+      "step": 1844
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.8449133634567261,
+      "learning_rate": 0.0004521810341156035,
+      "loss": 0.225,
+      "step": 1845
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.698259711265564,
+      "learning_rate": 0.00045202180981856555,
+      "loss": 0.1749,
+      "step": 1846
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.9894098043441772,
+      "learning_rate": 0.0004518625278814282,
+      "loss": 0.2954,
+      "step": 1847
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.8834659457206726,
+      "learning_rate": 0.0004517031883645841,
+      "loss": 0.242,
+      "step": 1848
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.8269351720809937,
+      "learning_rate": 0.000451543791328448,
+      "loss": 0.2278,
+      "step": 1849
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.7355552315711975,
+      "learning_rate": 0.00045138433683345657,
+      "loss": 0.1824,
+      "step": 1850
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.7778064608573914,
+      "learning_rate": 0.0004512248249400677,
+      "loss": 0.2029,
+      "step": 1851
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.8055093288421631,
+      "learning_rate": 0.0004510652557087616,
+      "loss": 0.2307,
+      "step": 1852
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.6711875796318054,
+      "learning_rate": 0.0004509056292000401,
+      "loss": 0.1772,
+      "step": 1853
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.877764105796814,
+      "learning_rate": 0.0004507459454744265,
+      "loss": 0.2387,
+      "step": 1854
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.9022437334060669,
+      "learning_rate": 0.00045058620459246596,
+      "loss": 0.2416,
+      "step": 1855
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.8407235741615295,
+      "learning_rate": 0.0004504264066147256,
+      "loss": 0.2387,
+      "step": 1856
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.9035999178886414,
+      "learning_rate": 0.00045026655160179354,
+      "loss": 0.3334,
+      "step": 1857
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.9116257429122925,
+      "learning_rate": 0.0004501066396142802,
+      "loss": 0.3003,
+      "step": 1858
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.7324634790420532,
+      "learning_rate": 0.00044994667071281705,
+      "loss": 0.2193,
+      "step": 1859
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.6285498142242432,
+      "learning_rate": 0.0004497866449580574,
+      "loss": 0.191,
+      "step": 1860
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.7754376530647278,
+      "learning_rate": 0.00044962656241067636,
+      "loss": 0.2205,
+      "step": 1861
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.8098081946372986,
+      "learning_rate": 0.00044946642313137,
+      "loss": 0.2861,
+      "step": 1862
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.7129320502281189,
+      "learning_rate": 0.0004493062271808564,
+      "loss": 0.1845,
+      "step": 1863
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.8500383496284485,
+      "learning_rate": 0.0004491459746198749,
+      "loss": 0.2505,
+      "step": 1864
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.8023073077201843,
+      "learning_rate": 0.0004489856655091863,
+      "loss": 0.211,
+      "step": 1865
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.8134183287620544,
+      "learning_rate": 0.00044882529990957295,
+      "loss": 0.2046,
+      "step": 1866
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.8146783709526062,
+      "learning_rate": 0.00044866487788183855,
+      "loss": 0.2641,
+      "step": 1867
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.7289921045303345,
+      "learning_rate": 0.0004485043994868082,
+      "loss": 0.2308,
+      "step": 1868
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.8416182398796082,
+      "learning_rate": 0.0004483438647853282,
+      "loss": 0.2419,
+      "step": 1869
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.7468903660774231,
+      "learning_rate": 0.0004481832738382666,
+      "loss": 0.2178,
+      "step": 1870
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.707642138004303,
+      "learning_rate": 0.00044802262670651255,
+      "loss": 0.1542,
+      "step": 1871
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.8503418564796448,
+      "learning_rate": 0.0004478619234509763,
+      "loss": 0.22,
+      "step": 1872
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 1.0577455759048462,
+      "learning_rate": 0.00044770116413258974,
+      "loss": 0.3233,
+      "step": 1873
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.8245779871940613,
+      "learning_rate": 0.0004475403488123057,
+      "loss": 0.1902,
+      "step": 1874
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 1.0501885414123535,
+      "learning_rate": 0.0004473794775510986,
+      "loss": 0.2953,
+      "step": 1875
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.6955844759941101,
+      "learning_rate": 0.00044721855040996374,
+      "loss": 0.1942,
+      "step": 1876
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.9016565084457397,
+      "learning_rate": 0.0004470575674499178,
+      "loss": 0.2424,
+      "step": 1877
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.7680657505989075,
+      "learning_rate": 0.00044689652873199846,
+      "loss": 0.2244,
+      "step": 1878
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.8058192729949951,
+      "learning_rate": 0.0004467354343172647,
+      "loss": 0.2792,
+      "step": 1879
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.6282344460487366,
+      "learning_rate": 0.00044657428426679666,
+      "loss": 0.1739,
+      "step": 1880
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.740366518497467,
+      "learning_rate": 0.0004464130786416953,
+      "loss": 0.2061,
+      "step": 1881
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.706928014755249,
+      "learning_rate": 0.000446251817503083,
+      "loss": 0.2362,
+      "step": 1882
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.6209346055984497,
+      "learning_rate": 0.00044609050091210287,
+      "loss": 0.154,
+      "step": 1883
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.797219455242157,
+      "learning_rate": 0.0004459291289299192,
+      "loss": 0.1994,
+      "step": 1884
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.8410423994064331,
+      "learning_rate": 0.00044576770161771734,
+      "loss": 0.2203,
+      "step": 1885
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.6617290377616882,
+      "learning_rate": 0.00044560621903670346,
+      "loss": 0.2365,
+      "step": 1886
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.7948521971702576,
+      "learning_rate": 0.00044544468124810494,
+      "loss": 0.2838,
+      "step": 1887
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.7432469129562378,
+      "learning_rate": 0.0004452830883131697,
+      "loss": 0.2094,
+      "step": 1888
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.7609319686889648,
+      "learning_rate": 0.000445121440293167,
+      "loss": 0.2505,
+      "step": 1889
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.7202736139297485,
+      "learning_rate": 0.00044495973724938674,
+      "loss": 0.1827,
+      "step": 1890
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.7481203079223633,
+      "learning_rate": 0.00044479797924313957,
+      "loss": 0.2316,
+      "step": 1891
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.7697124481201172,
+      "learning_rate": 0.0004446361663357573,
+      "loss": 0.2029,
+      "step": 1892
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.7221834659576416,
+      "learning_rate": 0.00044447429858859237,
+      "loss": 0.1756,
+      "step": 1893
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.8437337279319763,
+      "learning_rate": 0.000444312376063018,
+      "loss": 0.2514,
+      "step": 1894
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.8467068672180176,
+      "learning_rate": 0.00044415039882042816,
+      "loss": 0.2446,
+      "step": 1895
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.7756673097610474,
+      "learning_rate": 0.0004439883669222377,
+      "loss": 0.2322,
+      "step": 1896
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.771094560623169,
+      "learning_rate": 0.00044382628042988214,
+      "loss": 0.246,
+      "step": 1897
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.6726484298706055,
+      "learning_rate": 0.00044366413940481747,
+      "loss": 0.1607,
+      "step": 1898
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.6433727741241455,
+      "learning_rate": 0.00044350194390852085,
+      "loss": 0.201,
+      "step": 1899
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.7996774315834045,
+      "learning_rate": 0.00044333969400248965,
+      "loss": 0.1881,
+      "step": 1900
+    },
+    {
+      "epoch": 3.61,
+      "eval_blimp_filtered_avg": 0.5467164179104478,
+      "eval_blimp_filtered_std": 0.0057841328285657065,
+      "step": 1900
+    },
+    {
+      "epoch": 3.61,
+      "eval_blimp_supplement_avg": 0.5323275862068966,
+      "eval_blimp_supplement_std": 0.023142785420428583,
+      "step": 1900
+    },
+    {
+      "epoch": 3.61,
+      "eval_vqa_filtered_avg": 0.2,
+      "eval_vqa_filtered_std": 0.040201512610368445,
+      "step": 1900
+    },
+    {
+      "epoch": 3.61,
+      "eval_winoground_filtered_avg": 0.47,
+      "eval_winoground_filtered_std": 0.0501613558046592,
+      "step": 1900
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.9253387451171875,
+      "learning_rate": 0.00044317738974824193,
+      "loss": 0.295,
+      "step": 1901
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.7896431684494019,
+      "learning_rate": 0.0004430150312073166,
+      "loss": 0.2738,
+      "step": 1902
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.7488947510719299,
+      "learning_rate": 0.00044285261844127295,
+      "loss": 0.1607,
+      "step": 1903
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.8468368053436279,
+      "learning_rate": 0.000442690151511691,
+      "loss": 0.3364,
+      "step": 1904
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.690288782119751,
+      "learning_rate": 0.0004425276304801709,
+      "loss": 0.1957,
+      "step": 1905
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.6565365791320801,
+      "learning_rate": 0.0004423650554083338,
+      "loss": 0.1667,
+      "step": 1906
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.6743531227111816,
+      "learning_rate": 0.0004422024263578212,
+      "loss": 0.171,
+      "step": 1907
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.7053942680358887,
+      "learning_rate": 0.0004420397433902948,
+      "loss": 0.2142,
+      "step": 1908
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.8089833855628967,
+      "learning_rate": 0.00044187700656743716,
+      "loss": 0.2829,
+      "step": 1909
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.5970751047134399,
+      "learning_rate": 0.00044171421595095087,
+      "loss": 0.1732,
+      "step": 1910
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.7638328671455383,
+      "learning_rate": 0.0004415513716025592,
+      "loss": 0.239,
+      "step": 1911
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.6654527187347412,
+      "learning_rate": 0.0004413884735840057,
+      "loss": 0.2144,
+      "step": 1912
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.7514135241508484,
+      "learning_rate": 0.0004412255219570543,
+      "loss": 0.2047,
+      "step": 1913
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.836449921131134,
+      "learning_rate": 0.00044106251678348903,
+      "loss": 0.2254,
+      "step": 1914
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.8851816654205322,
+      "learning_rate": 0.0004408994581251145,
+      "loss": 0.2448,
+      "step": 1915
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.5841615200042725,
+      "learning_rate": 0.00044073634604375567,
+      "loss": 0.1323,
+      "step": 1916
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.8587937355041504,
+      "learning_rate": 0.0004405731806012573,
+      "loss": 0.2735,
+      "step": 1917
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.7121824026107788,
+      "learning_rate": 0.00044040996185948486,
+      "loss": 0.1565,
+      "step": 1918
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.8025166392326355,
+      "learning_rate": 0.0004402466898803238,
+      "loss": 0.2192,
+      "step": 1919
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.7203481793403625,
+      "learning_rate": 0.0004400833647256798,
+      "loss": 0.1957,
+      "step": 1920
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.7919506430625916,
+      "learning_rate": 0.0004399199864574786,
+      "loss": 0.2022,
+      "step": 1921
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.6177682876586914,
+      "learning_rate": 0.0004397565551376664,
+      "loss": 0.1501,
+      "step": 1922
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.6422441601753235,
+      "learning_rate": 0.0004395930708282091,
+      "loss": 0.1826,
+      "step": 1923
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.6111581921577454,
+      "learning_rate": 0.0004394295335910929,
+      "loss": 0.1516,
+      "step": 1924
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.7065783143043518,
+      "learning_rate": 0.00043926594348832424,
+      "loss": 0.2187,
+      "step": 1925
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.709506630897522,
+      "learning_rate": 0.00043910230058192916,
+      "loss": 0.1821,
+      "step": 1926
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.6595371961593628,
+      "learning_rate": 0.00043893860493395416,
+      "loss": 0.1636,
+      "step": 1927
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.7418705224990845,
+      "learning_rate": 0.0004387748566064655,
+      "loss": 0.2444,
+      "step": 1928
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.6724249124526978,
+      "learning_rate": 0.00043861105566154947,
+      "loss": 0.2183,
+      "step": 1929
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.7330010533332825,
+      "learning_rate": 0.00043844720216131217,
+      "loss": 0.2325,
+      "step": 1930
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.720460832118988,
+      "learning_rate": 0.00043828329616787997,
+      "loss": 0.186,
+      "step": 1931
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.6743378639221191,
+      "learning_rate": 0.00043811933774339884,
+      "loss": 0.18,
+      "step": 1932
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.8342899084091187,
+      "learning_rate": 0.0004379553269500348,
+      "loss": 0.2167,
+      "step": 1933
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.7605745196342468,
+      "learning_rate": 0.00043779126384997354,
+      "loss": 0.2239,
+      "step": 1934
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.5802068710327148,
+      "learning_rate": 0.0004376271485054208,
+      "loss": 0.1232,
+      "step": 1935
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.7242363095283508,
+      "learning_rate": 0.0004374629809786019,
+      "loss": 0.2223,
+      "step": 1936
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.6460543870925903,
+      "learning_rate": 0.00043729876133176225,
+      "loss": 0.1613,
+      "step": 1937
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.9159910678863525,
+      "learning_rate": 0.00043713448962716666,
+      "loss": 0.3205,
+      "step": 1938
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.6947113871574402,
+      "learning_rate": 0.00043697016592710003,
+      "loss": 0.2175,
+      "step": 1939
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.8850601315498352,
+      "learning_rate": 0.0004368057902938666,
+      "loss": 0.2295,
+      "step": 1940
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.7543456554412842,
+      "learning_rate": 0.0004366413627897907,
+      "loss": 0.2082,
+      "step": 1941
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.6688163876533508,
+      "learning_rate": 0.00043647688347721607,
+      "loss": 0.2243,
+      "step": 1942
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.8034172654151917,
+      "learning_rate": 0.00043631235241850617,
+      "loss": 0.2641,
+      "step": 1943
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.6980229616165161,
+      "learning_rate": 0.0004361477696760439,
+      "loss": 0.1976,
+      "step": 1944
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.6579756140708923,
+      "learning_rate": 0.0004359831353122322,
+      "loss": 0.1833,
+      "step": 1945
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.675949215888977,
+      "learning_rate": 0.00043581844938949315,
+      "loss": 0.1992,
+      "step": 1946
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.6434425115585327,
+      "learning_rate": 0.0004356537119702684,
+      "loss": 0.1928,
+      "step": 1947
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.7523457407951355,
+      "learning_rate": 0.00043548892311701955,
+      "loss": 0.2566,
+      "step": 1948
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.6517976522445679,
+      "learning_rate": 0.0004353240828922272,
+      "loss": 0.1721,
+      "step": 1949
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.7992874383926392,
+      "learning_rate": 0.00043515919135839176,
+      "loss": 0.2287,
+      "step": 1950
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.5808484554290771,
+      "learning_rate": 0.00043499424857803296,
+      "loss": 0.158,
+      "step": 1951
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.7970479726791382,
+      "learning_rate": 0.0004348292546136899,
+      "loss": 0.2164,
+      "step": 1952
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.5550775527954102,
+      "learning_rate": 0.0004346642095279211,
+      "loss": 0.0948,
+      "step": 1953
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.7288534641265869,
+      "learning_rate": 0.0004344991133833047,
+      "loss": 0.2199,
+      "step": 1954
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.9151400923728943,
+      "learning_rate": 0.0004343339662424379,
+      "loss": 0.2674,
+      "step": 1955
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.9504279494285583,
+      "learning_rate": 0.00043416876816793743,
+      "loss": 0.2746,
+      "step": 1956
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.8660331964492798,
+      "learning_rate": 0.0004340035192224392,
+      "loss": 0.1848,
+      "step": 1957
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.792060911655426,
+      "learning_rate": 0.0004338382194685985,
+      "loss": 0.2365,
+      "step": 1958
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 0.7538288235664368,
+      "learning_rate": 0.00043367286896908984,
+      "loss": 0.2446,
+      "step": 1959
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.6224275231361389,
+      "learning_rate": 0.000433507467786607,
+      "loss": 0.1717,
+      "step": 1960
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.6071409583091736,
+      "learning_rate": 0.0004333420159838629,
+      "loss": 0.167,
+      "step": 1961
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.6527204513549805,
+      "learning_rate": 0.00043317651362358985,
+      "loss": 0.2059,
+      "step": 1962
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.7043445706367493,
+      "learning_rate": 0.00043301096076853895,
+      "loss": 0.2034,
+      "step": 1963
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.700987696647644,
+      "learning_rate": 0.00043284535748148095,
+      "loss": 0.1353,
+      "step": 1964
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.7581907510757446,
+      "learning_rate": 0.0004326797038252053,
+      "loss": 0.2203,
+      "step": 1965
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.626243531703949,
+      "learning_rate": 0.0004325139998625208,
+      "loss": 0.1624,
+      "step": 1966
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.722376823425293,
+      "learning_rate": 0.0004323482456562551,
+      "loss": 0.1542,
+      "step": 1967
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.5801939368247986,
+      "learning_rate": 0.00043218244126925516,
+      "loss": 0.1339,
+      "step": 1968
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.7412282824516296,
+      "learning_rate": 0.00043201658676438677,
+      "loss": 0.2609,
+      "step": 1969
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.7019753456115723,
+      "learning_rate": 0.00043185068220453484,
+      "loss": 0.1799,
+      "step": 1970
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.7769423723220825,
+      "learning_rate": 0.0004316847276526031,
+      "loss": 0.196,
+      "step": 1971
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.6296120285987854,
+      "learning_rate": 0.0004315187231715145,
+      "loss": 0.1576,
+      "step": 1972
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.8018008470535278,
+      "learning_rate": 0.0004313526688242107,
+      "loss": 0.2124,
+      "step": 1973
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.6935921907424927,
+      "learning_rate": 0.0004311865646736523,
+      "loss": 0.1633,
+      "step": 1974
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.654003381729126,
+      "learning_rate": 0.0004310204107828189,
+      "loss": 0.1805,
+      "step": 1975
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.7695251107215881,
+      "learning_rate": 0.0004308542072147086,
+      "loss": 0.223,
+      "step": 1976
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.7324780225753784,
+      "learning_rate": 0.00043068795403233893,
+      "loss": 0.217,
+      "step": 1977
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.695796012878418,
+      "learning_rate": 0.00043052165129874584,
+      "loss": 0.2116,
+      "step": 1978
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.8567126989364624,
+      "learning_rate": 0.00043035529907698393,
+      "loss": 0.2563,
+      "step": 1979
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.6741450428962708,
+      "learning_rate": 0.0004301888974301271,
+      "loss": 0.2169,
+      "step": 1980
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.7490719556808472,
+      "learning_rate": 0.0004300224464212675,
+      "loss": 0.1823,
+      "step": 1981
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.8472000956535339,
+      "learning_rate": 0.0004298559461135161,
+      "loss": 0.28,
+      "step": 1982
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.5287483930587769,
+      "learning_rate": 0.0004296893965700026,
+      "loss": 0.1595,
+      "step": 1983
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.7403092384338379,
+      "learning_rate": 0.0004295227978538756,
+      "loss": 0.2429,
+      "step": 1984
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.6322824954986572,
+      "learning_rate": 0.00042935615002830194,
+      "loss": 0.1567,
+      "step": 1985
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.7314352989196777,
+      "learning_rate": 0.0004291894531564673,
+      "loss": 0.1864,
+      "step": 1986
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.7305060625076294,
+      "learning_rate": 0.0004290227073015759,
+      "loss": 0.2115,
+      "step": 1987
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.6552340984344482,
+      "learning_rate": 0.00042885591252685075,
+      "loss": 0.1368,
+      "step": 1988
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.8261328339576721,
+      "learning_rate": 0.000428689068895533,
+      "loss": 0.2705,
+      "step": 1989
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.7471053600311279,
+      "learning_rate": 0.00042852217647088264,
+      "loss": 0.248,
+      "step": 1990
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.7925482988357544,
+      "learning_rate": 0.00042835523531617806,
+      "loss": 0.2321,
+      "step": 1991
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.8159483075141907,
+      "learning_rate": 0.00042818824549471616,
+      "loss": 0.1852,
+      "step": 1992
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.6250635385513306,
+      "learning_rate": 0.00042802120706981214,
+      "loss": 0.157,
+      "step": 1993
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.7639796733856201,
+      "learning_rate": 0.00042785412010479994,
+      "loss": 0.2814,
+      "step": 1994
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.6358375549316406,
+      "learning_rate": 0.0004276869846630316,
+      "loss": 0.1497,
+      "step": 1995
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.5837895274162292,
+      "learning_rate": 0.00042751980080787765,
+      "loss": 0.1749,
+      "step": 1996
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.7442068457603455,
+      "learning_rate": 0.00042735256860272695,
+      "loss": 0.2255,
+      "step": 1997
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.7662954330444336,
+      "learning_rate": 0.00042718528811098685,
+      "loss": 0.2019,
+      "step": 1998
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.6226879358291626,
+      "learning_rate": 0.00042701795939608273,
+      "loss": 0.1721,
+      "step": 1999
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.6708007454872131,
+      "learning_rate": 0.0004268505825214585,
+      "loss": 0.1637,
+      "step": 2000
+    },
+    {
+      "epoch": 3.8,
+      "eval_blimp_filtered_avg": 0.5407462686567164,
+      "eval_blimp_filtered_std": 0.005861690075877737,
+      "step": 2000
+    },
+    {
+      "epoch": 3.8,
+      "eval_blimp_supplement_avg": 0.5237068965517241,
+      "eval_blimp_supplement_std": 0.023146842799946585,
+      "step": 2000
+    },
+    {
+      "epoch": 3.8,
+      "eval_vqa_filtered_avg": 0.11,
+      "eval_vqa_filtered_std": 0.031446603773522035,
+      "step": 2000
+    },
+    {
+      "epoch": 3.8,
+      "eval_winoground_filtered_avg": 0.46,
+      "eval_winoground_filtered_std": 0.05009082659620333,
+      "step": 2000
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.7335052490234375,
+      "learning_rate": 0.00042668315755057623,
+      "loss": 0.1711,
+      "step": 2001
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.8859219551086426,
+      "learning_rate": 0.0004265156845469162,
+      "loss": 0.2802,
+      "step": 2002
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.6194216012954712,
+      "learning_rate": 0.00042634816357397696,
+      "loss": 0.1456,
+      "step": 2003
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.9638639688491821,
+      "learning_rate": 0.0004261805946952753,
+      "loss": 0.2793,
+      "step": 2004
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.8402900099754333,
+      "learning_rate": 0.0004260129779743459,
+      "loss": 0.2203,
+      "step": 2005
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.7387112975120544,
+      "learning_rate": 0.00042584531347474204,
+      "loss": 0.1403,
+      "step": 2006
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.6464426517486572,
+      "learning_rate": 0.0004256776012600345,
+      "loss": 0.1501,
+      "step": 2007
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.755425751209259,
+      "learning_rate": 0.0004255098413938129,
+      "loss": 0.2167,
+      "step": 2008
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.6783706545829773,
+      "learning_rate": 0.0004253420339396842,
+      "loss": 0.1675,
+      "step": 2009
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.781607449054718,
+      "learning_rate": 0.00042517417896127387,
+      "loss": 0.1684,
+      "step": 2010
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.9447340369224548,
+      "learning_rate": 0.00042500627652222533,
+      "loss": 0.2632,
+      "step": 2011
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.6192470192909241,
+      "learning_rate": 0.00042483832668619977,
+      "loss": 0.1254,
+      "step": 2012
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.7143309712409973,
+      "learning_rate": 0.0004246703295168766,
+      "loss": 0.1883,
+      "step": 2013
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.8418563604354858,
+      "learning_rate": 0.00042450228507795316,
+      "loss": 0.2419,
+      "step": 2014
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.816008448600769,
+      "learning_rate": 0.00042433419343314445,
+      "loss": 0.2309,
+      "step": 2015
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.7812739610671997,
+      "learning_rate": 0.00042416605464618375,
+      "loss": 0.2357,
+      "step": 2016
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.7034612894058228,
+      "learning_rate": 0.000423997868780822,
+      "loss": 0.2228,
+      "step": 2017
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.5651074647903442,
+      "learning_rate": 0.00042382963590082784,
+      "loss": 0.1068,
+      "step": 2018
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.7861596941947937,
+      "learning_rate": 0.0004236613560699881,
+      "loss": 0.2067,
+      "step": 2019
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.6275096535682678,
+      "learning_rate": 0.00042349302935210705,
+      "loss": 0.1747,
+      "step": 2020
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.6645784974098206,
+      "learning_rate": 0.00042332465581100705,
+      "loss": 0.1796,
+      "step": 2021
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.6683154702186584,
+      "learning_rate": 0.0004231562355105279,
+      "loss": 0.171,
+      "step": 2022
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.6058634519577026,
+      "learning_rate": 0.00042298776851452746,
+      "loss": 0.1671,
+      "step": 2023
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.8612411618232727,
+      "learning_rate": 0.0004228192548868811,
+      "loss": 0.2182,
+      "step": 2024
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.8866882920265198,
+      "learning_rate": 0.00042265069469148186,
+      "loss": 0.3193,
+      "step": 2025
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.7684945464134216,
+      "learning_rate": 0.0004224820879922405,
+      "loss": 0.1852,
+      "step": 2026
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.8039574027061462,
+      "learning_rate": 0.00042231343485308546,
+      "loss": 0.2765,
+      "step": 2027
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.7666494846343994,
+      "learning_rate": 0.00042214473533796256,
+      "loss": 0.1424,
+      "step": 2028
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.6655550003051758,
+      "learning_rate": 0.00042197598951083557,
+      "loss": 0.165,
+      "step": 2029
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.8453596234321594,
+      "learning_rate": 0.00042180719743568543,
+      "loss": 0.2378,
+      "step": 2030
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.7555413842201233,
+      "learning_rate": 0.00042163835917651086,
+      "loss": 0.2058,
+      "step": 2031
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 0.8417167663574219,
+      "learning_rate": 0.000421469474797328,
+      "loss": 0.1889,
+      "step": 2032
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.7884296178817749,
+      "learning_rate": 0.0004213005443621707,
+      "loss": 0.2225,
+      "step": 2033
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.8812756538391113,
+      "learning_rate": 0.00042113156793508984,
+      "loss": 0.2408,
+      "step": 2034
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.7777116894721985,
+      "learning_rate": 0.000420962545580154,
+      "loss": 0.2057,
+      "step": 2035
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.7002843022346497,
+      "learning_rate": 0.0004207934773614493,
+      "loss": 0.2078,
+      "step": 2036
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.6729376316070557,
+      "learning_rate": 0.00042062436334307907,
+      "loss": 0.1899,
+      "step": 2037
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.7631048560142517,
+      "learning_rate": 0.00042045520358916384,
+      "loss": 0.21,
+      "step": 2038
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.7736886739730835,
+      "learning_rate": 0.0004202859981638419,
+      "loss": 0.3347,
+      "step": 2039
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.7264257669448853,
+      "learning_rate": 0.00042011674713126864,
+      "loss": 0.2369,
+      "step": 2040
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.6993554830551147,
+      "learning_rate": 0.0004199474505556167,
+      "loss": 0.1964,
+      "step": 2041
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.7944740653038025,
+      "learning_rate": 0.00041977810850107574,
+      "loss": 0.2607,
+      "step": 2042
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.8465865254402161,
+      "learning_rate": 0.00041960872103185334,
+      "loss": 0.2143,
+      "step": 2043
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.6680233478546143,
+      "learning_rate": 0.0004194392882121737,
+      "loss": 0.1763,
+      "step": 2044
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.7670724987983704,
+      "learning_rate": 0.0004192698101062784,
+      "loss": 0.226,
+      "step": 2045
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.7530050277709961,
+      "learning_rate": 0.00041910028677842647,
+      "loss": 0.1783,
+      "step": 2046
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.7233293056488037,
+      "learning_rate": 0.00041893071829289356,
+      "loss": 0.2365,
+      "step": 2047
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.8092796802520752,
+      "learning_rate": 0.0004187611047139727,
+      "loss": 0.2654,
+      "step": 2048
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.9415621757507324,
+      "learning_rate": 0.0004185914461059742,
+      "loss": 0.3769,
+      "step": 2049
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.8434984087944031,
+      "learning_rate": 0.0004184217425332252,
+      "loss": 0.214,
+      "step": 2050
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.6460445523262024,
+      "learning_rate": 0.00041825199406007,
+      "loss": 0.1396,
+      "step": 2051
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.7331963777542114,
+      "learning_rate": 0.00041808220075086977,
+      "loss": 0.2156,
+      "step": 2052
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.5924848914146423,
+      "learning_rate": 0.00041791236267000285,
+      "loss": 0.1593,
+      "step": 2053
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.591860294342041,
+      "learning_rate": 0.0004177424798818646,
+      "loss": 0.1502,
+      "step": 2054
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.7289583683013916,
+      "learning_rate": 0.00041757255245086707,
+      "loss": 0.1863,
+      "step": 2055
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.6765662431716919,
+      "learning_rate": 0.0004174025804414394,
+      "loss": 0.2078,
+      "step": 2056
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.4969986379146576,
+      "learning_rate": 0.00041723256391802785,
+      "loss": 0.1367,
+      "step": 2057
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.7648352384567261,
+      "learning_rate": 0.000417062502945095,
+      "loss": 0.2748,
+      "step": 2058
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.8076521158218384,
+      "learning_rate": 0.000416892397587121,
+      "loss": 0.2283,
+      "step": 2059
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.7752401232719421,
+      "learning_rate": 0.00041672224790860213,
+      "loss": 0.2271,
+      "step": 2060
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.7618404030799866,
+      "learning_rate": 0.00041655205397405195,
+      "loss": 0.2128,
+      "step": 2061
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.7256850600242615,
+      "learning_rate": 0.00041638181584800066,
+      "loss": 0.2321,
+      "step": 2062
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.6284517645835876,
+      "learning_rate": 0.0004162115335949951,
+      "loss": 0.1568,
+      "step": 2063
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.8240419030189514,
+      "learning_rate": 0.000416041207279599,
+      "loss": 0.2681,
+      "step": 2064
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.44685205817222595,
+      "learning_rate": 0.00041587083696639277,
+      "loss": 0.1186,
+      "step": 2065
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.7902135252952576,
+      "learning_rate": 0.0004157004227199734,
+      "loss": 0.2129,
+      "step": 2066
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.701676607131958,
+      "learning_rate": 0.00041552996460495465,
+      "loss": 0.1521,
+      "step": 2067
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.6692916750907898,
+      "learning_rate": 0.00041535946268596676,
+      "loss": 0.1696,
+      "step": 2068
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.6490834951400757,
+      "learning_rate": 0.00041518891702765686,
+      "loss": 0.1611,
+      "step": 2069
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.7425786256790161,
+      "learning_rate": 0.0004150183276946884,
+      "loss": 0.2011,
+      "step": 2070
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.804344892501831,
+      "learning_rate": 0.00041484769475174147,
+      "loss": 0.2662,
+      "step": 2071
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.5723490118980408,
+      "learning_rate": 0.00041467701826351273,
+      "loss": 0.1293,
+      "step": 2072
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.6959919929504395,
+      "learning_rate": 0.0004145062982947154,
+      "loss": 0.2224,
+      "step": 2073
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.748171865940094,
+      "learning_rate": 0.000414335534910079,
+      "loss": 0.2451,
+      "step": 2074
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.7793415188789368,
+      "learning_rate": 0.0004141647281743497,
+      "loss": 0.2944,
+      "step": 2075
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.5262612700462341,
+      "learning_rate": 0.00041399387815228996,
+      "loss": 0.0983,
+      "step": 2076
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.7707334160804749,
+      "learning_rate": 0.0004138229849086787,
+      "loss": 0.2294,
+      "step": 2077
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.5858818292617798,
+      "learning_rate": 0.00041365204850831144,
+      "loss": 0.1477,
+      "step": 2078
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.6947721242904663,
+      "learning_rate": 0.00041348106901599967,
+      "loss": 0.2136,
+      "step": 2079
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.5880098342895508,
+      "learning_rate": 0.0004133100464965715,
+      "loss": 0.1552,
+      "step": 2080
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.7437090277671814,
+      "learning_rate": 0.0004131389810148713,
+      "loss": 0.219,
+      "step": 2081
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.8400521874427795,
+      "learning_rate": 0.0004129678726357597,
+      "loss": 0.2626,
+      "step": 2082
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.654587984085083,
+      "learning_rate": 0.00041279672142411353,
+      "loss": 0.1941,
+      "step": 2083
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.5364904403686523,
+      "learning_rate": 0.000412625527444826,
+      "loss": 0.1222,
+      "step": 2084
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.7094141840934753,
+      "learning_rate": 0.0004124542907628065,
+      "loss": 0.2143,
+      "step": 2085
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 0.6904459595680237,
+      "learning_rate": 0.00041228301144298057,
+      "loss": 0.2079,
+      "step": 2086
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 0.9072972536087036,
+      "learning_rate": 0.0004121116895502898,
+      "loss": 0.2384,
+      "step": 2087
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 0.7342488765716553,
+      "learning_rate": 0.0004119403251496922,
+      "loss": 0.2194,
+      "step": 2088
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 0.7403815388679504,
+      "learning_rate": 0.00041176891830616165,
+      "loss": 0.1686,
+      "step": 2089
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 0.7273654937744141,
+      "learning_rate": 0.0004115974690846883,
+      "loss": 0.1965,
+      "step": 2090
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.8523355722427368,
+      "learning_rate": 0.0004114259775502781,
+      "loss": 0.2914,
+      "step": 2091
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.7413622140884399,
+      "learning_rate": 0.00041125444376795343,
+      "loss": 0.2629,
+      "step": 2092
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.5585724115371704,
+      "learning_rate": 0.00041108286780275246,
+      "loss": 0.1352,
+      "step": 2093
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.6443831324577332,
+      "learning_rate": 0.00041091124971972924,
+      "loss": 0.1883,
+      "step": 2094
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.7514685988426208,
+      "learning_rate": 0.00041073958958395406,
+      "loss": 0.2058,
+      "step": 2095
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.7157757878303528,
+      "learning_rate": 0.0004105678874605131,
+      "loss": 0.2004,
+      "step": 2096
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.7385727763175964,
+      "learning_rate": 0.0004103961434145082,
+      "loss": 0.2145,
+      "step": 2097
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.7565996050834656,
+      "learning_rate": 0.0004102243575110573,
+      "loss": 0.1623,
+      "step": 2098
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.6105378866195679,
+      "learning_rate": 0.0004100525298152943,
+      "loss": 0.1367,
+      "step": 2099
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.6439101696014404,
+      "learning_rate": 0.0004098806603923687,
+      "loss": 0.1341,
+      "step": 2100
+    },
+    {
+      "epoch": 3.99,
+      "eval_blimp_filtered_avg": 0.542089552238806,
+      "eval_blimp_filtered_std": 0.005769662468258094,
+      "step": 2100
+    },
+    {
+      "epoch": 3.99,
+      "eval_blimp_supplement_avg": 0.540948275862069,
+      "eval_blimp_supplement_std": 0.02300874559762897,
+      "step": 2100
+    },
+    {
+      "epoch": 3.99,
+      "eval_vqa_filtered_avg": 0.2,
+      "eval_vqa_filtered_std": 0.04020151261036845,
+      "step": 2100
+    },
+    {
+      "epoch": 3.99,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 2100
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.8754507899284363,
+      "learning_rate": 0.00040970874930744606,
+      "loss": 0.2375,
+      "step": 2101
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.688088595867157,
+      "learning_rate": 0.0004095367966257077,
+      "loss": 0.1972,
+      "step": 2102
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.8064619302749634,
+      "learning_rate": 0.00040936480241235044,
+      "loss": 0.2233,
+      "step": 2103
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.6206127405166626,
+      "learning_rate": 0.00040919276673258715,
+      "loss": 0.129,
+      "step": 2104
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.7309362292289734,
+      "learning_rate": 0.0004090206896516464,
+      "loss": 0.1183,
+      "step": 2105
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.5982339978218079,
+      "learning_rate": 0.0004088485712347723,
+      "loss": 0.0975,
+      "step": 2106
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.6308634281158447,
+      "learning_rate": 0.00040867641154722465,
+      "loss": 0.1303,
+      "step": 2107
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.49192479252815247,
+      "learning_rate": 0.00040850421065427914,
+      "loss": 0.0897,
+      "step": 2108
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.5120988488197327,
+      "learning_rate": 0.0004083319686212268,
+      "loss": 0.0867,
+      "step": 2109
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.7253944873809814,
+      "learning_rate": 0.00040815968551337437,
+      "loss": 0.1536,
+      "step": 2110
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.7093410491943359,
+      "learning_rate": 0.0004079873613960441,
+      "loss": 0.138,
+      "step": 2111
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.7062446475028992,
+      "learning_rate": 0.00040781499633457403,
+      "loss": 0.1247,
+      "step": 2112
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.5792815685272217,
+      "learning_rate": 0.00040764259039431725,
+      "loss": 0.0982,
+      "step": 2113
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.5664187669754028,
+      "learning_rate": 0.00040747014364064295,
+      "loss": 0.1057,
+      "step": 2114
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.748189389705658,
+      "learning_rate": 0.0004072976561389353,
+      "loss": 0.1463,
+      "step": 2115
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.7681527137756348,
+      "learning_rate": 0.0004071251279545942,
+      "loss": 0.1429,
+      "step": 2116
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.6443457007408142,
+      "learning_rate": 0.0004069525591530347,
+      "loss": 0.1409,
+      "step": 2117
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.6721271872520447,
+      "learning_rate": 0.00040677994979968765,
+      "loss": 0.1191,
+      "step": 2118
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.646342933177948,
+      "learning_rate": 0.00040660729995999897,
+      "loss": 0.1069,
+      "step": 2119
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.5387141108512878,
+      "learning_rate": 0.00040643460969943,
+      "loss": 0.0862,
+      "step": 2120
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.6352238655090332,
+      "learning_rate": 0.0004062618790834575,
+      "loss": 0.1216,
+      "step": 2121
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.6716883182525635,
+      "learning_rate": 0.0004060891081775733,
+      "loss": 0.1088,
+      "step": 2122
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.6244571208953857,
+      "learning_rate": 0.0004059162970472847,
+      "loss": 0.1368,
+      "step": 2123
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.6330448389053345,
+      "learning_rate": 0.00040574344575811436,
+      "loss": 0.1131,
+      "step": 2124
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.5787625312805176,
+      "learning_rate": 0.00040557055437559996,
+      "loss": 0.1267,
+      "step": 2125
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.7390915155410767,
+      "learning_rate": 0.0004053976229652942,
+      "loss": 0.156,
+      "step": 2126
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.5939272046089172,
+      "learning_rate": 0.0004052246515927656,
+      "loss": 0.1056,
+      "step": 2127
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.5305801630020142,
+      "learning_rate": 0.0004050516403235971,
+      "loss": 0.0978,
+      "step": 2128
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.640679657459259,
+      "learning_rate": 0.0004048785892233873,
+      "loss": 0.117,
+      "step": 2129
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.6518338322639465,
+      "learning_rate": 0.00040470549835774954,
+      "loss": 0.1332,
+      "step": 2130
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.5508995056152344,
+      "learning_rate": 0.00040453236779231254,
+      "loss": 0.108,
+      "step": 2131
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.6370252966880798,
+      "learning_rate": 0.0004043591975927198,
+      "loss": 0.1202,
+      "step": 2132
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.6855735182762146,
+      "learning_rate": 0.0004041859878246301,
+      "loss": 0.1258,
+      "step": 2133
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.5619186758995056,
+      "learning_rate": 0.000404012738553717,
+      "loss": 0.1323,
+      "step": 2134
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.6208124160766602,
+      "learning_rate": 0.00040383944984566915,
+      "loss": 0.0974,
+      "step": 2135
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.5874891877174377,
+      "learning_rate": 0.00040366612176619013,
+      "loss": 0.121,
+      "step": 2136
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.5135067105293274,
+      "learning_rate": 0.0004034927543809985,
+      "loss": 0.1032,
+      "step": 2137
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.62488853931427,
+      "learning_rate": 0.0004033193477558277,
+      "loss": 0.1277,
+      "step": 2138
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.577577531337738,
+      "learning_rate": 0.000403145901956426,
+      "loss": 0.0987,
+      "step": 2139
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.7395797371864319,
+      "learning_rate": 0.00040297241704855636,
+      "loss": 0.1454,
+      "step": 2140
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.5604895353317261,
+      "learning_rate": 0.0004027988930979971,
+      "loss": 0.1111,
+      "step": 2141
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.47653451561927795,
+      "learning_rate": 0.0004026253301705408,
+      "loss": 0.0785,
+      "step": 2142
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.5344717502593994,
+      "learning_rate": 0.00040245172833199504,
+      "loss": 0.0927,
+      "step": 2143
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.5514301061630249,
+      "learning_rate": 0.00040227808764818217,
+      "loss": 0.1009,
+      "step": 2144
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.6423225998878479,
+      "learning_rate": 0.00040210440818493927,
+      "loss": 0.1248,
+      "step": 2145
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.7451153993606567,
+      "learning_rate": 0.0004019306900081179,
+      "loss": 0.1327,
+      "step": 2146
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.6342983841896057,
+      "learning_rate": 0.0004017569331835847,
+      "loss": 0.108,
+      "step": 2147
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.575021505355835,
+      "learning_rate": 0.00040158313777722063,
+      "loss": 0.124,
+      "step": 2148
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.6633716821670532,
+      "learning_rate": 0.00040140930385492135,
+      "loss": 0.1263,
+      "step": 2149
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.6933222413063049,
+      "learning_rate": 0.0004012354314825974,
+      "loss": 0.1248,
+      "step": 2150
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.6213657855987549,
+      "learning_rate": 0.0004010615207261734,
+      "loss": 0.1106,
+      "step": 2151
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.5865564942359924,
+      "learning_rate": 0.00040088757165158897,
+      "loss": 0.1125,
+      "step": 2152
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.729832112789154,
+      "learning_rate": 0.0004007135843247979,
+      "loss": 0.1488,
+      "step": 2153
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.5331519842147827,
+      "learning_rate": 0.0004005395588117689,
+      "loss": 0.0904,
+      "step": 2154
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.5137259364128113,
+      "learning_rate": 0.00040036549517848477,
+      "loss": 0.0899,
+      "step": 2155
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.5746036171913147,
+      "learning_rate": 0.00040019139349094295,
+      "loss": 0.101,
+      "step": 2156
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.5467351078987122,
+      "learning_rate": 0.00040001725381515527,
+      "loss": 0.0934,
+      "step": 2157
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.6925759315490723,
+      "learning_rate": 0.000399843076217148,
+      "loss": 0.1303,
+      "step": 2158
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.6524451971054077,
+      "learning_rate": 0.0003996688607629616,
+      "loss": 0.1027,
+      "step": 2159
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.5795971155166626,
+      "learning_rate": 0.0003994946075186513,
+      "loss": 0.1076,
+      "step": 2160
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.4877747595310211,
+      "learning_rate": 0.0003993203165502862,
+      "loss": 0.0782,
+      "step": 2161
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.6154145002365112,
+      "learning_rate": 0.00039914598792395,
+      "loss": 0.1178,
+      "step": 2162
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.5283840298652649,
+      "learning_rate": 0.00039897162170574057,
+      "loss": 0.0938,
+      "step": 2163
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.7919886708259583,
+      "learning_rate": 0.00039879721796176994,
+      "loss": 0.1265,
+      "step": 2164
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.721454381942749,
+      "learning_rate": 0.00039862277675816463,
+      "loss": 0.1027,
+      "step": 2165
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.6500721573829651,
+      "learning_rate": 0.00039844829816106515,
+      "loss": 0.0968,
+      "step": 2166
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.6547823548316956,
+      "learning_rate": 0.00039827378223662615,
+      "loss": 0.1288,
+      "step": 2167
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.6055137515068054,
+      "learning_rate": 0.0003980992290510167,
+      "loss": 0.082,
+      "step": 2168
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.6073504686355591,
+      "learning_rate": 0.0003979246386704195,
+      "loss": 0.1044,
+      "step": 2169
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.6827021241188049,
+      "learning_rate": 0.00039775001116103216,
+      "loss": 0.1268,
+      "step": 2170
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.5636672377586365,
+      "learning_rate": 0.0003975753465890655,
+      "loss": 0.0961,
+      "step": 2171
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.6115701794624329,
+      "learning_rate": 0.00039740064502074483,
+      "loss": 0.0973,
+      "step": 2172
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.8162551522254944,
+      "learning_rate": 0.00039722590652230955,
+      "loss": 0.1412,
+      "step": 2173
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.7374301552772522,
+      "learning_rate": 0.0003970511311600129,
+      "loss": 0.1617,
+      "step": 2174
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.5391994118690491,
+      "learning_rate": 0.0003968763190001221,
+      "loss": 0.0829,
+      "step": 2175
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.5712114572525024,
+      "learning_rate": 0.0003967014701089185,
+      "loss": 0.0943,
+      "step": 2176
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.6882869005203247,
+      "learning_rate": 0.00039652658455269706,
+      "loss": 0.102,
+      "step": 2177
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.6553488373756409,
+      "learning_rate": 0.000396351662397767,
+      "loss": 0.1201,
+      "step": 2178
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.6372990012168884,
+      "learning_rate": 0.0003961767037104511,
+      "loss": 0.1129,
+      "step": 2179
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.6226000189781189,
+      "learning_rate": 0.0003960017085570862,
+      "loss": 0.1092,
+      "step": 2180
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.7061742544174194,
+      "learning_rate": 0.0003958266770040229,
+      "loss": 0.1296,
+      "step": 2181
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.6357200145721436,
+      "learning_rate": 0.00039565160911762544,
+      "loss": 0.1121,
+      "step": 2182
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.7205974459648132,
+      "learning_rate": 0.0003954765049642723,
+      "loss": 0.1186,
+      "step": 2183
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.6357987523078918,
+      "learning_rate": 0.0003953013646103552,
+      "loss": 0.114,
+      "step": 2184
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.609764039516449,
+      "learning_rate": 0.00039512618812227973,
+      "loss": 0.0869,
+      "step": 2185
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.5889382362365723,
+      "learning_rate": 0.0003949509755664654,
+      "loss": 0.0987,
+      "step": 2186
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.714795708656311,
+      "learning_rate": 0.0003947757270093453,
+      "loss": 0.1577,
+      "step": 2187
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.6461151838302612,
+      "learning_rate": 0.0003946004425173659,
+      "loss": 0.1456,
+      "step": 2188
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.5184183120727539,
+      "learning_rate": 0.0003944251221569876,
+      "loss": 0.098,
+      "step": 2189
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.6074878573417664,
+      "learning_rate": 0.00039424976599468424,
+      "loss": 0.128,
+      "step": 2190
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.6596125960350037,
+      "learning_rate": 0.0003940743740969434,
+      "loss": 0.1092,
+      "step": 2191
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.5922967195510864,
+      "learning_rate": 0.0003938989465302661,
+      "loss": 0.1035,
+      "step": 2192
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.6435729265213013,
+      "learning_rate": 0.0003937234833611668,
+      "loss": 0.1037,
+      "step": 2193
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.6862114667892456,
+      "learning_rate": 0.0003935479846561736,
+      "loss": 0.1369,
+      "step": 2194
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.7501999735832214,
+      "learning_rate": 0.00039337245048182807,
+      "loss": 0.1366,
+      "step": 2195
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.5986515879631042,
+      "learning_rate": 0.00039319688090468514,
+      "loss": 0.0896,
+      "step": 2196
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.6902726888656616,
+      "learning_rate": 0.0003930212759913132,
+      "loss": 0.0996,
+      "step": 2197
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.7295423150062561,
+      "learning_rate": 0.000392845635808294,
+      "loss": 0.1265,
+      "step": 2198
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.8583167195320129,
+      "learning_rate": 0.00039266996042222277,
+      "loss": 0.1244,
+      "step": 2199
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.6034978628158569,
+      "learning_rate": 0.00039249424989970804,
+      "loss": 0.0957,
+      "step": 2200
+    },
+    {
+      "epoch": 4.18,
+      "eval_blimp_filtered_avg": 0.5486567164179105,
+      "eval_blimp_filtered_std": 0.005783024002486732,
+      "step": 2200
+    },
+    {
+      "epoch": 4.18,
+      "eval_blimp_supplement_avg": 0.5,
+      "eval_blimp_supplement_std": 0.023209831140070733,
+      "step": 2200
+    },
+    {
+      "epoch": 4.18,
+      "eval_vqa_filtered_avg": 0.18,
+      "eval_vqa_filtered_std": 0.03861229196653697,
+      "step": 2200
+    },
+    {
+      "epoch": 4.18,
+      "eval_winoground_filtered_avg": 0.49,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 2200
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 0.501226007938385,
+      "learning_rate": 0.00039231850430737164,
+      "loss": 0.0673,
+      "step": 2201
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.7723246812820435,
+      "learning_rate": 0.0003921427237118486,
+      "loss": 0.1304,
+      "step": 2202
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.8419126868247986,
+      "learning_rate": 0.00039196690817978744,
+      "loss": 0.1283,
+      "step": 2203
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.658560037612915,
+      "learning_rate": 0.0003917910577778498,
+      "loss": 0.0998,
+      "step": 2204
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.651067316532135,
+      "learning_rate": 0.00039161517257271035,
+      "loss": 0.1109,
+      "step": 2205
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 0.5639516115188599,
+      "learning_rate": 0.0003914392526310575,
+      "loss": 0.0959,
+      "step": 2206
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.7931994199752808,
+      "learning_rate": 0.0003912632980195923,
+      "loss": 0.1785,
+      "step": 2207
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.6774352788925171,
+      "learning_rate": 0.000391087308805029,
+      "loss": 0.1127,
+      "step": 2208
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.6180725693702698,
+      "learning_rate": 0.0003909112850540954,
+      "loss": 0.1023,
+      "step": 2209
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.6442916393280029,
+      "learning_rate": 0.0003907352268335319,
+      "loss": 0.1428,
+      "step": 2210
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 0.7152258157730103,
+      "learning_rate": 0.00039055913421009227,
+      "loss": 0.1355,
+      "step": 2211
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.6308382153511047,
+      "learning_rate": 0.0003903830072505431,
+      "loss": 0.1077,
+      "step": 2212
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.7285107374191284,
+      "learning_rate": 0.0003902068460216643,
+      "loss": 0.1264,
+      "step": 2213
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.7395067811012268,
+      "learning_rate": 0.00039003065059024837,
+      "loss": 0.1311,
+      "step": 2214
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.857064425945282,
+      "learning_rate": 0.00038985442102310106,
+      "loss": 0.1868,
+      "step": 2215
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.8554385304450989,
+      "learning_rate": 0.0003896781573870412,
+      "loss": 0.1343,
+      "step": 2216
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 0.6736313700675964,
+      "learning_rate": 0.00038950185974890016,
+      "loss": 0.1434,
+      "step": 2217
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.5530020594596863,
+      "learning_rate": 0.00038932552817552235,
+      "loss": 0.1025,
+      "step": 2218
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.6488381624221802,
+      "learning_rate": 0.00038914916273376525,
+      "loss": 0.1124,
+      "step": 2219
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.6129838824272156,
+      "learning_rate": 0.0003889727634904989,
+      "loss": 0.0895,
+      "step": 2220
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.6773309111595154,
+      "learning_rate": 0.0003887963305126062,
+      "loss": 0.1372,
+      "step": 2221
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 0.7317503094673157,
+      "learning_rate": 0.0003886198638669831,
+      "loss": 0.1593,
+      "step": 2222
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.6877882480621338,
+      "learning_rate": 0.00038844336362053807,
+      "loss": 0.1335,
+      "step": 2223
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.6695659756660461,
+      "learning_rate": 0.00038826682984019223,
+      "loss": 0.1308,
+      "step": 2224
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.6234405040740967,
+      "learning_rate": 0.0003880902625928797,
+      "loss": 0.1066,
+      "step": 2225
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.6832362413406372,
+      "learning_rate": 0.0003879136619455471,
+      "loss": 0.1159,
+      "step": 2226
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 0.6020423769950867,
+      "learning_rate": 0.0003877370279651537,
+      "loss": 0.1049,
+      "step": 2227
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.8546380996704102,
+      "learning_rate": 0.0003875603607186716,
+      "loss": 0.1636,
+      "step": 2228
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.6549291610717773,
+      "learning_rate": 0.00038738366027308527,
+      "loss": 0.0946,
+      "step": 2229
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.584298312664032,
+      "learning_rate": 0.000387206926695392,
+      "loss": 0.0808,
+      "step": 2230
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.6756349802017212,
+      "learning_rate": 0.0003870301600526014,
+      "loss": 0.1097,
+      "step": 2231
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.5899354815483093,
+      "learning_rate": 0.0003868533604117359,
+      "loss": 0.1015,
+      "step": 2232
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.592678964138031,
+      "learning_rate": 0.0003866765278398301,
+      "loss": 0.0983,
+      "step": 2233
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.649116039276123,
+      "learning_rate": 0.00038649966240393135,
+      "loss": 0.1262,
+      "step": 2234
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.5182325839996338,
+      "learning_rate": 0.0003863227641710994,
+      "loss": 0.0998,
+      "step": 2235
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.586738646030426,
+      "learning_rate": 0.00038614583320840636,
+      "loss": 0.0959,
+      "step": 2236
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.6215356588363647,
+      "learning_rate": 0.00038596886958293684,
+      "loss": 0.1029,
+      "step": 2237
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.598528265953064,
+      "learning_rate": 0.0003857918733617878,
+      "loss": 0.0905,
+      "step": 2238
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.6475570797920227,
+      "learning_rate": 0.0003856148446120685,
+      "loss": 0.1361,
+      "step": 2239
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.5558689832687378,
+      "learning_rate": 0.00038543778340090077,
+      "loss": 0.0854,
+      "step": 2240
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.67322838306427,
+      "learning_rate": 0.0003852606897954183,
+      "loss": 0.108,
+      "step": 2241
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.7611307501792908,
+      "learning_rate": 0.0003850835638627675,
+      "loss": 0.1311,
+      "step": 2242
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 0.6097924709320068,
+      "learning_rate": 0.0003849064056701069,
+      "loss": 0.1053,
+      "step": 2243
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.5731884837150574,
+      "learning_rate": 0.00038472921528460703,
+      "loss": 0.1095,
+      "step": 2244
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.48946335911750793,
+      "learning_rate": 0.0003845519927734511,
+      "loss": 0.0724,
+      "step": 2245
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.6059260368347168,
+      "learning_rate": 0.00038437473820383404,
+      "loss": 0.1029,
+      "step": 2246
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.5329364538192749,
+      "learning_rate": 0.0003841974516429632,
+      "loss": 0.1063,
+      "step": 2247
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 0.5585803985595703,
+      "learning_rate": 0.0003840201331580579,
+      "loss": 0.1246,
+      "step": 2248
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.8123360872268677,
+      "learning_rate": 0.0003838427828163497,
+      "loss": 0.179,
+      "step": 2249
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.5752469897270203,
+      "learning_rate": 0.00038366540068508217,
+      "loss": 0.0991,
+      "step": 2250
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.5910701155662537,
+      "learning_rate": 0.00038348798683151094,
+      "loss": 0.0994,
+      "step": 2251
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.6380770802497864,
+      "learning_rate": 0.0003833105413229037,
+      "loss": 0.1101,
+      "step": 2252
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.7234246730804443,
+      "learning_rate": 0.0003831330642265402,
+      "loss": 0.1178,
+      "step": 2253
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.6192782521247864,
+      "learning_rate": 0.0003829555556097118,
+      "loss": 0.1131,
+      "step": 2254
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.5918615460395813,
+      "learning_rate": 0.0003827780155397224,
+      "loss": 0.1029,
+      "step": 2255
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.6804864406585693,
+      "learning_rate": 0.0003826004440838875,
+      "loss": 0.1181,
+      "step": 2256
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.6206170320510864,
+      "learning_rate": 0.0003824228413095343,
+      "loss": 0.0977,
+      "step": 2257
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.5906144380569458,
+      "learning_rate": 0.0003822452072840022,
+      "loss": 0.096,
+      "step": 2258
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 0.6860532760620117,
+      "learning_rate": 0.00038206754207464254,
+      "loss": 0.1459,
+      "step": 2259
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.6666074991226196,
+      "learning_rate": 0.00038188984574881793,
+      "loss": 0.1183,
+      "step": 2260
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.6854516863822937,
+      "learning_rate": 0.0003817121183739034,
+      "loss": 0.1193,
+      "step": 2261
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.8664455413818359,
+      "learning_rate": 0.0003815343600172854,
+      "loss": 0.1675,
+      "step": 2262
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.6370760798454285,
+      "learning_rate": 0.0003813565707463623,
+      "loss": 0.1245,
+      "step": 2263
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.7337343096733093,
+      "learning_rate": 0.00038117875062854383,
+      "loss": 0.139,
+      "step": 2264
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.49056363105773926,
+      "learning_rate": 0.00038100089973125206,
+      "loss": 0.0703,
+      "step": 2265
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.5646297931671143,
+      "learning_rate": 0.0003808230181219202,
+      "loss": 0.0875,
+      "step": 2266
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.6676580905914307,
+      "learning_rate": 0.00038064510586799323,
+      "loss": 0.1041,
+      "step": 2267
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.802716851234436,
+      "learning_rate": 0.0003804671630369278,
+      "loss": 0.1341,
+      "step": 2268
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 0.6803567409515381,
+      "learning_rate": 0.0003802891896961922,
+      "loss": 0.1411,
+      "step": 2269
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.5597631335258484,
+      "learning_rate": 0.00038011118591326626,
+      "loss": 0.1101,
+      "step": 2270
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.5911812782287598,
+      "learning_rate": 0.00037993315175564114,
+      "loss": 0.1021,
+      "step": 2271
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.7688585519790649,
+      "learning_rate": 0.00037975508729081993,
+      "loss": 0.1467,
+      "step": 2272
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.5936682820320129,
+      "learning_rate": 0.0003795769925863167,
+      "loss": 0.1155,
+      "step": 2273
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 0.7135878205299377,
+      "learning_rate": 0.0003793988677096575,
+      "loss": 0.1417,
+      "step": 2274
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.6752962470054626,
+      "learning_rate": 0.00037922071272837956,
+      "loss": 0.124,
+      "step": 2275
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.655805766582489,
+      "learning_rate": 0.0003790425277100315,
+      "loss": 0.122,
+      "step": 2276
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.6466427445411682,
+      "learning_rate": 0.0003788643127221732,
+      "loss": 0.0969,
+      "step": 2277
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.6170642375946045,
+      "learning_rate": 0.00037868606783237637,
+      "loss": 0.1308,
+      "step": 2278
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.8220329284667969,
+      "learning_rate": 0.0003785077931082237,
+      "loss": 0.134,
+      "step": 2279
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.46200326085090637,
+      "learning_rate": 0.0003783294886173092,
+      "loss": 0.088,
+      "step": 2280
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.5459378957748413,
+      "learning_rate": 0.0003781511544272382,
+      "loss": 0.0925,
+      "step": 2281
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.7910642027854919,
+      "learning_rate": 0.0003779727906056274,
+      "loss": 0.1652,
+      "step": 2282
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.5624299049377441,
+      "learning_rate": 0.00037779439722010455,
+      "loss": 0.097,
+      "step": 2283
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.5606541037559509,
+      "learning_rate": 0.00037761597433830875,
+      "loss": 0.0918,
+      "step": 2284
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 0.6552658677101135,
+      "learning_rate": 0.0003774375220278903,
+      "loss": 0.1211,
+      "step": 2285
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.6808488965034485,
+      "learning_rate": 0.00037725904035651063,
+      "loss": 0.1613,
+      "step": 2286
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.5724557042121887,
+      "learning_rate": 0.0003770805293918421,
+      "loss": 0.0955,
+      "step": 2287
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.6262295842170715,
+      "learning_rate": 0.00037690198920156855,
+      "loss": 0.1034,
+      "step": 2288
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.5716241002082825,
+      "learning_rate": 0.00037672341985338457,
+      "loss": 0.0864,
+      "step": 2289
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.7186647057533264,
+      "learning_rate": 0.0003765448214149959,
+      "loss": 0.1276,
+      "step": 2290
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.7748983502388,
+      "learning_rate": 0.0003763661939541196,
+      "loss": 0.1413,
+      "step": 2291
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.6123780608177185,
+      "learning_rate": 0.00037618753753848323,
+      "loss": 0.0984,
+      "step": 2292
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.6866722702980042,
+      "learning_rate": 0.0003760088522358257,
+      "loss": 0.1465,
+      "step": 2293
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.5656173229217529,
+      "learning_rate": 0.0003758301381138967,
+      "loss": 0.0851,
+      "step": 2294
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.6368588209152222,
+      "learning_rate": 0.0003756513952404569,
+      "loss": 0.1119,
+      "step": 2295
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.6274756193161011,
+      "learning_rate": 0.0003754726236832779,
+      "loss": 0.1511,
+      "step": 2296
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.8282131552696228,
+      "learning_rate": 0.0003752938235101421,
+      "loss": 0.1643,
+      "step": 2297
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.5847483277320862,
+      "learning_rate": 0.00037511499478884295,
+      "loss": 0.1181,
+      "step": 2298
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.5761601328849792,
+      "learning_rate": 0.0003749361375871844,
+      "loss": 0.1087,
+      "step": 2299
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.590646505355835,
+      "learning_rate": 0.0003747572519729813,
+      "loss": 0.1274,
+      "step": 2300
+    },
+    {
+      "epoch": 4.37,
+      "eval_blimp_filtered_avg": 0.5523880597014925,
+      "eval_blimp_filtered_std": 0.00581559713555244,
+      "step": 2300
+    },
+    {
+      "epoch": 4.37,
+      "eval_blimp_supplement_avg": 0.4956896551724138,
+      "eval_blimp_supplement_std": 0.023070838993464658,
+      "step": 2300
+    },
+    {
+      "epoch": 4.37,
+      "eval_vqa_filtered_avg": 0.2,
+      "eval_vqa_filtered_std": 0.040201512610368466,
+      "step": 2300
+    },
+    {
+      "epoch": 4.37,
+      "eval_winoground_filtered_avg": 0.51,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 2300
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 0.5721964836120605,
+      "learning_rate": 0.0003745783380140597,
+      "loss": 0.104,
+      "step": 2301
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.5373708605766296,
+      "learning_rate": 0.0003743993957782557,
+      "loss": 0.0981,
+      "step": 2302
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.6379767656326294,
+      "learning_rate": 0.0003742204253334165,
+      "loss": 0.1307,
+      "step": 2303
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.5864633321762085,
+      "learning_rate": 0.00037404142674740015,
+      "loss": 0.1209,
+      "step": 2304
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.7076834440231323,
+      "learning_rate": 0.00037386240008807506,
+      "loss": 0.109,
+      "step": 2305
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 0.6014961004257202,
+      "learning_rate": 0.0003736833454233203,
+      "loss": 0.1056,
+      "step": 2306
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.7625566124916077,
+      "learning_rate": 0.0003735042628210258,
+      "loss": 0.1726,
+      "step": 2307
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.6192310452461243,
+      "learning_rate": 0.00037332515234909193,
+      "loss": 0.1068,
+      "step": 2308
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.7367005348205566,
+      "learning_rate": 0.0003731460140754295,
+      "loss": 0.1448,
+      "step": 2309
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.5022045373916626,
+      "learning_rate": 0.0003729668480679602,
+      "loss": 0.0763,
+      "step": 2310
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 0.5090523362159729,
+      "learning_rate": 0.00037278765439461594,
+      "loss": 0.0993,
+      "step": 2311
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.7820708751678467,
+      "learning_rate": 0.0003726084331233393,
+      "loss": 0.1197,
+      "step": 2312
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.6847142577171326,
+      "learning_rate": 0.00037242918432208306,
+      "loss": 0.1206,
+      "step": 2313
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.7505772113800049,
+      "learning_rate": 0.00037224990805881077,
+      "loss": 0.0985,
+      "step": 2314
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.7501308917999268,
+      "learning_rate": 0.00037207060440149614,
+      "loss": 0.1515,
+      "step": 2315
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.72861647605896,
+      "learning_rate": 0.00037189127341812344,
+      "loss": 0.1183,
+      "step": 2316
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.6827106475830078,
+      "learning_rate": 0.00037171191517668725,
+      "loss": 0.1314,
+      "step": 2317
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.6931393146514893,
+      "learning_rate": 0.00037153252974519237,
+      "loss": 0.1386,
+      "step": 2318
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.5125062465667725,
+      "learning_rate": 0.00037135311719165394,
+      "loss": 0.0771,
+      "step": 2319
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.639116644859314,
+      "learning_rate": 0.0003711736775840977,
+      "loss": 0.1195,
+      "step": 2320
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.6365360617637634,
+      "learning_rate": 0.0003709942109905592,
+      "loss": 0.1128,
+      "step": 2321
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 0.6280671954154968,
+      "learning_rate": 0.0003708147174790844,
+      "loss": 0.0969,
+      "step": 2322
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.563705563545227,
+      "learning_rate": 0.0003706351971177297,
+      "loss": 0.1048,
+      "step": 2323
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.6614601016044617,
+      "learning_rate": 0.0003704556499745612,
+      "loss": 0.1237,
+      "step": 2324
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.672018826007843,
+      "learning_rate": 0.0003702760761176556,
+      "loss": 0.1309,
+      "step": 2325
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.7849199175834656,
+      "learning_rate": 0.0003700964756150994,
+      "loss": 0.1443,
+      "step": 2326
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.45391160249710083,
+      "learning_rate": 0.0003699168485349894,
+      "loss": 0.0768,
+      "step": 2327
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.6664903163909912,
+      "learning_rate": 0.00036973719494543244,
+      "loss": 0.1382,
+      "step": 2328
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.60687255859375,
+      "learning_rate": 0.0003695575149145454,
+      "loss": 0.1071,
+      "step": 2329
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.6902353167533875,
+      "learning_rate": 0.0003693778085104552,
+      "loss": 0.1463,
+      "step": 2330
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.576877772808075,
+      "learning_rate": 0.0003691980758012988,
+      "loss": 0.0802,
+      "step": 2331
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 0.5559332370758057,
+      "learning_rate": 0.0003690183168552228,
+      "loss": 0.0927,
+      "step": 2332
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.5945702195167542,
+      "learning_rate": 0.00036883853174038433,
+      "loss": 0.1056,
+      "step": 2333
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.7116044163703918,
+      "learning_rate": 0.00036865872052495004,
+      "loss": 0.1285,
+      "step": 2334
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.6097710132598877,
+      "learning_rate": 0.00036847888327709653,
+      "loss": 0.1105,
+      "step": 2335
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.6382853388786316,
+      "learning_rate": 0.00036829902006501025,
+      "loss": 0.1189,
+      "step": 2336
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.5259060263633728,
+      "learning_rate": 0.0003681191309568877,
+      "loss": 0.1087,
+      "step": 2337
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 0.6142178773880005,
+      "learning_rate": 0.0003679392160209351,
+      "loss": 0.1162,
+      "step": 2338
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.729465901851654,
+      "learning_rate": 0.0003677592753253682,
+      "loss": 0.1626,
+      "step": 2339
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.6042198538780212,
+      "learning_rate": 0.0003675793089384128,
+      "loss": 0.1249,
+      "step": 2340
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.8017492890357971,
+      "learning_rate": 0.00036739931692830453,
+      "loss": 0.1311,
+      "step": 2341
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.6672855615615845,
+      "learning_rate": 0.0003672192993632884,
+      "loss": 0.112,
+      "step": 2342
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.561383068561554,
+      "learning_rate": 0.0003670392563116194,
+      "loss": 0.0964,
+      "step": 2343
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.737882673740387,
+      "learning_rate": 0.00036685918784156194,
+      "loss": 0.1462,
+      "step": 2344
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.574994683265686,
+      "learning_rate": 0.00036667909402139035,
+      "loss": 0.0939,
+      "step": 2345
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.5887920260429382,
+      "learning_rate": 0.00036649897491938837,
+      "loss": 0.0969,
+      "step": 2346
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.5045207142829895,
+      "learning_rate": 0.0003663188306038494,
+      "loss": 0.0576,
+      "step": 2347
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.5979520678520203,
+      "learning_rate": 0.00036613866114307633,
+      "loss": 0.1278,
+      "step": 2348
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.44078439474105835,
+      "learning_rate": 0.00036595846660538166,
+      "loss": 0.0516,
+      "step": 2349
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.6361558437347412,
+      "learning_rate": 0.00036577824705908736,
+      "loss": 0.1155,
+      "step": 2350
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.5816084146499634,
+      "learning_rate": 0.0003655980025725249,
+      "loss": 0.0983,
+      "step": 2351
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.6562865376472473,
+      "learning_rate": 0.00036541773321403516,
+      "loss": 0.1065,
+      "step": 2352
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 0.6859402656555176,
+      "learning_rate": 0.0003652374390519686,
+      "loss": 0.1292,
+      "step": 2353
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.5858909487724304,
+      "learning_rate": 0.0003650571201546848,
+      "loss": 0.1094,
+      "step": 2354
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.6184085011482239,
+      "learning_rate": 0.0003648767765905531,
+      "loss": 0.0983,
+      "step": 2355
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.5549909472465515,
+      "learning_rate": 0.00036469640842795174,
+      "loss": 0.0847,
+      "step": 2356
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.6637457609176636,
+      "learning_rate": 0.0003645160157352688,
+      "loss": 0.1168,
+      "step": 2357
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.6394730806350708,
+      "learning_rate": 0.0003643355985809012,
+      "loss": 0.1465,
+      "step": 2358
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 0.698337197303772,
+      "learning_rate": 0.00036415515703325534,
+      "loss": 0.1169,
+      "step": 2359
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.6428084969520569,
+      "learning_rate": 0.000363974691160747,
+      "loss": 0.0951,
+      "step": 2360
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.6346538662910461,
+      "learning_rate": 0.0003637942010318008,
+      "loss": 0.1115,
+      "step": 2361
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.6214971542358398,
+      "learning_rate": 0.0003636136867148511,
+      "loss": 0.1078,
+      "step": 2362
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.5514646172523499,
+      "learning_rate": 0.00036343314827834086,
+      "loss": 0.1053,
+      "step": 2363
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 0.5496067404747009,
+      "learning_rate": 0.0003632525857907226,
+      "loss": 0.0947,
+      "step": 2364
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.6322467923164368,
+      "learning_rate": 0.0003630719993204577,
+      "loss": 0.1235,
+      "step": 2365
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.6454159021377563,
+      "learning_rate": 0.00036289138893601686,
+      "loss": 0.1113,
+      "step": 2366
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.5399425029754639,
+      "learning_rate": 0.00036271075470587963,
+      "loss": 0.0881,
+      "step": 2367
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.6995037794113159,
+      "learning_rate": 0.0003625300966985347,
+      "loss": 0.1211,
+      "step": 2368
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.5016408562660217,
+      "learning_rate": 0.00036234941498248,
+      "loss": 0.095,
+      "step": 2369
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.6358677744865417,
+      "learning_rate": 0.00036216870962622194,
+      "loss": 0.1413,
+      "step": 2370
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.6576187610626221,
+      "learning_rate": 0.00036198798069827636,
+      "loss": 0.1314,
+      "step": 2371
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.6394796371459961,
+      "learning_rate": 0.00036180722826716777,
+      "loss": 0.1336,
+      "step": 2372
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.6270532011985779,
+      "learning_rate": 0.0003616264524014297,
+      "loss": 0.1272,
+      "step": 2373
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 0.6427406668663025,
+      "learning_rate": 0.00036144565316960456,
+      "loss": 0.118,
+      "step": 2374
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.6995803117752075,
+      "learning_rate": 0.0003612648306402436,
+      "loss": 0.1474,
+      "step": 2375
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.6441980600357056,
+      "learning_rate": 0.0003610839848819068,
+      "loss": 0.1334,
+      "step": 2376
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.589851975440979,
+      "learning_rate": 0.00036090311596316326,
+      "loss": 0.1071,
+      "step": 2377
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.43165698647499084,
+      "learning_rate": 0.0003607222239525905,
+      "loss": 0.0744,
+      "step": 2378
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.5401017665863037,
+      "learning_rate": 0.000360541308918775,
+      "loss": 0.0862,
+      "step": 2379
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 0.43925753235816956,
+      "learning_rate": 0.00036036037093031196,
+      "loss": 0.0815,
+      "step": 2380
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.6398507952690125,
+      "learning_rate": 0.0003601794100558052,
+      "loss": 0.1252,
+      "step": 2381
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.6328310966491699,
+      "learning_rate": 0.0003599984263638672,
+      "loss": 0.1194,
+      "step": 2382
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.643570601940155,
+      "learning_rate": 0.00035981741992311935,
+      "loss": 0.121,
+      "step": 2383
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.5710518956184387,
+      "learning_rate": 0.0003596363908021913,
+      "loss": 0.0932,
+      "step": 2384
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 0.5184112191200256,
+      "learning_rate": 0.0003594553390697215,
+      "loss": 0.0794,
+      "step": 2385
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.621701180934906,
+      "learning_rate": 0.000359274264794357,
+      "loss": 0.0948,
+      "step": 2386
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.4309825599193573,
+      "learning_rate": 0.00035909316804475324,
+      "loss": 0.0588,
+      "step": 2387
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.6497613191604614,
+      "learning_rate": 0.00035891204888957427,
+      "loss": 0.142,
+      "step": 2388
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.6979169845581055,
+      "learning_rate": 0.0003587309073974928,
+      "loss": 0.104,
+      "step": 2389
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 0.6334772109985352,
+      "learning_rate": 0.00035854974363718975,
+      "loss": 0.1218,
+      "step": 2390
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.5316715836524963,
+      "learning_rate": 0.00035836855767735446,
+      "loss": 0.0888,
+      "step": 2391
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.6660722494125366,
+      "learning_rate": 0.0003581873495866851,
+      "loss": 0.1673,
+      "step": 2392
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.5991950035095215,
+      "learning_rate": 0.0003580061194338878,
+      "loss": 0.0831,
+      "step": 2393
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.6906603574752808,
+      "learning_rate": 0.0003578248672876772,
+      "loss": 0.1573,
+      "step": 2394
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.6573761105537415,
+      "learning_rate": 0.00035764359321677626,
+      "loss": 0.1419,
+      "step": 2395
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.627439022064209,
+      "learning_rate": 0.0003574622972899163,
+      "loss": 0.1133,
+      "step": 2396
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.4194423258304596,
+      "learning_rate": 0.0003572809795758368,
+      "loss": 0.0676,
+      "step": 2397
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.5952950716018677,
+      "learning_rate": 0.00035709964014328576,
+      "loss": 0.1183,
+      "step": 2398
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.573484480381012,
+      "learning_rate": 0.00035691827906101915,
+      "loss": 0.0923,
+      "step": 2399
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.5142375826835632,
+      "learning_rate": 0.0003567368963978014,
+      "loss": 0.1084,
+      "step": 2400
+    },
+    {
+      "epoch": 4.56,
+      "eval_blimp_filtered_avg": 0.5404477611940298,
+      "eval_blimp_filtered_std": 0.005806536154939915,
+      "step": 2400
+    },
+    {
+      "epoch": 4.56,
+      "eval_blimp_supplement_avg": 0.5150862068965517,
+      "eval_blimp_supplement_std": 0.02322708350522436,
+      "step": 2400
+    },
+    {
+      "epoch": 4.56,
+      "eval_vqa_filtered_avg": 0.12,
+      "eval_vqa_filtered_std": 0.03265986323710906,
+      "step": 2400
+    },
+    {
+      "epoch": 4.56,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 2400
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 0.631787121295929,
+      "learning_rate": 0.00035655549222240475,
+      "loss": 0.0914,
+      "step": 2401
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.5728902816772461,
+      "learning_rate": 0.00035637406660360996,
+      "loss": 0.1299,
+      "step": 2402
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.6180467009544373,
+      "learning_rate": 0.0003561926196102058,
+      "loss": 0.1184,
+      "step": 2403
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.6311930418014526,
+      "learning_rate": 0.000356011151310989,
+      "loss": 0.1133,
+      "step": 2404
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.5987362265586853,
+      "learning_rate": 0.00035582966177476465,
+      "loss": 0.1191,
+      "step": 2405
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 0.6291629076004028,
+      "learning_rate": 0.00035564815107034565,
+      "loss": 0.1278,
+      "step": 2406
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.5104873776435852,
+      "learning_rate": 0.00035546661926655305,
+      "loss": 0.0801,
+      "step": 2407
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.5455119609832764,
+      "learning_rate": 0.0003552850664322159,
+      "loss": 0.119,
+      "step": 2408
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.6001309156417847,
+      "learning_rate": 0.0003551034926361711,
+      "loss": 0.1097,
+      "step": 2409
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.7429015636444092,
+      "learning_rate": 0.0003549218979472636,
+      "loss": 0.119,
+      "step": 2410
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 0.6921620965003967,
+      "learning_rate": 0.00035474028243434624,
+      "loss": 0.1356,
+      "step": 2411
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.6262943148612976,
+      "learning_rate": 0.00035455864616627987,
+      "loss": 0.0986,
+      "step": 2412
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.6199610233306885,
+      "learning_rate": 0.00035437698921193307,
+      "loss": 0.1074,
+      "step": 2413
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.5386694073677063,
+      "learning_rate": 0.00035419531164018216,
+      "loss": 0.0889,
+      "step": 2414
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.6236569285392761,
+      "learning_rate": 0.00035401361351991166,
+      "loss": 0.0982,
+      "step": 2415
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.6449233293533325,
+      "learning_rate": 0.00035383189492001356,
+      "loss": 0.1463,
+      "step": 2416
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.5636408925056458,
+      "learning_rate": 0.00035365015590938756,
+      "loss": 0.1004,
+      "step": 2417
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.43737295269966125,
+      "learning_rate": 0.0003534683965569415,
+      "loss": 0.0735,
+      "step": 2418
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.6006467938423157,
+      "learning_rate": 0.0003532866169315904,
+      "loss": 0.1346,
+      "step": 2419
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.5108967423439026,
+      "learning_rate": 0.0003531048171022574,
+      "loss": 0.0914,
+      "step": 2420
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.5442214608192444,
+      "learning_rate": 0.00035292299713787306,
+      "loss": 0.1035,
+      "step": 2421
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.5852410793304443,
+      "learning_rate": 0.0003527411571073757,
+      "loss": 0.1195,
+      "step": 2422
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.7534271478652954,
+      "learning_rate": 0.00035255929707971126,
+      "loss": 0.1581,
+      "step": 2423
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.5752830505371094,
+      "learning_rate": 0.00035237741712383305,
+      "loss": 0.1108,
+      "step": 2424
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.5280358791351318,
+      "learning_rate": 0.0003521955173087022,
+      "loss": 0.0954,
+      "step": 2425
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.6789833307266235,
+      "learning_rate": 0.00035201359770328723,
+      "loss": 0.1487,
+      "step": 2426
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 0.5741812586784363,
+      "learning_rate": 0.00035183165837656414,
+      "loss": 0.0889,
+      "step": 2427
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.7208979725837708,
+      "learning_rate": 0.0003516496993975165,
+      "loss": 0.144,
+      "step": 2428
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.6706939339637756,
+      "learning_rate": 0.0003514677208351354,
+      "loss": 0.1472,
+      "step": 2429
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.42230281233787537,
+      "learning_rate": 0.00035128572275841904,
+      "loss": 0.0805,
+      "step": 2430
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.4893317222595215,
+      "learning_rate": 0.0003511037052363734,
+      "loss": 0.0888,
+      "step": 2431
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 0.647075891494751,
+      "learning_rate": 0.0003509216683380115,
+      "loss": 0.1209,
+      "step": 2432
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.5984809398651123,
+      "learning_rate": 0.00035073961213235395,
+      "loss": 0.109,
+      "step": 2433
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.6465935111045837,
+      "learning_rate": 0.0003505575366884286,
+      "loss": 0.1229,
+      "step": 2434
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.5404577255249023,
+      "learning_rate": 0.00035037544207527064,
+      "loss": 0.1341,
+      "step": 2435
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.5898814797401428,
+      "learning_rate": 0.0003501933283619224,
+      "loss": 0.1215,
+      "step": 2436
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.5689233541488647,
+      "learning_rate": 0.00035001119561743345,
+      "loss": 0.0751,
+      "step": 2437
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 0.5174353122711182,
+      "learning_rate": 0.0003498290439108608,
+      "loss": 0.0981,
+      "step": 2438
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.6658139228820801,
+      "learning_rate": 0.0003496468733112685,
+      "loss": 0.1633,
+      "step": 2439
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.5880412459373474,
+      "learning_rate": 0.00034946468388772766,
+      "loss": 0.125,
+      "step": 2440
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.6458158493041992,
+      "learning_rate": 0.00034928247570931674,
+      "loss": 0.1119,
+      "step": 2441
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.5550180673599243,
+      "learning_rate": 0.00034910024884512113,
+      "loss": 0.1072,
+      "step": 2442
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 0.6061705946922302,
+      "learning_rate": 0.0003489180033642333,
+      "loss": 0.1229,
+      "step": 2443
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.648802638053894,
+      "learning_rate": 0.0003487357393357532,
+      "loss": 0.1275,
+      "step": 2444
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.49662721157073975,
+      "learning_rate": 0.00034855345682878713,
+      "loss": 0.0785,
+      "step": 2445
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.5387270450592041,
+      "learning_rate": 0.0003483711559124488,
+      "loss": 0.0878,
+      "step": 2446
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.5279498100280762,
+      "learning_rate": 0.00034818883665585883,
+      "loss": 0.1251,
+      "step": 2447
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 0.5602584481239319,
+      "learning_rate": 0.000348006499128145,
+      "loss": 0.1227,
+      "step": 2448
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.6310223937034607,
+      "learning_rate": 0.00034782414339844166,
+      "loss": 0.1248,
+      "step": 2449
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.586723268032074,
+      "learning_rate": 0.00034764176953589013,
+      "loss": 0.0987,
+      "step": 2450
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.5968186259269714,
+      "learning_rate": 0.00034745937760963886,
+      "loss": 0.1201,
+      "step": 2451
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.47636082768440247,
+      "learning_rate": 0.000347276967688843,
+      "loss": 0.0656,
+      "step": 2452
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 0.5886270999908447,
+      "learning_rate": 0.0003470945398426644,
+      "loss": 0.1323,
+      "step": 2453
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.5182158350944519,
+      "learning_rate": 0.0003469120941402718,
+      "loss": 0.087,
+      "step": 2454
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.4822755753993988,
+      "learning_rate": 0.00034672963065084084,
+      "loss": 0.0843,
+      "step": 2455
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.6982547640800476,
+      "learning_rate": 0.0003465471494435536,
+      "loss": 0.1235,
+      "step": 2456
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.4048895835876465,
+      "learning_rate": 0.00034636465058759936,
+      "loss": 0.079,
+      "step": 2457
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.5396956205368042,
+      "learning_rate": 0.00034618213415217353,
+      "loss": 0.0717,
+      "step": 2458
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 0.5466949939727783,
+      "learning_rate": 0.0003459996002064786,
+      "loss": 0.1154,
+      "step": 2459
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.45758068561553955,
+      "learning_rate": 0.00034581704881972344,
+      "loss": 0.0583,
+      "step": 2460
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.6710917949676514,
+      "learning_rate": 0.00034563448006112377,
+      "loss": 0.1472,
+      "step": 2461
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.5833229422569275,
+      "learning_rate": 0.00034545189399990165,
+      "loss": 0.1302,
+      "step": 2462
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.6149641871452332,
+      "learning_rate": 0.00034526929070528595,
+      "loss": 0.0978,
+      "step": 2463
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 0.6006848216056824,
+      "learning_rate": 0.00034508667024651187,
+      "loss": 0.0899,
+      "step": 2464
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 0.5661689639091492,
+      "learning_rate": 0.0003449040326928212,
+      "loss": 0.0905,
+      "step": 2465
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 0.627905011177063,
+      "learning_rate": 0.0003447213781134621,
+      "loss": 0.12,
+      "step": 2466
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 0.5093738436698914,
+      "learning_rate": 0.00034453870657768956,
+      "loss": 0.0639,
+      "step": 2467
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 0.8187510371208191,
+      "learning_rate": 0.00034435601815476456,
+      "loss": 0.1599,
+      "step": 2468
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 0.44370996952056885,
+      "learning_rate": 0.00034417331291395454,
+      "loss": 0.0633,
+      "step": 2469
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 0.5135902762413025,
+      "learning_rate": 0.0003439905909245337,
+      "loss": 0.0845,
+      "step": 2470
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 0.7205265760421753,
+      "learning_rate": 0.0003438078522557821,
+      "loss": 0.1349,
+      "step": 2471
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 0.6324465870857239,
+      "learning_rate": 0.0003436250969769865,
+      "loss": 0.1233,
+      "step": 2472
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 0.455877423286438,
+      "learning_rate": 0.0003434423251574396,
+      "loss": 0.0539,
+      "step": 2473
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 0.7055237293243408,
+      "learning_rate": 0.0003432595368664408,
+      "loss": 0.1303,
+      "step": 2474
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 0.49157801270484924,
+      "learning_rate": 0.0003430767321732954,
+      "loss": 0.0782,
+      "step": 2475
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 0.5151515007019043,
+      "learning_rate": 0.0003428939111473151,
+      "loss": 0.0807,
+      "step": 2476
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 0.5193693041801453,
+      "learning_rate": 0.0003427110738578176,
+      "loss": 0.1156,
+      "step": 2477
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 0.5535184144973755,
+      "learning_rate": 0.0003425282203741271,
+      "loss": 0.1026,
+      "step": 2478
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 0.4764206111431122,
+      "learning_rate": 0.00034234535076557346,
+      "loss": 0.0827,
+      "step": 2479
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 0.6120935678482056,
+      "learning_rate": 0.0003421624651014932,
+      "loss": 0.1326,
+      "step": 2480
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.5984523892402649,
+      "learning_rate": 0.00034197956345122856,
+      "loss": 0.123,
+      "step": 2481
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.4574005901813507,
+      "learning_rate": 0.00034179664588412787,
+      "loss": 0.0855,
+      "step": 2482
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.613044261932373,
+      "learning_rate": 0.00034161371246954565,
+      "loss": 0.1307,
+      "step": 2483
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.6082948446273804,
+      "learning_rate": 0.0003414307632768424,
+      "loss": 0.106,
+      "step": 2484
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 0.5069074630737305,
+      "learning_rate": 0.0003412477983753844,
+      "loss": 0.0973,
+      "step": 2485
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.5656412243843079,
+      "learning_rate": 0.00034106481783454415,
+      "loss": 0.098,
+      "step": 2486
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.6260591745376587,
+      "learning_rate": 0.0003408818217236999,
+      "loss": 0.1213,
+      "step": 2487
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.5079365968704224,
+      "learning_rate": 0.0003406988101122359,
+      "loss": 0.1025,
+      "step": 2488
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.6834782958030701,
+      "learning_rate": 0.0003405157830695421,
+      "loss": 0.1283,
+      "step": 2489
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 0.5567680597305298,
+      "learning_rate": 0.00034033274066501467,
+      "loss": 0.0886,
+      "step": 2490
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 0.5290331840515137,
+      "learning_rate": 0.0003401496829680553,
+      "loss": 0.0957,
+      "step": 2491
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 0.38408592343330383,
+      "learning_rate": 0.0003399666100480714,
+      "loss": 0.0596,
+      "step": 2492
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 0.5968230962753296,
+      "learning_rate": 0.0003397835219744766,
+      "loss": 0.1257,
+      "step": 2493
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 0.6120256185531616,
+      "learning_rate": 0.00033960041881668973,
+      "loss": 0.1509,
+      "step": 2494
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 0.5389651656150818,
+      "learning_rate": 0.00033941730064413573,
+      "loss": 0.0912,
+      "step": 2495
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.6902221441268921,
+      "learning_rate": 0.00033923416752624494,
+      "loss": 0.1103,
+      "step": 2496
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.7500288486480713,
+      "learning_rate": 0.00033905101953245377,
+      "loss": 0.1265,
+      "step": 2497
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.6434916257858276,
+      "learning_rate": 0.00033886785673220386,
+      "loss": 0.1306,
+      "step": 2498
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.5839682817459106,
+      "learning_rate": 0.0003386846791949427,
+      "loss": 0.1197,
+      "step": 2499
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.509861171245575,
+      "learning_rate": 0.00033850148699012314,
+      "loss": 0.0886,
+      "step": 2500
+    },
+    {
+      "epoch": 4.75,
+      "eval_blimp_filtered_avg": 0.54,
+      "eval_blimp_filtered_std": 0.005831072680562093,
+      "step": 2500
+    },
+    {
+      "epoch": 4.75,
+      "eval_blimp_supplement_avg": 0.5129310344827587,
+      "eval_blimp_supplement_std": 0.023200536091122624,
+      "step": 2500
+    },
+    {
+      "epoch": 4.75,
+      "eval_vqa_filtered_avg": 0.18,
+      "eval_vqa_filtered_std": 0.038612291966536934,
+      "step": 2500
+    },
+    {
+      "epoch": 4.75,
+      "eval_winoground_filtered_avg": 0.45,
+      "eval_winoground_filtered_std": 0.05,
+      "step": 2500
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.40043821930885315,
+      "learning_rate": 0.00033831828018720395,
+      "loss": 0.0619,
+      "step": 2501
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 0.45643943548202515,
+      "learning_rate": 0.000338135058855649,
+      "loss": 0.0709,
+      "step": 2502
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 0.6159443855285645,
+      "learning_rate": 0.0003379518230649282,
+      "loss": 0.1097,
+      "step": 2503
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 0.5198758840560913,
+      "learning_rate": 0.0003377685728845164,
+      "loss": 0.0726,
+      "step": 2504
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 0.44191399216651917,
+      "learning_rate": 0.0003375853083838942,
+      "loss": 0.0904,
+      "step": 2505
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 0.5884183645248413,
+      "learning_rate": 0.00033740202963254766,
+      "loss": 0.1066,
+      "step": 2506
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 0.5733084678649902,
+      "learning_rate": 0.0003372187366999681,
+      "loss": 0.0899,
+      "step": 2507
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 0.6540478467941284,
+      "learning_rate": 0.0003370354296556523,
+      "loss": 0.1157,
+      "step": 2508
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 0.6422880291938782,
+      "learning_rate": 0.0003368521085691023,
+      "loss": 0.1153,
+      "step": 2509
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 0.7753595113754272,
+      "learning_rate": 0.0003366687735098256,
+      "loss": 0.1346,
+      "step": 2510
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 0.6775477528572083,
+      "learning_rate": 0.0003364854245473349,
+      "loss": 0.1035,
+      "step": 2511
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 0.5548525452613831,
+      "learning_rate": 0.00033630206175114827,
+      "loss": 0.1176,
+      "step": 2512
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 0.6273616552352905,
+      "learning_rate": 0.00033611868519078887,
+      "loss": 0.1349,
+      "step": 2513
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 0.4649050235748291,
+      "learning_rate": 0.00033593529493578515,
+      "loss": 0.0917,
+      "step": 2514
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 0.6342312097549438,
+      "learning_rate": 0.00033575189105567084,
+      "loss": 0.0951,
+      "step": 2515
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 0.5309105515480042,
+      "learning_rate": 0.0003355684736199847,
+      "loss": 0.0819,
+      "step": 2516
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 0.8025500774383545,
+      "learning_rate": 0.00033538504269827086,
+      "loss": 0.1639,
+      "step": 2517
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 0.5558556914329529,
+      "learning_rate": 0.0003352015983600781,
+      "loss": 0.085,
+      "step": 2518
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 0.620010256767273,
+      "learning_rate": 0.0003350181406749609,
+      "loss": 0.1123,
+      "step": 2519
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 0.49416583776474,
+      "learning_rate": 0.0003348346697124783,
+      "loss": 0.0846,
+      "step": 2520
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 0.6176611185073853,
+      "learning_rate": 0.0003346511855421946,
+      "loss": 0.1001,
+      "step": 2521
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 0.5738742351531982,
+      "learning_rate": 0.00033446768823367905,
+      "loss": 0.1014,
+      "step": 2522
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.5969063639640808,
+      "learning_rate": 0.00033428417785650596,
+      "loss": 0.0934,
+      "step": 2523
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.549521267414093,
+      "learning_rate": 0.0003341006544802545,
+      "loss": 0.0998,
+      "step": 2524
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.6934860348701477,
+      "learning_rate": 0.0003339171181745087,
+      "loss": 0.1361,
+      "step": 2525
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.47549256682395935,
+      "learning_rate": 0.0003337335690088576,
+      "loss": 0.0733,
+      "step": 2526
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.593654453754425,
+      "learning_rate": 0.0003335500070528954,
+      "loss": 0.0924,
+      "step": 2527
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.48575350642204285,
+      "learning_rate": 0.00033336643237622047,
+      "loss": 0.0665,
+      "step": 2528
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.623408317565918,
+      "learning_rate": 0.00033318284504843667,
+      "loss": 0.1095,
+      "step": 2529
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.525310754776001,
+      "learning_rate": 0.0003329992451391523,
+      "loss": 0.077,
+      "step": 2530
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.5856096744537354,
+      "learning_rate": 0.00033281563271798047,
+      "loss": 0.1148,
+      "step": 2531
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 0.5986639857292175,
+      "learning_rate": 0.00033263200785453915,
+      "loss": 0.0827,
+      "step": 2532
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 0.6396752595901489,
+      "learning_rate": 0.00033244837061845086,
+      "loss": 0.1106,
+      "step": 2533
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 0.666419506072998,
+      "learning_rate": 0.00033226472107934286,
+      "loss": 0.1341,
+      "step": 2534
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 0.675774872303009,
+      "learning_rate": 0.00033208105930684727,
+      "loss": 0.1359,
+      "step": 2535
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 0.5463175773620605,
+      "learning_rate": 0.0003318973853706006,
+      "loss": 0.0888,
+      "step": 2536
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 0.47457072138786316,
+      "learning_rate": 0.00033171369934024416,
+      "loss": 0.0601,
+      "step": 2537
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.6740870475769043,
+      "learning_rate": 0.00033153000128542356,
+      "loss": 0.1154,
+      "step": 2538
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.5972588658332825,
+      "learning_rate": 0.0003313462912757894,
+      "loss": 0.1212,
+      "step": 2539
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.6392324566841125,
+      "learning_rate": 0.0003311625693809964,
+      "loss": 0.1122,
+      "step": 2540
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.6360642313957214,
+      "learning_rate": 0.000330978835670704,
+      "loss": 0.1239,
+      "step": 2541
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.5680141448974609,
+      "learning_rate": 0.0003307950902145762,
+      "loss": 0.1307,
+      "step": 2542
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 0.6143139004707336,
+      "learning_rate": 0.00033061133308228124,
+      "loss": 0.1242,
+      "step": 2543
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.5031746029853821,
+      "learning_rate": 0.0003304275643434918,
+      "loss": 0.0734,
+      "step": 2544
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.708976149559021,
+      "learning_rate": 0.0003302437840678852,
+      "loss": 0.1137,
+      "step": 2545
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.4944307506084442,
+      "learning_rate": 0.0003300599923251429,
+      "loss": 0.0879,
+      "step": 2546
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.46304020285606384,
+      "learning_rate": 0.00032987618918495074,
+      "loss": 0.0742,
+      "step": 2547
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 0.4216800630092621,
+      "learning_rate": 0.000329692374716999,
+      "loss": 0.0695,
+      "step": 2548
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 0.6729946136474609,
+      "learning_rate": 0.0003295085489909822,
+      "loss": 0.1038,
+      "step": 2549
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 0.6376661658287048,
+      "learning_rate": 0.000329324712076599,
+      "loss": 0.1257,
+      "step": 2550
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 0.5135642290115356,
+      "learning_rate": 0.0003291408640435524,
+      "loss": 0.0958,
+      "step": 2551
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 0.5697847008705139,
+      "learning_rate": 0.0003289570049615499,
+      "loss": 0.1095,
+      "step": 2552
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 0.6046176552772522,
+      "learning_rate": 0.00032877313490030255,
+      "loss": 0.1123,
+      "step": 2553
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 0.5319065451622009,
+      "learning_rate": 0.00032858925392952623,
+      "loss": 0.094,
+      "step": 2554
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 0.6588727235794067,
+      "learning_rate": 0.00032840536211894045,
+      "loss": 0.1002,
+      "step": 2555
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 0.5910625457763672,
+      "learning_rate": 0.00032822145953826916,
+      "loss": 0.1243,
+      "step": 2556
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 0.5319292545318604,
+      "learning_rate": 0.0003280375462572402,
+      "loss": 0.1077,
+      "step": 2557
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 0.6285613179206848,
+      "learning_rate": 0.0003278536223455855,
+      "loss": 0.1531,
+      "step": 2558
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 0.5655633807182312,
+      "learning_rate": 0.00032766968787304124,
+      "loss": 0.0788,
+      "step": 2559
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 0.5915897488594055,
+      "learning_rate": 0.00032748574290934724,
+      "loss": 0.1056,
+      "step": 2560
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 0.6051650047302246,
+      "learning_rate": 0.00032730178752424753,
+      "loss": 0.1222,
+      "step": 2561
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 0.6582232117652893,
+      "learning_rate": 0.00032711782178749003,
+      "loss": 0.1147,
+      "step": 2562
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 0.49483126401901245,
+      "learning_rate": 0.0003269338457688267,
+      "loss": 0.0779,
+      "step": 2563
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 0.5742583870887756,
+      "learning_rate": 0.00032674985953801307,
+      "loss": 0.1151,
+      "step": 2564
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.5805327296257019,
+      "learning_rate": 0.000326565863164809,
+      "loss": 0.0958,
+      "step": 2565
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.4902883470058441,
+      "learning_rate": 0.0003263818567189779,
+      "loss": 0.0712,
+      "step": 2566
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.4799145758152008,
+      "learning_rate": 0.000326197840270287,
+      "loss": 0.0558,
+      "step": 2567
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.5933427810668945,
+      "learning_rate": 0.0003260138138885074,
+      "loss": 0.1119,
+      "step": 2568
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.5421668887138367,
+      "learning_rate": 0.00032582977764341405,
+      "loss": 0.1063,
+      "step": 2569
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.5681478977203369,
+      "learning_rate": 0.0003256457316047854,
+      "loss": 0.1018,
+      "step": 2570
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.5525860786437988,
+      "learning_rate": 0.00032546167584240367,
+      "loss": 0.0987,
+      "step": 2571
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.540233850479126,
+      "learning_rate": 0.00032527761042605514,
+      "loss": 0.1109,
+      "step": 2572
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.5202035903930664,
+      "learning_rate": 0.00032509353542552926,
+      "loss": 0.0939,
+      "step": 2573
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 0.4349798858165741,
+      "learning_rate": 0.00032490945091061935,
+      "loss": 0.066,
+      "step": 2574
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.580929160118103,
+      "learning_rate": 0.00032472535695112235,
+      "loss": 0.107,
+      "step": 2575
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.4513215124607086,
+      "learning_rate": 0.00032454125361683865,
+      "loss": 0.075,
+      "step": 2576
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.5819218158721924,
+      "learning_rate": 0.00032435714097757244,
+      "loss": 0.0947,
+      "step": 2577
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.4282824993133545,
+      "learning_rate": 0.000324173019103131,
+      "loss": 0.0788,
+      "step": 2578
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.5710247755050659,
+      "learning_rate": 0.00032398888806332567,
+      "loss": 0.0964,
+      "step": 2579
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.46373143792152405,
+      "learning_rate": 0.0003238047479279707,
+      "loss": 0.0732,
+      "step": 2580
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.5840041041374207,
+      "learning_rate": 0.0003236205987668843,
+      "loss": 0.1005,
+      "step": 2581
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.6133108139038086,
+      "learning_rate": 0.00032343644064988786,
+      "loss": 0.1196,
+      "step": 2582
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.5582835674285889,
+      "learning_rate": 0.0003232522736468061,
+      "loss": 0.1171,
+      "step": 2583
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.4288780391216278,
+      "learning_rate": 0.0003230680978274671,
+      "loss": 0.0651,
+      "step": 2584
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.4993967115879059,
+      "learning_rate": 0.00032288391326170253,
+      "loss": 0.0805,
+      "step": 2585
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.41591164469718933,
+      "learning_rate": 0.0003226997200193472,
+      "loss": 0.0653,
+      "step": 2586
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.6035439968109131,
+      "learning_rate": 0.0003225155181702391,
+      "loss": 0.1113,
+      "step": 2587
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.4216573238372803,
+      "learning_rate": 0.00032233130778421974,
+      "loss": 0.0539,
+      "step": 2588
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.47905951738357544,
+      "learning_rate": 0.0003221470889311337,
+      "loss": 0.0808,
+      "step": 2589
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.5926640033721924,
+      "learning_rate": 0.0003219628616808288,
+      "loss": 0.0983,
+      "step": 2590
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 0.834683895111084,
+      "learning_rate": 0.0003217786261031561,
+      "loss": 0.1434,
+      "step": 2591
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 0.7158082723617554,
+      "learning_rate": 0.00032159438226796964,
+      "loss": 0.1578,
+      "step": 2592
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 0.5338995456695557,
+      "learning_rate": 0.0003214101302451268,
+      "loss": 0.0701,
+      "step": 2593
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 0.6891649961471558,
+      "learning_rate": 0.00032122587010448793,
+      "loss": 0.1076,
+      "step": 2594
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 0.6767675876617432,
+      "learning_rate": 0.0003210416019159166,
+      "loss": 0.141,
+      "step": 2595
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.8885285258293152,
+      "learning_rate": 0.0003208573257492794,
+      "loss": 0.2072,
+      "step": 2596
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.5555295944213867,
+      "learning_rate": 0.00032067304167444563,
+      "loss": 0.0958,
+      "step": 2597
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.5715036392211914,
+      "learning_rate": 0.0003204887497612881,
+      "loss": 0.0974,
+      "step": 2598
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.5321238040924072,
+      "learning_rate": 0.0003203044500796822,
+      "loss": 0.1014,
+      "step": 2599
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.3409011662006378,
+      "learning_rate": 0.00032012014269950647,
+      "loss": 0.0421,
+      "step": 2600
+    },
+    {
+      "epoch": 4.94,
+      "eval_blimp_filtered_avg": 0.5294029850746269,
+      "eval_blimp_filtered_std": 0.0058471999947428115,
+      "step": 2600
+    },
+    {
+      "epoch": 4.94,
+      "eval_blimp_supplement_avg": 0.5538793103448276,
+      "eval_blimp_supplement_std": 0.02308355545522312,
+      "step": 2600
+    },
+    {
+      "epoch": 4.94,
+      "eval_vqa_filtered_avg": 0.17,
+      "eval_vqa_filtered_std": 0.03775251680686371,
+      "step": 2600
+    },
+    {
+      "epoch": 4.94,
+      "eval_winoground_filtered_avg": 0.46,
+      "eval_winoground_filtered_std": 0.05009082659620333,
+      "step": 2600
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 0.6032962203025818,
+      "learning_rate": 0.0003199358276906423,
+      "loss": 0.1288,
+      "step": 2601
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.5322068333625793,
+      "learning_rate": 0.000319751505122974,
+      "loss": 0.0898,
+      "step": 2602
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.57469242811203,
+      "learning_rate": 0.0003195671750663885,
+      "loss": 0.0835,
+      "step": 2603
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.5454330444335938,
+      "learning_rate": 0.00031938283759077603,
+      "loss": 0.1081,
+      "step": 2604
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.5250727534294128,
+      "learning_rate": 0.0003191984927660294,
+      "loss": 0.1048,
+      "step": 2605
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.48337051272392273,
+      "learning_rate": 0.00031901414066204414,
+      "loss": 0.1075,
+      "step": 2606
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.4703986346721649,
+      "learning_rate": 0.0003188297813487185,
+      "loss": 0.0764,
+      "step": 2607
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.48361656069755554,
+      "learning_rate": 0.0003186454148959537,
+      "loss": 0.0913,
+      "step": 2608
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.5230949521064758,
+      "learning_rate": 0.00031846104137365344,
+      "loss": 0.0882,
+      "step": 2609
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.47167307138442993,
+      "learning_rate": 0.0003182766608517242,
+      "loss": 0.0709,
+      "step": 2610
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 0.5961464643478394,
+      "learning_rate": 0.00031809227340007517,
+      "loss": 0.1071,
+      "step": 2611
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 0.5541611313819885,
+      "learning_rate": 0.00031790787908861807,
+      "loss": 0.0848,
+      "step": 2612
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 0.65880286693573,
+      "learning_rate": 0.0003177234779872673,
+      "loss": 0.1071,
+      "step": 2613
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 0.5279577374458313,
+      "learning_rate": 0.00031753907016593966,
+      "loss": 0.1076,
+      "step": 2614
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 0.507997989654541,
+      "learning_rate": 0.0003173546556945548,
+      "loss": 0.0863,
+      "step": 2615
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 0.6160945296287537,
+      "learning_rate": 0.00031717023464303447,
+      "loss": 0.1211,
+      "step": 2616
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.6595787405967712,
+      "learning_rate": 0.0003169858070813034,
+      "loss": 0.1207,
+      "step": 2617
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.425027996301651,
+      "learning_rate": 0.00031680137307928846,
+      "loss": 0.0607,
+      "step": 2618
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.6258980631828308,
+      "learning_rate": 0.00031661693270691906,
+      "loss": 0.1204,
+      "step": 2619
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.6928203701972961,
+      "learning_rate": 0.0003164324860341269,
+      "loss": 0.1397,
+      "step": 2620
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.6261894106864929,
+      "learning_rate": 0.0003162480331308464,
+      "loss": 0.1169,
+      "step": 2621
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 0.4223698377609253,
+      "learning_rate": 0.0003160635740670141,
+      "loss": 0.0691,
+      "step": 2622
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 0.504023551940918,
+      "learning_rate": 0.0003158791089125687,
+      "loss": 0.1047,
+      "step": 2623
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 0.49146324396133423,
+      "learning_rate": 0.0003156946377374516,
+      "loss": 0.0683,
+      "step": 2624
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 0.5999441742897034,
+      "learning_rate": 0.00031551016061160623,
+      "loss": 0.1151,
+      "step": 2625
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 0.4722380042076111,
+      "learning_rate": 0.00031532567760497824,
+      "loss": 0.0611,
+      "step": 2626
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 0.6382120251655579,
+      "learning_rate": 0.00031514118878751585,
+      "loss": 0.0989,
+      "step": 2627
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.5013511776924133,
+      "learning_rate": 0.00031495669422916913,
+      "loss": 0.0828,
+      "step": 2628
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.42499831318855286,
+      "learning_rate": 0.0003147721939998903,
+      "loss": 0.0761,
+      "step": 2629
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.45410820841789246,
+      "learning_rate": 0.0003145876881696341,
+      "loss": 0.0657,
+      "step": 2630
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.5169359445571899,
+      "learning_rate": 0.000314403176808357,
+      "loss": 0.039,
+      "step": 2631
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.41324782371520996,
+      "learning_rate": 0.0003142186599860179,
+      "loss": 0.0542,
+      "step": 2632
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 0.445215106010437,
+      "learning_rate": 0.0003140341377725774,
+      "loss": 0.0507,
+      "step": 2633
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 0.4344821870326996,
+      "learning_rate": 0.00031384961023799853,
+      "loss": 0.0499,
+      "step": 2634
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 0.508225679397583,
+      "learning_rate": 0.00031366507745224603,
+      "loss": 0.058,
+      "step": 2635
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 0.406578928232193,
+      "learning_rate": 0.0003134805394852869,
+      "loss": 0.0379,
+      "step": 2636
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 0.5052029490470886,
+      "learning_rate": 0.0003132959964070897,
+      "loss": 0.06,
+      "step": 2637
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 0.4931176006793976,
+      "learning_rate": 0.0003131114482876255,
+      "loss": 0.0538,
+      "step": 2638
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 0.3985958695411682,
+      "learning_rate": 0.0003129268951968667,
+      "loss": 0.0488,
+      "step": 2639
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 0.43642348051071167,
+      "learning_rate": 0.0003127423372047881,
+      "loss": 0.0651,
+      "step": 2640
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 0.3645976781845093,
+      "learning_rate": 0.00031255777438136587,
+      "loss": 0.034,
+      "step": 2641
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 0.39518365263938904,
+      "learning_rate": 0.00031237320679657846,
+      "loss": 0.0461,
+      "step": 2642
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 0.45141878724098206,
+      "learning_rate": 0.00031218863452040575,
+      "loss": 0.0503,
+      "step": 2643
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 0.4226429760456085,
+      "learning_rate": 0.0003120040576228298,
+      "loss": 0.0529,
+      "step": 2644
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 0.3775617480278015,
+      "learning_rate": 0.00031181947617383394,
+      "loss": 0.0498,
+      "step": 2645
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 0.36932629346847534,
+      "learning_rate": 0.00031163489024340366,
+      "loss": 0.0497,
+      "step": 2646
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 0.46100884675979614,
+      "learning_rate": 0.0003114502999015259,
+      "loss": 0.038,
+      "step": 2647
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 0.44295117259025574,
+      "learning_rate": 0.0003112657052181892,
+      "loss": 0.0604,
+      "step": 2648
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 0.5597884058952332,
+      "learning_rate": 0.00031108110626338405,
+      "loss": 0.0544,
+      "step": 2649
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 0.5484944581985474,
+      "learning_rate": 0.00031089650310710243,
+      "loss": 0.0515,
+      "step": 2650
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 0.48890072107315063,
+      "learning_rate": 0.00031071189581933785,
+      "loss": 0.0742,
+      "step": 2651
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 0.4310765564441681,
+      "learning_rate": 0.0003105272844700852,
+      "loss": 0.0538,
+      "step": 2652
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 0.36987587809562683,
+      "learning_rate": 0.00031034266912934147,
+      "loss": 0.0475,
+      "step": 2653
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 0.46720781922340393,
+      "learning_rate": 0.00031015804986710463,
+      "loss": 0.0808,
+      "step": 2654
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 0.3207816779613495,
+      "learning_rate": 0.0003099734267533743,
+      "loss": 0.0437,
+      "step": 2655
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 0.5102361440658569,
+      "learning_rate": 0.0003097887998581515,
+      "loss": 0.0757,
+      "step": 2656
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 0.39079058170318604,
+      "learning_rate": 0.0003096041692514391,
+      "loss": 0.0488,
+      "step": 2657
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 0.41659146547317505,
+      "learning_rate": 0.0003094195350032407,
+      "loss": 0.0518,
+      "step": 2658
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 0.3961039185523987,
+      "learning_rate": 0.0003092348971835619,
+      "loss": 0.036,
+      "step": 2659
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 0.4981707036495209,
+      "learning_rate": 0.00030905025586240917,
+      "loss": 0.0699,
+      "step": 2660
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 0.4812929034233093,
+      "learning_rate": 0.0003088656111097906,
+      "loss": 0.0539,
+      "step": 2661
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 0.4777328073978424,
+      "learning_rate": 0.0003086809629957154,
+      "loss": 0.05,
+      "step": 2662
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 0.5160920023918152,
+      "learning_rate": 0.0003084963115901943,
+      "loss": 0.0583,
+      "step": 2663
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 0.44199061393737793,
+      "learning_rate": 0.0003083116569632392,
+      "loss": 0.046,
+      "step": 2664
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 0.4930119514465332,
+      "learning_rate": 0.0003081269991848629,
+      "loss": 0.0717,
+      "step": 2665
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 0.4718412756919861,
+      "learning_rate": 0.00030794233832507977,
+      "loss": 0.0666,
+      "step": 2666
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 0.4780096709728241,
+      "learning_rate": 0.00030775767445390535,
+      "loss": 0.0506,
+      "step": 2667
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 0.5419859290122986,
+      "learning_rate": 0.00030757300764135603,
+      "loss": 0.0826,
+      "step": 2668
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 0.4335952699184418,
+      "learning_rate": 0.0003073883379574496,
+      "loss": 0.0545,
+      "step": 2669
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 0.4504346251487732,
+      "learning_rate": 0.00030720366547220466,
+      "loss": 0.057,
+      "step": 2670
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 0.40275707840919495,
+      "learning_rate": 0.0003070189902556412,
+      "loss": 0.044,
+      "step": 2671
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 0.47973746061325073,
+      "learning_rate": 0.00030683431237777997,
+      "loss": 0.0504,
+      "step": 2672
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 0.40677502751350403,
+      "learning_rate": 0.0003066496319086429,
+      "loss": 0.044,
+      "step": 2673
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 0.505141019821167,
+      "learning_rate": 0.0003064649489182528,
+      "loss": 0.0698,
+      "step": 2674
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 0.45909202098846436,
+      "learning_rate": 0.00030628026347663355,
+      "loss": 0.0482,
+      "step": 2675
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 0.40435469150543213,
+      "learning_rate": 0.0003060955756538098,
+      "loss": 0.0353,
+      "step": 2676
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 0.4769234359264374,
+      "learning_rate": 0.0003059108855198073,
+      "loss": 0.0559,
+      "step": 2677
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 0.4823453724384308,
+      "learning_rate": 0.0003057261931446524,
+      "loss": 0.0819,
+      "step": 2678
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 0.3868640065193176,
+      "learning_rate": 0.00030554149859837254,
+      "loss": 0.0424,
+      "step": 2679
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 0.40343084931373596,
+      "learning_rate": 0.000305356801950996,
+      "loss": 0.042,
+      "step": 2680
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 0.35556212067604065,
+      "learning_rate": 0.0003051721032725518,
+      "loss": 0.0411,
+      "step": 2681
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 0.5034061670303345,
+      "learning_rate": 0.00030498740263306946,
+      "loss": 0.0399,
+      "step": 2682
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 0.4321735203266144,
+      "learning_rate": 0.0003048027001025797,
+      "loss": 0.0591,
+      "step": 2683
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 0.4233998954296112,
+      "learning_rate": 0.0003046179957511136,
+      "loss": 0.0431,
+      "step": 2684
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 0.3716711401939392,
+      "learning_rate": 0.0003044332896487031,
+      "loss": 0.0439,
+      "step": 2685
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 0.4412180483341217,
+      "learning_rate": 0.00030424858186538077,
+      "loss": 0.0346,
+      "step": 2686
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 0.4242977201938629,
+      "learning_rate": 0.00030406387247117986,
+      "loss": 0.0355,
+      "step": 2687
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 0.3777371346950531,
+      "learning_rate": 0.0003038791615361341,
+      "loss": 0.0369,
+      "step": 2688
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 0.3569364547729492,
+      "learning_rate": 0.00030369444913027797,
+      "loss": 0.0382,
+      "step": 2689
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 0.3983069956302643,
+      "learning_rate": 0.00030350973532364647,
+      "loss": 0.0434,
+      "step": 2690
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 0.5013437271118164,
+      "learning_rate": 0.00030332502018627503,
+      "loss": 0.0536,
+      "step": 2691
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 0.43859267234802246,
+      "learning_rate": 0.0003031403037881997,
+      "loss": 0.0523,
+      "step": 2692
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 0.44409245252609253,
+      "learning_rate": 0.0003029555861994568,
+      "loss": 0.0636,
+      "step": 2693
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 0.36404159665107727,
+      "learning_rate": 0.00030277086749008346,
+      "loss": 0.0418,
+      "step": 2694
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 0.3989045321941376,
+      "learning_rate": 0.0003025861477301168,
+      "loss": 0.0431,
+      "step": 2695
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 0.5098609924316406,
+      "learning_rate": 0.00030240142698959477,
+      "loss": 0.0554,
+      "step": 2696
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 0.46752968430519104,
+      "learning_rate": 0.0003022167053385555,
+      "loss": 0.0436,
+      "step": 2697
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 0.3948163390159607,
+      "learning_rate": 0.00030203198284703723,
+      "loss": 0.0454,
+      "step": 2698
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 0.34104257822036743,
+      "learning_rate": 0.00030184725958507896,
+      "loss": 0.0345,
+      "step": 2699
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 0.5428547859191895,
+      "learning_rate": 0.0003016625356227198,
+      "loss": 0.0907,
+      "step": 2700
+    },
+    {
+      "epoch": 5.13,
+      "eval_blimp_filtered_avg": 0.5382089552238806,
+      "eval_blimp_filtered_std": 0.005887411000263488,
+      "step": 2700
+    },
+    {
+      "epoch": 5.13,
+      "eval_blimp_supplement_avg": 0.5560344827586207,
+      "eval_blimp_supplement_std": 0.023153624643285433,
+      "step": 2700
+    },
+    {
+      "epoch": 5.13,
+      "eval_vqa_filtered_avg": 0.17,
+      "eval_vqa_filtered_std": 0.03775251680686371,
+      "step": 2700
+    },
+    {
+      "epoch": 5.13,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 2700
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 0.3808356523513794,
+      "learning_rate": 0.00030147781102999883,
+      "loss": 0.0464,
+      "step": 2701
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 0.37783917784690857,
+      "learning_rate": 0.0003012930858769559,
+      "loss": 0.0415,
+      "step": 2702
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 0.4322805404663086,
+      "learning_rate": 0.00030110836023363065,
+      "loss": 0.0434,
+      "step": 2703
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 0.550209105014801,
+      "learning_rate": 0.00030092363417006307,
+      "loss": 0.0356,
+      "step": 2704
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 0.41808760166168213,
+      "learning_rate": 0.0003007389077562934,
+      "loss": 0.0505,
+      "step": 2705
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 0.3170402944087982,
+      "learning_rate": 0.0003005541810623619,
+      "loss": 0.0353,
+      "step": 2706
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 0.4534017741680145,
+      "learning_rate": 0.0003003694541583088,
+      "loss": 0.0419,
+      "step": 2707
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 0.37658482789993286,
+      "learning_rate": 0.00030018472711417464,
+      "loss": 0.0315,
+      "step": 2708
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 0.3348631262779236,
+      "learning_rate": 0.0003,
+      "loss": 0.0333,
+      "step": 2709
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 0.35804057121276855,
+      "learning_rate": 0.00029981527288582525,
+      "loss": 0.0263,
+      "step": 2710
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 0.35514193773269653,
+      "learning_rate": 0.00029963054584169124,
+      "loss": 0.0314,
+      "step": 2711
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 0.5579752326011658,
+      "learning_rate": 0.0002994458189376381,
+      "loss": 0.0817,
+      "step": 2712
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 0.3677256107330322,
+      "learning_rate": 0.00029926109224370654,
+      "loss": 0.0501,
+      "step": 2713
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 0.5723695158958435,
+      "learning_rate": 0.0002990763658299369,
+      "loss": 0.064,
+      "step": 2714
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 0.33839330077171326,
+      "learning_rate": 0.00029889163976636935,
+      "loss": 0.0351,
+      "step": 2715
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 0.3184431791305542,
+      "learning_rate": 0.00029870691412304406,
+      "loss": 0.0402,
+      "step": 2716
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 0.4733424186706543,
+      "learning_rate": 0.00029852218897000117,
+      "loss": 0.0472,
+      "step": 2717
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 0.29669249057769775,
+      "learning_rate": 0.00029833746437728025,
+      "loss": 0.0292,
+      "step": 2718
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 0.49216800928115845,
+      "learning_rate": 0.00029815274041492093,
+      "loss": 0.0534,
+      "step": 2719
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 0.46354708075523376,
+      "learning_rate": 0.0002979680171529627,
+      "loss": 0.0441,
+      "step": 2720
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 0.3705219030380249,
+      "learning_rate": 0.00029778329466144453,
+      "loss": 0.0421,
+      "step": 2721
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 0.3699526786804199,
+      "learning_rate": 0.00029759857301040523,
+      "loss": 0.035,
+      "step": 2722
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 0.33144092559814453,
+      "learning_rate": 0.0002974138522698832,
+      "loss": 0.0305,
+      "step": 2723
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 0.462199330329895,
+      "learning_rate": 0.00029722913250991654,
+      "loss": 0.0538,
+      "step": 2724
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 0.4759749472141266,
+      "learning_rate": 0.0002970444138005432,
+      "loss": 0.028,
+      "step": 2725
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 0.3873721659183502,
+      "learning_rate": 0.0002968596962118003,
+      "loss": 0.0401,
+      "step": 2726
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 0.45062869787216187,
+      "learning_rate": 0.000296674979813725,
+      "loss": 0.0479,
+      "step": 2727
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 0.49729955196380615,
+      "learning_rate": 0.0002964902646763535,
+      "loss": 0.0543,
+      "step": 2728
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 0.38274499773979187,
+      "learning_rate": 0.000296305550869722,
+      "loss": 0.049,
+      "step": 2729
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 0.4341006577014923,
+      "learning_rate": 0.00029612083846386586,
+      "loss": 0.0388,
+      "step": 2730
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 0.43312782049179077,
+      "learning_rate": 0.00029593612752882025,
+      "loss": 0.0415,
+      "step": 2731
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 0.5027369260787964,
+      "learning_rate": 0.00029575141813461924,
+      "loss": 0.0523,
+      "step": 2732
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 0.4587538540363312,
+      "learning_rate": 0.00029556671035129686,
+      "loss": 0.0577,
+      "step": 2733
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 0.46108585596084595,
+      "learning_rate": 0.00029538200424888647,
+      "loss": 0.0409,
+      "step": 2734
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 0.4875165820121765,
+      "learning_rate": 0.0002951972998974203,
+      "loss": 0.0567,
+      "step": 2735
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 0.4829827547073364,
+      "learning_rate": 0.0002950125973669305,
+      "loss": 0.0428,
+      "step": 2736
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 0.44230854511260986,
+      "learning_rate": 0.0002948278967274482,
+      "loss": 0.0459,
+      "step": 2737
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 0.35536667704582214,
+      "learning_rate": 0.00029464319804900393,
+      "loss": 0.0382,
+      "step": 2738
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 0.42008543014526367,
+      "learning_rate": 0.00029445850140162735,
+      "loss": 0.0496,
+      "step": 2739
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 0.44505542516708374,
+      "learning_rate": 0.0002942738068553476,
+      "loss": 0.0547,
+      "step": 2740
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 0.6243768334388733,
+      "learning_rate": 0.0002940891144801927,
+      "loss": 0.0691,
+      "step": 2741
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 0.38714170455932617,
+      "learning_rate": 0.00029390442434619013,
+      "loss": 0.0412,
+      "step": 2742
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 0.48365822434425354,
+      "learning_rate": 0.0002937197365233665,
+      "loss": 0.0535,
+      "step": 2743
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 0.3442947268486023,
+      "learning_rate": 0.00029353505108174716,
+      "loss": 0.0367,
+      "step": 2744
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 0.5332748889923096,
+      "learning_rate": 0.00029335036809135704,
+      "loss": 0.0612,
+      "step": 2745
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 0.29757511615753174,
+      "learning_rate": 0.0002931656876222201,
+      "loss": 0.0364,
+      "step": 2746
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 0.48824310302734375,
+      "learning_rate": 0.0002929810097443588,
+      "loss": 0.0598,
+      "step": 2747
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 0.5269879102706909,
+      "learning_rate": 0.0002927963345277954,
+      "loss": 0.0587,
+      "step": 2748
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 0.3221952021121979,
+      "learning_rate": 0.00029261166204255047,
+      "loss": 0.0321,
+      "step": 2749
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 0.3070559799671173,
+      "learning_rate": 0.0002924269923586439,
+      "loss": 0.0358,
+      "step": 2750
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 0.5049554109573364,
+      "learning_rate": 0.00029224232554609465,
+      "loss": 0.0607,
+      "step": 2751
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 0.3793417811393738,
+      "learning_rate": 0.0002920576616749202,
+      "loss": 0.0342,
+      "step": 2752
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 0.3775675296783447,
+      "learning_rate": 0.000291873000815137,
+      "loss": 0.0423,
+      "step": 2753
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 0.5031003952026367,
+      "learning_rate": 0.00029168834303676085,
+      "loss": 0.0594,
+      "step": 2754
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 0.40328970551490784,
+      "learning_rate": 0.00029150368840980565,
+      "loss": 0.0462,
+      "step": 2755
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 0.4050198197364807,
+      "learning_rate": 0.0002913190370042845,
+      "loss": 0.0383,
+      "step": 2756
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 0.35440972447395325,
+      "learning_rate": 0.0002911343888902095,
+      "loss": 0.0334,
+      "step": 2757
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 0.3839973211288452,
+      "learning_rate": 0.00029094974413759083,
+      "loss": 0.0392,
+      "step": 2758
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 0.4825478196144104,
+      "learning_rate": 0.0002907651028164381,
+      "loss": 0.0433,
+      "step": 2759
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 0.5040748715400696,
+      "learning_rate": 0.0002905804649967592,
+      "loss": 0.0541,
+      "step": 2760
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 0.3576754629611969,
+      "learning_rate": 0.0002903958307485609,
+      "loss": 0.0341,
+      "step": 2761
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 0.45123717188835144,
+      "learning_rate": 0.00029021120014184837,
+      "loss": 0.0387,
+      "step": 2762
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 0.44147297739982605,
+      "learning_rate": 0.00029002657324662576,
+      "loss": 0.0431,
+      "step": 2763
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 0.4540736973285675,
+      "learning_rate": 0.0002898419501328954,
+      "loss": 0.0409,
+      "step": 2764
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 0.45698055624961853,
+      "learning_rate": 0.0002896573308706585,
+      "loss": 0.0586,
+      "step": 2765
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 0.4165416359901428,
+      "learning_rate": 0.00028947271552991476,
+      "loss": 0.0446,
+      "step": 2766
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 0.4641492962837219,
+      "learning_rate": 0.00028928810418066216,
+      "loss": 0.0486,
+      "step": 2767
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 0.48310142755508423,
+      "learning_rate": 0.00028910349689289746,
+      "loss": 0.0647,
+      "step": 2768
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 0.4961323142051697,
+      "learning_rate": 0.0002889188937366159,
+      "loss": 0.0514,
+      "step": 2769
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 0.5789635181427002,
+      "learning_rate": 0.00028873429478181074,
+      "loss": 0.0656,
+      "step": 2770
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 0.30838021636009216,
+      "learning_rate": 0.00028854970009847423,
+      "loss": 0.0227,
+      "step": 2771
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 0.3947138786315918,
+      "learning_rate": 0.00028836510975659634,
+      "loss": 0.0312,
+      "step": 2772
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 0.3990902602672577,
+      "learning_rate": 0.00028818052382616606,
+      "loss": 0.0388,
+      "step": 2773
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 0.49722957611083984,
+      "learning_rate": 0.0002879959423771702,
+      "loss": 0.0602,
+      "step": 2774
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 0.3216679096221924,
+      "learning_rate": 0.0002878113654795942,
+      "loss": 0.0238,
+      "step": 2775
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 0.34644851088523865,
+      "learning_rate": 0.00028762679320342154,
+      "loss": 0.0364,
+      "step": 2776
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 0.3966728448867798,
+      "learning_rate": 0.00028744222561863413,
+      "loss": 0.0394,
+      "step": 2777
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 0.4966990053653717,
+      "learning_rate": 0.00028725766279521196,
+      "loss": 0.0411,
+      "step": 2778
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 0.45808470249176025,
+      "learning_rate": 0.00028707310480313323,
+      "loss": 0.0514,
+      "step": 2779
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 0.4943194091320038,
+      "learning_rate": 0.0002868885517123746,
+      "loss": 0.064,
+      "step": 2780
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 0.5746262669563293,
+      "learning_rate": 0.00028670400359291024,
+      "loss": 0.0647,
+      "step": 2781
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 0.41655635833740234,
+      "learning_rate": 0.00028651946051471317,
+      "loss": 0.0436,
+      "step": 2782
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 0.3899505138397217,
+      "learning_rate": 0.00028633492254775397,
+      "loss": 0.0441,
+      "step": 2783
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 0.4480680823326111,
+      "learning_rate": 0.00028615038976200147,
+      "loss": 0.0558,
+      "step": 2784
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 0.3700203597545624,
+      "learning_rate": 0.0002859658622274225,
+      "loss": 0.0316,
+      "step": 2785
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 0.5466910600662231,
+      "learning_rate": 0.0002857813400139822,
+      "loss": 0.049,
+      "step": 2786
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 0.43631547689437866,
+      "learning_rate": 0.00028559682319164294,
+      "loss": 0.0416,
+      "step": 2787
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 0.3977242708206177,
+      "learning_rate": 0.0002854123118303658,
+      "loss": 0.046,
+      "step": 2788
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 0.4939308166503906,
+      "learning_rate": 0.0002852278060001097,
+      "loss": 0.0447,
+      "step": 2789
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 0.36392533779144287,
+      "learning_rate": 0.0002850433057708309,
+      "loss": 0.0384,
+      "step": 2790
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 0.4104593098163605,
+      "learning_rate": 0.00028485881121248404,
+      "loss": 0.0489,
+      "step": 2791
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 0.5253554582595825,
+      "learning_rate": 0.0002846743223950217,
+      "loss": 0.0739,
+      "step": 2792
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 0.2664570212364197,
+      "learning_rate": 0.00028448983938839377,
+      "loss": 0.0255,
+      "step": 2793
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 0.4084486961364746,
+      "learning_rate": 0.00028430536226254835,
+      "loss": 0.0562,
+      "step": 2794
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 0.32501524686813354,
+      "learning_rate": 0.0002841208910874313,
+      "loss": 0.0313,
+      "step": 2795
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 0.5583932995796204,
+      "learning_rate": 0.0002839364259329859,
+      "loss": 0.0811,
+      "step": 2796
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 0.504929780960083,
+      "learning_rate": 0.00028375196686915353,
+      "loss": 0.0693,
+      "step": 2797
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 0.4051588475704193,
+      "learning_rate": 0.00028356751396587304,
+      "loss": 0.0489,
+      "step": 2798
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 0.35525748133659363,
+      "learning_rate": 0.0002833830672930809,
+      "loss": 0.0409,
+      "step": 2799
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 0.4288542866706848,
+      "learning_rate": 0.00028319862692071155,
+      "loss": 0.0499,
+      "step": 2800
+    },
+    {
+      "epoch": 5.32,
+      "eval_blimp_filtered_avg": 0.538955223880597,
+      "eval_blimp_filtered_std": 0.005815171508264413,
+      "step": 2800
+    },
+    {
+      "epoch": 5.32,
+      "eval_blimp_supplement_avg": 0.5150862068965517,
+      "eval_blimp_supplement_std": 0.02318250446073278,
+      "step": 2800
+    },
+    {
+      "epoch": 5.32,
+      "eval_vqa_filtered_avg": 0.17,
+      "eval_vqa_filtered_std": 0.03775251680686371,
+      "step": 2800
+    },
+    {
+      "epoch": 5.32,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 2800
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 0.3115423619747162,
+      "learning_rate": 0.00028301419291869653,
+      "loss": 0.0331,
+      "step": 2801
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 0.3514610230922699,
+      "learning_rate": 0.0002828297653569654,
+      "loss": 0.0516,
+      "step": 2802
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 0.33785974979400635,
+      "learning_rate": 0.00028264534430544525,
+      "loss": 0.0431,
+      "step": 2803
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 0.5198425650596619,
+      "learning_rate": 0.0002824609298340603,
+      "loss": 0.0652,
+      "step": 2804
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 0.3130654990673065,
+      "learning_rate": 0.0002822765220127327,
+      "loss": 0.0419,
+      "step": 2805
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 0.39017173647880554,
+      "learning_rate": 0.0002820921209113819,
+      "loss": 0.052,
+      "step": 2806
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 0.4749644100666046,
+      "learning_rate": 0.00028190772659992484,
+      "loss": 0.047,
+      "step": 2807
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 0.4178813099861145,
+      "learning_rate": 0.00028172333914827576,
+      "loss": 0.0451,
+      "step": 2808
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 0.4907459020614624,
+      "learning_rate": 0.0002815389586263466,
+      "loss": 0.0578,
+      "step": 2809
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 0.49824410676956177,
+      "learning_rate": 0.0002813545851040463,
+      "loss": 0.047,
+      "step": 2810
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 0.4893423914909363,
+      "learning_rate": 0.00028117021865128145,
+      "loss": 0.0526,
+      "step": 2811
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 0.5088124871253967,
+      "learning_rate": 0.0002809858593379559,
+      "loss": 0.0602,
+      "step": 2812
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 0.4538384675979614,
+      "learning_rate": 0.00028080150723397055,
+      "loss": 0.0545,
+      "step": 2813
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 0.37545207142829895,
+      "learning_rate": 0.0002806171624092238,
+      "loss": 0.0424,
+      "step": 2814
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 0.5451890230178833,
+      "learning_rate": 0.0002804328249336115,
+      "loss": 0.0715,
+      "step": 2815
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 0.34766528010368347,
+      "learning_rate": 0.00028024849487702605,
+      "loss": 0.0353,
+      "step": 2816
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 0.42961448431015015,
+      "learning_rate": 0.0002800641723093577,
+      "loss": 0.0413,
+      "step": 2817
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 0.36667171120643616,
+      "learning_rate": 0.0002798798573004935,
+      "loss": 0.0345,
+      "step": 2818
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 0.48910245299339294,
+      "learning_rate": 0.0002796955499203178,
+      "loss": 0.0525,
+      "step": 2819
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 0.4113122224807739,
+      "learning_rate": 0.0002795112502387119,
+      "loss": 0.0576,
+      "step": 2820
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 0.49643710255622864,
+      "learning_rate": 0.00027932695832555437,
+      "loss": 0.0418,
+      "step": 2821
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 0.3768112361431122,
+      "learning_rate": 0.0002791426742507206,
+      "loss": 0.0397,
+      "step": 2822
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 0.43932485580444336,
+      "learning_rate": 0.00027895839808408343,
+      "loss": 0.0564,
+      "step": 2823
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 0.36575108766555786,
+      "learning_rate": 0.000278774129895512,
+      "loss": 0.0384,
+      "step": 2824
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 0.5122680068016052,
+      "learning_rate": 0.0002785898697548732,
+      "loss": 0.0552,
+      "step": 2825
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 0.4354902505874634,
+      "learning_rate": 0.0002784056177320304,
+      "loss": 0.0431,
+      "step": 2826
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 0.5654293894767761,
+      "learning_rate": 0.0002782213738968439,
+      "loss": 0.0804,
+      "step": 2827
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 0.43214184045791626,
+      "learning_rate": 0.00027803713831917113,
+      "loss": 0.0616,
+      "step": 2828
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 0.3421386182308197,
+      "learning_rate": 0.0002778529110688663,
+      "loss": 0.0393,
+      "step": 2829
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 0.35445839166641235,
+      "learning_rate": 0.00027766869221578026,
+      "loss": 0.034,
+      "step": 2830
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 0.4979478716850281,
+      "learning_rate": 0.00027748448182976087,
+      "loss": 0.0883,
+      "step": 2831
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 0.46812954545021057,
+      "learning_rate": 0.0002773002799806529,
+      "loss": 0.0385,
+      "step": 2832
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 0.4242936670780182,
+      "learning_rate": 0.0002771160867382975,
+      "loss": 0.0576,
+      "step": 2833
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 0.3198358714580536,
+      "learning_rate": 0.0002769319021725328,
+      "loss": 0.0278,
+      "step": 2834
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 0.4773464500904083,
+      "learning_rate": 0.000276747726353194,
+      "loss": 0.0465,
+      "step": 2835
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 0.4128437042236328,
+      "learning_rate": 0.00027656355935011214,
+      "loss": 0.0483,
+      "step": 2836
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 0.47906726598739624,
+      "learning_rate": 0.00027637940123311557,
+      "loss": 0.0495,
+      "step": 2837
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 0.2760261297225952,
+      "learning_rate": 0.0002761952520720293,
+      "loss": 0.0285,
+      "step": 2838
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 0.38681235909461975,
+      "learning_rate": 0.00027601111193667434,
+      "loss": 0.0366,
+      "step": 2839
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 0.40353187918663025,
+      "learning_rate": 0.00027582698089686895,
+      "loss": 0.0553,
+      "step": 2840
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 0.4160085618495941,
+      "learning_rate": 0.0002756428590224276,
+      "loss": 0.0538,
+      "step": 2841
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 0.5189279317855835,
+      "learning_rate": 0.0002754587463831613,
+      "loss": 0.0434,
+      "step": 2842
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 0.3554925322532654,
+      "learning_rate": 0.0002752746430488776,
+      "loss": 0.0374,
+      "step": 2843
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 0.2795903980731964,
+      "learning_rate": 0.0002750905490893806,
+      "loss": 0.0219,
+      "step": 2844
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 0.3387110233306885,
+      "learning_rate": 0.0002749064645744707,
+      "loss": 0.0273,
+      "step": 2845
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 0.4420880079269409,
+      "learning_rate": 0.00027472238957394486,
+      "loss": 0.0331,
+      "step": 2846
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 0.2687292695045471,
+      "learning_rate": 0.0002745383241575963,
+      "loss": 0.0278,
+      "step": 2847
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 0.4204709529876709,
+      "learning_rate": 0.0002743542683952146,
+      "loss": 0.0545,
+      "step": 2848
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 0.3335634171962738,
+      "learning_rate": 0.000274170222356586,
+      "loss": 0.0266,
+      "step": 2849
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 0.4041450619697571,
+      "learning_rate": 0.0002739861861114925,
+      "loss": 0.0522,
+      "step": 2850
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 0.5606263279914856,
+      "learning_rate": 0.000273802159729713,
+      "loss": 0.0661,
+      "step": 2851
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 0.43006280064582825,
+      "learning_rate": 0.0002736181432810221,
+      "loss": 0.0469,
+      "step": 2852
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 0.5179421305656433,
+      "learning_rate": 0.000273434136835191,
+      "loss": 0.0404,
+      "step": 2853
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 0.3463863134384155,
+      "learning_rate": 0.0002732501404619868,
+      "loss": 0.0357,
+      "step": 2854
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 0.5452454686164856,
+      "learning_rate": 0.0002730661542311734,
+      "loss": 0.0657,
+      "step": 2855
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 0.5281289219856262,
+      "learning_rate": 0.00027288217821250997,
+      "loss": 0.0471,
+      "step": 2856
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 0.46149930357933044,
+      "learning_rate": 0.0002726982124757524,
+      "loss": 0.0498,
+      "step": 2857
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 0.5534565448760986,
+      "learning_rate": 0.0002725142570906528,
+      "loss": 0.0544,
+      "step": 2858
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 0.4322470426559448,
+      "learning_rate": 0.00027233031212695876,
+      "loss": 0.0364,
+      "step": 2859
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 0.4070103168487549,
+      "learning_rate": 0.0002721463776544144,
+      "loss": 0.0475,
+      "step": 2860
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 0.4376852810382843,
+      "learning_rate": 0.00027196245374275983,
+      "loss": 0.0407,
+      "step": 2861
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 0.3668881952762604,
+      "learning_rate": 0.0002717785404617308,
+      "loss": 0.0475,
+      "step": 2862
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 0.46823757886886597,
+      "learning_rate": 0.0002715946378810595,
+      "loss": 0.0398,
+      "step": 2863
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 0.3883928656578064,
+      "learning_rate": 0.00027141074607047377,
+      "loss": 0.0458,
+      "step": 2864
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 0.3741036355495453,
+      "learning_rate": 0.0002712268650996974,
+      "loss": 0.028,
+      "step": 2865
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 0.46202999353408813,
+      "learning_rate": 0.00027104299503845006,
+      "loss": 0.0499,
+      "step": 2866
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 0.6126335263252258,
+      "learning_rate": 0.0002708591359564476,
+      "loss": 0.077,
+      "step": 2867
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 0.4092845618724823,
+      "learning_rate": 0.000270675287923401,
+      "loss": 0.0374,
+      "step": 2868
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 0.4037605822086334,
+      "learning_rate": 0.0002704914510090179,
+      "loss": 0.0413,
+      "step": 2869
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 0.4286487400531769,
+      "learning_rate": 0.000270307625283001,
+      "loss": 0.0474,
+      "step": 2870
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 0.5369948148727417,
+      "learning_rate": 0.0002701238108150492,
+      "loss": 0.0973,
+      "step": 2871
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 0.5346407294273376,
+      "learning_rate": 0.00026994000767485715,
+      "loss": 0.0574,
+      "step": 2872
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 0.4451616406440735,
+      "learning_rate": 0.00026975621593211476,
+      "loss": 0.0596,
+      "step": 2873
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 0.3268740177154541,
+      "learning_rate": 0.00026957243565650814,
+      "loss": 0.0312,
+      "step": 2874
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 0.40490013360977173,
+      "learning_rate": 0.00026938866691771877,
+      "loss": 0.0395,
+      "step": 2875
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 0.33140814304351807,
+      "learning_rate": 0.0002692049097854238,
+      "loss": 0.0451,
+      "step": 2876
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 0.4777286648750305,
+      "learning_rate": 0.0002690211643292959,
+      "loss": 0.0386,
+      "step": 2877
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 0.42780759930610657,
+      "learning_rate": 0.00026883743061900363,
+      "loss": 0.0477,
+      "step": 2878
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 0.3630886375904083,
+      "learning_rate": 0.0002686537087242106,
+      "loss": 0.0461,
+      "step": 2879
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 0.3754870593547821,
+      "learning_rate": 0.00026846999871457634,
+      "loss": 0.0384,
+      "step": 2880
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 0.4762609302997589,
+      "learning_rate": 0.0002682863006597559,
+      "loss": 0.05,
+      "step": 2881
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 0.5779047608375549,
+      "learning_rate": 0.0002681026146293994,
+      "loss": 0.0599,
+      "step": 2882
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 0.4262893795967102,
+      "learning_rate": 0.0002679189406931527,
+      "loss": 0.0453,
+      "step": 2883
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 0.28400659561157227,
+      "learning_rate": 0.00026773527892065714,
+      "loss": 0.0348,
+      "step": 2884
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 0.3568739891052246,
+      "learning_rate": 0.0002675516293815492,
+      "loss": 0.0326,
+      "step": 2885
+    },
+    {
+      "epoch": 5.49,
+      "grad_norm": 0.434099942445755,
+      "learning_rate": 0.00026736799214546086,
+      "loss": 0.0346,
+      "step": 2886
+    },
+    {
+      "epoch": 5.49,
+      "grad_norm": 0.4433208405971527,
+      "learning_rate": 0.00026718436728201953,
+      "loss": 0.0537,
+      "step": 2887
+    },
+    {
+      "epoch": 5.49,
+      "grad_norm": 0.5115588307380676,
+      "learning_rate": 0.0002670007548608477,
+      "loss": 0.0469,
+      "step": 2888
+    },
+    {
+      "epoch": 5.49,
+      "grad_norm": 0.42881473898887634,
+      "learning_rate": 0.0002668171549515632,
+      "loss": 0.0491,
+      "step": 2889
+    },
+    {
+      "epoch": 5.49,
+      "grad_norm": 0.3860151767730713,
+      "learning_rate": 0.00026663356762377953,
+      "loss": 0.0406,
+      "step": 2890
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 0.4953013062477112,
+      "learning_rate": 0.0002664499929471046,
+      "loss": 0.0577,
+      "step": 2891
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 0.43686702847480774,
+      "learning_rate": 0.0002662664309911422,
+      "loss": 0.0402,
+      "step": 2892
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 0.46144405007362366,
+      "learning_rate": 0.0002660828818254914,
+      "loss": 0.0413,
+      "step": 2893
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 0.5204452276229858,
+      "learning_rate": 0.0002658993455197455,
+      "loss": 0.0502,
+      "step": 2894
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 0.43161049485206604,
+      "learning_rate": 0.0002657158221434941,
+      "loss": 0.0465,
+      "step": 2895
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 0.4888586401939392,
+      "learning_rate": 0.0002655323117663209,
+      "loss": 0.0563,
+      "step": 2896
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 0.6166937351226807,
+      "learning_rate": 0.0002653488144578054,
+      "loss": 0.078,
+      "step": 2897
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 0.3125656545162201,
+      "learning_rate": 0.00026516533028752166,
+      "loss": 0.0265,
+      "step": 2898
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 0.31984585523605347,
+      "learning_rate": 0.0002649818593250391,
+      "loss": 0.0347,
+      "step": 2899
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 0.4876939654350281,
+      "learning_rate": 0.00026479840163992174,
+      "loss": 0.0614,
+      "step": 2900
+    },
+    {
+      "epoch": 5.51,
+      "eval_blimp_filtered_avg": 0.5446268656716418,
+      "eval_blimp_filtered_std": 0.005787477472170901,
+      "step": 2900
+    },
+    {
+      "epoch": 5.51,
+      "eval_blimp_supplement_avg": 0.521551724137931,
+      "eval_blimp_supplement_std": 0.02308285623188346,
+      "step": 2900
+    },
+    {
+      "epoch": 5.51,
+      "eval_vqa_filtered_avg": 0.17,
+      "eval_vqa_filtered_std": 0.03775251680686371,
+      "step": 2900
+    },
+    {
+      "epoch": 5.51,
+      "eval_winoground_filtered_avg": 0.47,
+      "eval_winoground_filtered_std": 0.05016135580465919,
+      "step": 2900
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 0.4979645013809204,
+      "learning_rate": 0.0002646149573017292,
+      "loss": 0.0533,
+      "step": 2901
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 0.3132162392139435,
+      "learning_rate": 0.00026443152638001524,
+      "loss": 0.0299,
+      "step": 2902
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 0.5044434666633606,
+      "learning_rate": 0.0002642481089443291,
+      "loss": 0.0609,
+      "step": 2903
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 0.43881818652153015,
+      "learning_rate": 0.00026406470506421486,
+      "loss": 0.0367,
+      "step": 2904
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 0.5495441555976868,
+      "learning_rate": 0.00026388131480921113,
+      "loss": 0.0686,
+      "step": 2905
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 0.4598351716995239,
+      "learning_rate": 0.0002636979382488517,
+      "loss": 0.0538,
+      "step": 2906
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 0.412082314491272,
+      "learning_rate": 0.0002635145754526651,
+      "loss": 0.0526,
+      "step": 2907
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 0.308606892824173,
+      "learning_rate": 0.00026333122649017433,
+      "loss": 0.0251,
+      "step": 2908
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 0.5011122226715088,
+      "learning_rate": 0.00026314789143089766,
+      "loss": 0.0511,
+      "step": 2909
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 0.3463582396507263,
+      "learning_rate": 0.0002629645703443477,
+      "loss": 0.0413,
+      "step": 2910
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 0.43941983580589294,
+      "learning_rate": 0.00026278126330003186,
+      "loss": 0.0527,
+      "step": 2911
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 0.3931686282157898,
+      "learning_rate": 0.0002625979703674523,
+      "loss": 0.0486,
+      "step": 2912
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 0.37004947662353516,
+      "learning_rate": 0.0002624146916161058,
+      "loss": 0.0371,
+      "step": 2913
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 0.43847203254699707,
+      "learning_rate": 0.0002622314271154836,
+      "loss": 0.0486,
+      "step": 2914
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 0.3489550054073334,
+      "learning_rate": 0.00026204817693507176,
+      "loss": 0.0419,
+      "step": 2915
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 0.4067748486995697,
+      "learning_rate": 0.00026186494114435093,
+      "loss": 0.043,
+      "step": 2916
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 0.3405396640300751,
+      "learning_rate": 0.00026168171981279606,
+      "loss": 0.0444,
+      "step": 2917
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 0.43018588423728943,
+      "learning_rate": 0.0002614985130098769,
+      "loss": 0.0445,
+      "step": 2918
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 0.34504497051239014,
+      "learning_rate": 0.00026131532080505735,
+      "loss": 0.0247,
+      "step": 2919
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 0.4800398349761963,
+      "learning_rate": 0.0002611321432677961,
+      "loss": 0.0509,
+      "step": 2920
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 0.45939138531684875,
+      "learning_rate": 0.00026094898046754623,
+      "loss": 0.0584,
+      "step": 2921
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 0.42499038577079773,
+      "learning_rate": 0.000260765832473755,
+      "loss": 0.0338,
+      "step": 2922
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 0.4249691963195801,
+      "learning_rate": 0.00026058269935586427,
+      "loss": 0.0444,
+      "step": 2923
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 0.4069173038005829,
+      "learning_rate": 0.0002603995811833103,
+      "loss": 0.0419,
+      "step": 2924
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 0.30068477988243103,
+      "learning_rate": 0.0002602164780255234,
+      "loss": 0.0216,
+      "step": 2925
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 0.3432515561580658,
+      "learning_rate": 0.0002600333899519285,
+      "loss": 0.0396,
+      "step": 2926
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 0.495832622051239,
+      "learning_rate": 0.00025985031703194475,
+      "loss": 0.0601,
+      "step": 2927
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 0.435499370098114,
+      "learning_rate": 0.0002596672593349853,
+      "loss": 0.0537,
+      "step": 2928
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 0.5717827081680298,
+      "learning_rate": 0.0002594842169304578,
+      "loss": 0.0618,
+      "step": 2929
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 0.45532914996147156,
+      "learning_rate": 0.0002593011898877642,
+      "loss": 0.0449,
+      "step": 2930
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 0.5685784220695496,
+      "learning_rate": 0.0002591181782763001,
+      "loss": 0.0696,
+      "step": 2931
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 0.3464115262031555,
+      "learning_rate": 0.00025893518216545586,
+      "loss": 0.037,
+      "step": 2932
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 0.5351083874702454,
+      "learning_rate": 0.00025875220162461555,
+      "loss": 0.0518,
+      "step": 2933
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 0.4169665277004242,
+      "learning_rate": 0.0002585692367231576,
+      "loss": 0.0384,
+      "step": 2934
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 0.3790469467639923,
+      "learning_rate": 0.00025838628753045424,
+      "loss": 0.0427,
+      "step": 2935
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 0.40669143199920654,
+      "learning_rate": 0.00025820335411587213,
+      "loss": 0.0543,
+      "step": 2936
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 0.388621985912323,
+      "learning_rate": 0.00025802043654877144,
+      "loss": 0.0462,
+      "step": 2937
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 0.5100823044776917,
+      "learning_rate": 0.00025783753489850675,
+      "loss": 0.0589,
+      "step": 2938
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 0.36394989490509033,
+      "learning_rate": 0.00025765464923442654,
+      "loss": 0.0345,
+      "step": 2939
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 0.419136106967926,
+      "learning_rate": 0.0002574717796258729,
+      "loss": 0.04,
+      "step": 2940
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 0.3006397783756256,
+      "learning_rate": 0.0002572889261421823,
+      "loss": 0.0288,
+      "step": 2941
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 0.5168293118476868,
+      "learning_rate": 0.0002571060888526849,
+      "loss": 0.0512,
+      "step": 2942
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.39365410804748535,
+      "learning_rate": 0.0002569232678267046,
+      "loss": 0.0456,
+      "step": 2943
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.3430064618587494,
+      "learning_rate": 0.00025674046313355916,
+      "loss": 0.0346,
+      "step": 2944
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.40416017174720764,
+      "learning_rate": 0.0002565576748425604,
+      "loss": 0.0548,
+      "step": 2945
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.37429681420326233,
+      "learning_rate": 0.00025637490302301354,
+      "loss": 0.0278,
+      "step": 2946
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.3777844309806824,
+      "learning_rate": 0.00025619214774421793,
+      "loss": 0.0389,
+      "step": 2947
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.4469503164291382,
+      "learning_rate": 0.00025600940907546634,
+      "loss": 0.0488,
+      "step": 2948
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 0.3601633906364441,
+      "learning_rate": 0.00025582668708604535,
+      "loss": 0.0345,
+      "step": 2949
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 0.3557094633579254,
+      "learning_rate": 0.0002556439818452355,
+      "loss": 0.0261,
+      "step": 2950
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 0.37793466448783875,
+      "learning_rate": 0.0002554612934223104,
+      "loss": 0.0369,
+      "step": 2951
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 0.4145800769329071,
+      "learning_rate": 0.00025527862188653777,
+      "loss": 0.0624,
+      "step": 2952
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 0.3975084125995636,
+      "learning_rate": 0.00025509596730717885,
+      "loss": 0.0405,
+      "step": 2953
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 0.4459436535835266,
+      "learning_rate": 0.00025491332975348814,
+      "loss": 0.0465,
+      "step": 2954
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 0.3743772804737091,
+      "learning_rate": 0.00025473070929471405,
+      "loss": 0.0547,
+      "step": 2955
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 0.524311900138855,
+      "learning_rate": 0.0002545481060000983,
+      "loss": 0.0601,
+      "step": 2956
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 0.35471147298812866,
+      "learning_rate": 0.00025436551993887624,
+      "loss": 0.0357,
+      "step": 2957
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 0.3998315632343292,
+      "learning_rate": 0.0002541829511802765,
+      "loss": 0.0313,
+      "step": 2958
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 0.45675143599510193,
+      "learning_rate": 0.00025400039979352147,
+      "loss": 0.0474,
+      "step": 2959
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 0.356831818819046,
+      "learning_rate": 0.0002538178658478265,
+      "loss": 0.0447,
+      "step": 2960
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 0.32891204953193665,
+      "learning_rate": 0.0002536353494124006,
+      "loss": 0.0415,
+      "step": 2961
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 0.45393189787864685,
+      "learning_rate": 0.00025345285055644633,
+      "loss": 0.0542,
+      "step": 2962
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 0.3583466708660126,
+      "learning_rate": 0.0002532703693491591,
+      "loss": 0.0368,
+      "step": 2963
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 0.40638118982315063,
+      "learning_rate": 0.0002530879058597281,
+      "loss": 0.0614,
+      "step": 2964
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 0.4496307373046875,
+      "learning_rate": 0.0002529054601573356,
+      "loss": 0.0598,
+      "step": 2965
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 0.4396252930164337,
+      "learning_rate": 0.000252723032311157,
+      "loss": 0.0471,
+      "step": 2966
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 0.32210424542427063,
+      "learning_rate": 0.0002525406223903611,
+      "loss": 0.0265,
+      "step": 2967
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 0.6607413291931152,
+      "learning_rate": 0.00025235823046410987,
+      "loss": 0.0802,
+      "step": 2968
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 0.46392929553985596,
+      "learning_rate": 0.0002521758566015584,
+      "loss": 0.0674,
+      "step": 2969
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 0.3126295804977417,
+      "learning_rate": 0.00025199350087185506,
+      "loss": 0.0272,
+      "step": 2970
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 0.4043475389480591,
+      "learning_rate": 0.0002518111633441411,
+      "loss": 0.0455,
+      "step": 2971
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 0.512768566608429,
+      "learning_rate": 0.0002516288440875512,
+      "loss": 0.0608,
+      "step": 2972
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 0.43042272329330444,
+      "learning_rate": 0.000251446543171213,
+      "loss": 0.0549,
+      "step": 2973
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 0.34047746658325195,
+      "learning_rate": 0.0002512642606642468,
+      "loss": 0.0368,
+      "step": 2974
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 0.3789752125740051,
+      "learning_rate": 0.00025108199663576654,
+      "loss": 0.0391,
+      "step": 2975
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 0.4403546452522278,
+      "learning_rate": 0.0002508997511548789,
+      "loss": 0.0472,
+      "step": 2976
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 0.3363628685474396,
+      "learning_rate": 0.00025071752429068326,
+      "loss": 0.0363,
+      "step": 2977
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 0.378674179315567,
+      "learning_rate": 0.00025053531611227234,
+      "loss": 0.0436,
+      "step": 2978
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 0.5137527585029602,
+      "learning_rate": 0.00025035312668873153,
+      "loss": 0.0485,
+      "step": 2979
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 0.47312799096107483,
+      "learning_rate": 0.0002501709560891392,
+      "loss": 0.0579,
+      "step": 2980
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 0.4063437581062317,
+      "learning_rate": 0.0002499888043825665,
+      "loss": 0.0399,
+      "step": 2981
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 0.35661932826042175,
+      "learning_rate": 0.00024980667163807766,
+      "loss": 0.0317,
+      "step": 2982
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 0.37903380393981934,
+      "learning_rate": 0.00024962455792472936,
+      "loss": 0.0309,
+      "step": 2983
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 0.5285203456878662,
+      "learning_rate": 0.0002494424633115713,
+      "loss": 0.0506,
+      "step": 2984
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 0.3365808427333832,
+      "learning_rate": 0.00024926038786764605,
+      "loss": 0.0442,
+      "step": 2985
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 0.4597436785697937,
+      "learning_rate": 0.00024907833166198847,
+      "loss": 0.0459,
+      "step": 2986
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 0.36294305324554443,
+      "learning_rate": 0.00024889629476362656,
+      "loss": 0.039,
+      "step": 2987
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 0.43336328864097595,
+      "learning_rate": 0.00024871427724158096,
+      "loss": 0.0364,
+      "step": 2988
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 0.35182854533195496,
+      "learning_rate": 0.0002485322791648646,
+      "loss": 0.0446,
+      "step": 2989
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 0.3300316035747528,
+      "learning_rate": 0.0002483503006024834,
+      "loss": 0.0385,
+      "step": 2990
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 0.4220917522907257,
+      "learning_rate": 0.00024816834162343587,
+      "loss": 0.0397,
+      "step": 2991
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 0.41422444581985474,
+      "learning_rate": 0.00024798640229671277,
+      "loss": 0.0383,
+      "step": 2992
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 0.3906545341014862,
+      "learning_rate": 0.00024780448269129786,
+      "loss": 0.0447,
+      "step": 2993
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 0.45704177021980286,
+      "learning_rate": 0.00024762258287616695,
+      "loss": 0.0353,
+      "step": 2994
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 0.4478970468044281,
+      "learning_rate": 0.0002474407029202887,
+      "loss": 0.0402,
+      "step": 2995
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 0.4143120348453522,
+      "learning_rate": 0.0002472588428926243,
+      "loss": 0.0304,
+      "step": 2996
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 0.43559446930885315,
+      "learning_rate": 0.0002470770028621269,
+      "loss": 0.041,
+      "step": 2997
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 0.4283859431743622,
+      "learning_rate": 0.00024689518289774254,
+      "loss": 0.0417,
+      "step": 2998
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 0.4785545766353607,
+      "learning_rate": 0.0002467133830684096,
+      "loss": 0.0605,
+      "step": 2999
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 0.3670130670070648,
+      "learning_rate": 0.0002465316034430585,
+      "loss": 0.0409,
+      "step": 3000
+    },
+    {
+      "epoch": 5.7,
+      "eval_blimp_filtered_avg": 0.5446268656716418,
+      "eval_blimp_filtered_std": 0.005820315642177976,
+      "step": 3000
+    },
+    {
+      "epoch": 5.7,
+      "eval_blimp_supplement_avg": 0.5344827586206896,
+      "eval_blimp_supplement_std": 0.02311766454994662,
+      "step": 3000
+    },
+    {
+      "epoch": 5.7,
+      "eval_vqa_filtered_avg": 0.15,
+      "eval_vqa_filtered_std": 0.03588702812826371,
+      "step": 3000
+    },
+    {
+      "epoch": 5.7,
+      "eval_winoground_filtered_avg": 0.52,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 3000
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 0.442426860332489,
+      "learning_rate": 0.0002463498440906124,
+      "loss": 0.0595,
+      "step": 3001
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 0.3948839008808136,
+      "learning_rate": 0.00024616810507998644,
+      "loss": 0.0403,
+      "step": 3002
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 0.3048163056373596,
+      "learning_rate": 0.0002459863864800883,
+      "loss": 0.0388,
+      "step": 3003
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 0.24428093433380127,
+      "learning_rate": 0.00024580468835981773,
+      "loss": 0.0152,
+      "step": 3004
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 0.39419296383857727,
+      "learning_rate": 0.00024562301078806704,
+      "loss": 0.0484,
+      "step": 3005
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 0.31799840927124023,
+      "learning_rate": 0.00024544135383372013,
+      "loss": 0.0351,
+      "step": 3006
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 0.3885533809661865,
+      "learning_rate": 0.0002452597175656537,
+      "loss": 0.044,
+      "step": 3007
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 0.5409144759178162,
+      "learning_rate": 0.00024507810205273646,
+      "loss": 0.0681,
+      "step": 3008
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 0.2788480520248413,
+      "learning_rate": 0.0002448965073638289,
+      "loss": 0.0216,
+      "step": 3009
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 0.2916741967201233,
+      "learning_rate": 0.00024471493356778405,
+      "loss": 0.0196,
+      "step": 3010
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 0.3180861175060272,
+      "learning_rate": 0.0002445333807334469,
+      "loss": 0.0295,
+      "step": 3011
+    },
+    {
+      "epoch": 5.73,
+      "grad_norm": 0.37642645835876465,
+      "learning_rate": 0.00024435184892965435,
+      "loss": 0.0481,
+      "step": 3012
+    },
+    {
+      "epoch": 5.73,
+      "grad_norm": 0.40529298782348633,
+      "learning_rate": 0.0002441703382252353,
+      "loss": 0.0358,
+      "step": 3013
+    },
+    {
+      "epoch": 5.73,
+      "grad_norm": 0.5886270999908447,
+      "learning_rate": 0.00024398884868901102,
+      "loss": 0.0664,
+      "step": 3014
+    },
+    {
+      "epoch": 5.73,
+      "grad_norm": 0.3947683274745941,
+      "learning_rate": 0.00024380738038979423,
+      "loss": 0.028,
+      "step": 3015
+    },
+    {
+      "epoch": 5.73,
+      "grad_norm": 0.35609379410743713,
+      "learning_rate": 0.0002436259333963901,
+      "loss": 0.0341,
+      "step": 3016
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 0.3925468623638153,
+      "learning_rate": 0.00024344450777759528,
+      "loss": 0.038,
+      "step": 3017
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 0.4578824043273926,
+      "learning_rate": 0.00024326310360219862,
+      "loss": 0.0597,
+      "step": 3018
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 0.46013832092285156,
+      "learning_rate": 0.00024308172093898085,
+      "loss": 0.0567,
+      "step": 3019
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 0.4215681552886963,
+      "learning_rate": 0.00024290035985671422,
+      "loss": 0.0275,
+      "step": 3020
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 0.38679444789886475,
+      "learning_rate": 0.0002427190204241631,
+      "loss": 0.044,
+      "step": 3021
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 0.27499550580978394,
+      "learning_rate": 0.00024253770271008378,
+      "loss": 0.0199,
+      "step": 3022
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 0.37019601464271545,
+      "learning_rate": 0.00024235640678322375,
+      "loss": 0.0327,
+      "step": 3023
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 0.3852914273738861,
+      "learning_rate": 0.00024217513271232278,
+      "loss": 0.0464,
+      "step": 3024
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 0.6182989478111267,
+      "learning_rate": 0.00024199388056611219,
+      "loss": 0.0673,
+      "step": 3025
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 0.42302730679512024,
+      "learning_rate": 0.0002418126504133149,
+      "loss": 0.0406,
+      "step": 3026
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 0.3673613667488098,
+      "learning_rate": 0.0002416314423226454,
+      "loss": 0.0471,
+      "step": 3027
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 0.45144200325012207,
+      "learning_rate": 0.0002414502563628103,
+      "loss": 0.0493,
+      "step": 3028
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 0.3667828440666199,
+      "learning_rate": 0.00024126909260250722,
+      "loss": 0.0376,
+      "step": 3029
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 0.3781014680862427,
+      "learning_rate": 0.00024108795111042562,
+      "loss": 0.0437,
+      "step": 3030
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 0.29906395077705383,
+      "learning_rate": 0.00024090683195524681,
+      "loss": 0.0258,
+      "step": 3031
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 0.3437903821468353,
+      "learning_rate": 0.000240725735205643,
+      "loss": 0.0304,
+      "step": 3032
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 0.3354373574256897,
+      "learning_rate": 0.0002405446609302784,
+      "loss": 0.0324,
+      "step": 3033
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 0.34696927666664124,
+      "learning_rate": 0.00024036360919780868,
+      "loss": 0.0459,
+      "step": 3034
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 0.41768476366996765,
+      "learning_rate": 0.00024018258007688065,
+      "loss": 0.0428,
+      "step": 3035
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 0.4441203474998474,
+      "learning_rate": 0.00024000157363613265,
+      "loss": 0.0551,
+      "step": 3036
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 0.36954382061958313,
+      "learning_rate": 0.0002398205899441948,
+      "loss": 0.0304,
+      "step": 3037
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 0.4440646171569824,
+      "learning_rate": 0.000239639629069688,
+      "loss": 0.0536,
+      "step": 3038
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 0.3333814740180969,
+      "learning_rate": 0.0002394586910812249,
+      "loss": 0.0225,
+      "step": 3039
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 0.36834025382995605,
+      "learning_rate": 0.00023927777604740953,
+      "loss": 0.0332,
+      "step": 3040
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 0.32448500394821167,
+      "learning_rate": 0.00023909688403683674,
+      "loss": 0.0309,
+      "step": 3041
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 0.42295992374420166,
+      "learning_rate": 0.00023891601511809323,
+      "loss": 0.0428,
+      "step": 3042
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 0.38845908641815186,
+      "learning_rate": 0.00023873516935975645,
+      "loss": 0.0317,
+      "step": 3043
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 0.408076673746109,
+      "learning_rate": 0.0002385543468303954,
+      "loss": 0.0373,
+      "step": 3044
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 0.3614228069782257,
+      "learning_rate": 0.00023837354759857036,
+      "loss": 0.0344,
+      "step": 3045
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 0.4713466763496399,
+      "learning_rate": 0.0002381927717328322,
+      "loss": 0.0432,
+      "step": 3046
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 0.42560234665870667,
+      "learning_rate": 0.00023801201930172361,
+      "loss": 0.0379,
+      "step": 3047
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 0.43267738819122314,
+      "learning_rate": 0.00023783129037377804,
+      "loss": 0.0401,
+      "step": 3048
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 0.3814980685710907,
+      "learning_rate": 0.00023765058501752,
+      "loss": 0.0364,
+      "step": 3049
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 0.4342520833015442,
+      "learning_rate": 0.00023746990330146515,
+      "loss": 0.0563,
+      "step": 3050
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 0.30293241143226624,
+      "learning_rate": 0.0002372892452941204,
+      "loss": 0.036,
+      "step": 3051
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 0.60707688331604,
+      "learning_rate": 0.00023710861106398314,
+      "loss": 0.0722,
+      "step": 3052
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 0.39471951127052307,
+      "learning_rate": 0.00023692800067954224,
+      "loss": 0.041,
+      "step": 3053
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 0.3401859700679779,
+      "learning_rate": 0.00023674741420927743,
+      "loss": 0.0217,
+      "step": 3054
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 0.4813999533653259,
+      "learning_rate": 0.00023656685172165911,
+      "loss": 0.0516,
+      "step": 3055
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 0.4148545265197754,
+      "learning_rate": 0.00023638631328514885,
+      "loss": 0.047,
+      "step": 3056
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 0.35554996132850647,
+      "learning_rate": 0.0002362057989681991,
+      "loss": 0.0376,
+      "step": 3057
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 0.39951813220977783,
+      "learning_rate": 0.00023602530883925303,
+      "loss": 0.0448,
+      "step": 3058
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 0.2676348090171814,
+      "learning_rate": 0.00023584484296674459,
+      "loss": 0.0321,
+      "step": 3059
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 0.45693063735961914,
+      "learning_rate": 0.00023566440141909885,
+      "loss": 0.0537,
+      "step": 3060
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 0.4101620316505432,
+      "learning_rate": 0.0002354839842647312,
+      "loss": 0.0414,
+      "step": 3061
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 0.34870341420173645,
+      "learning_rate": 0.00023530359157204815,
+      "loss": 0.0229,
+      "step": 3062
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 0.36582043766975403,
+      "learning_rate": 0.00023512322340944695,
+      "loss": 0.0431,
+      "step": 3063
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 0.46426981687545776,
+      "learning_rate": 0.00023494287984531512,
+      "loss": 0.0363,
+      "step": 3064
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 0.3857658803462982,
+      "learning_rate": 0.00023476256094803147,
+      "loss": 0.0426,
+      "step": 3065
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 0.4577321410179138,
+      "learning_rate": 0.00023458226678596482,
+      "loss": 0.0332,
+      "step": 3066
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 0.2657248377799988,
+      "learning_rate": 0.00023440199742747505,
+      "loss": 0.0272,
+      "step": 3067
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 0.3731611967086792,
+      "learning_rate": 0.00023422175294091267,
+      "loss": 0.0413,
+      "step": 3068
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 0.41995424032211304,
+      "learning_rate": 0.00023404153339461834,
+      "loss": 0.0439,
+      "step": 3069
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 0.38145866990089417,
+      "learning_rate": 0.00023386133885692367,
+      "loss": 0.0289,
+      "step": 3070
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 0.5210149884223938,
+      "learning_rate": 0.0002336811693961506,
+      "loss": 0.0452,
+      "step": 3071
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 0.3920343816280365,
+      "learning_rate": 0.00023350102508061163,
+      "loss": 0.0308,
+      "step": 3072
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 0.42188817262649536,
+      "learning_rate": 0.0002333209059786096,
+      "loss": 0.0341,
+      "step": 3073
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 0.46101319789886475,
+      "learning_rate": 0.00023314081215843808,
+      "loss": 0.0454,
+      "step": 3074
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 0.3901765048503876,
+      "learning_rate": 0.00023296074368838065,
+      "loss": 0.0272,
+      "step": 3075
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 0.4215056598186493,
+      "learning_rate": 0.00023278070063671158,
+      "loss": 0.0348,
+      "step": 3076
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 0.3967689573764801,
+      "learning_rate": 0.00023260068307169552,
+      "loss": 0.0347,
+      "step": 3077
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 0.5569227337837219,
+      "learning_rate": 0.00023242069106158716,
+      "loss": 0.0851,
+      "step": 3078
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 0.45878419280052185,
+      "learning_rate": 0.00023224072467463176,
+      "loss": 0.0536,
+      "step": 3079
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 0.24166451394557953,
+      "learning_rate": 0.00023206078397906493,
+      "loss": 0.0196,
+      "step": 3080
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 0.4917944371700287,
+      "learning_rate": 0.00023188086904311225,
+      "loss": 0.045,
+      "step": 3081
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 0.5125945210456848,
+      "learning_rate": 0.00023170097993498967,
+      "loss": 0.095,
+      "step": 3082
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 0.38339632749557495,
+      "learning_rate": 0.00023152111672290355,
+      "loss": 0.0317,
+      "step": 3083
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 0.3542669415473938,
+      "learning_rate": 0.00023134127947504996,
+      "loss": 0.0356,
+      "step": 3084
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 0.4589560627937317,
+      "learning_rate": 0.0002311614682596156,
+      "loss": 0.0551,
+      "step": 3085
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 0.3813551068305969,
+      "learning_rate": 0.0002309816831447772,
+      "loss": 0.0281,
+      "step": 3086
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 0.3179730474948883,
+      "learning_rate": 0.00023080192419870125,
+      "loss": 0.0429,
+      "step": 3087
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 0.31902262568473816,
+      "learning_rate": 0.0002306221914895447,
+      "loss": 0.0223,
+      "step": 3088
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 0.37548893690109253,
+      "learning_rate": 0.00023044248508545455,
+      "loss": 0.0417,
+      "step": 3089
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 0.36814895272254944,
+      "learning_rate": 0.00023026280505456748,
+      "loss": 0.0323,
+      "step": 3090
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 0.26369038224220276,
+      "learning_rate": 0.00023008315146501063,
+      "loss": 0.0219,
+      "step": 3091
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 0.37084677815437317,
+      "learning_rate": 0.00022990352438490063,
+      "loss": 0.0394,
+      "step": 3092
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 0.3798152804374695,
+      "learning_rate": 0.00022972392388234444,
+      "loss": 0.037,
+      "step": 3093
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 0.25273677706718445,
+      "learning_rate": 0.00022954435002543882,
+      "loss": 0.0198,
+      "step": 3094
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 0.31462985277175903,
+      "learning_rate": 0.00022936480288227032,
+      "loss": 0.0256,
+      "step": 3095
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 0.37683939933776855,
+      "learning_rate": 0.00022918528252091548,
+      "loss": 0.0381,
+      "step": 3096
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 0.3123359680175781,
+      "learning_rate": 0.00022900578900944082,
+      "loss": 0.0286,
+      "step": 3097
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 0.36059266328811646,
+      "learning_rate": 0.0002288263224159023,
+      "loss": 0.0305,
+      "step": 3098
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 0.5642847418785095,
+      "learning_rate": 0.00022864688280834598,
+      "loss": 0.0619,
+      "step": 3099
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 0.4070831835269928,
+      "learning_rate": 0.00022846747025480772,
+      "loss": 0.0456,
+      "step": 3100
+    },
+    {
+      "epoch": 5.89,
+      "eval_blimp_filtered_avg": 0.5444776119402985,
+      "eval_blimp_filtered_std": 0.005836626892526587,
+      "step": 3100
+    },
+    {
+      "epoch": 5.89,
+      "eval_blimp_supplement_avg": 0.5301724137931034,
+      "eval_blimp_supplement_std": 0.023145511550693363,
+      "step": 3100
+    },
+    {
+      "epoch": 5.89,
+      "eval_vqa_filtered_avg": 0.18,
+      "eval_vqa_filtered_std": 0.03861229196653694,
+      "step": 3100
+    },
+    {
+      "epoch": 5.89,
+      "eval_winoground_filtered_avg": 0.54,
+      "eval_winoground_filtered_std": 0.05009082659620333,
+      "step": 3100
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 0.3735406696796417,
+      "learning_rate": 0.00022828808482331278,
+      "loss": 0.0412,
+      "step": 3101
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 0.4649685323238373,
+      "learning_rate": 0.00022810872658187657,
+      "loss": 0.0456,
+      "step": 3102
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 0.3176480531692505,
+      "learning_rate": 0.00022792939559850389,
+      "loss": 0.0404,
+      "step": 3103
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 0.24864183366298676,
+      "learning_rate": 0.00022775009194118926,
+      "loss": 0.021,
+      "step": 3104
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 0.40167272090911865,
+      "learning_rate": 0.0002275708156779169,
+      "loss": 0.0366,
+      "step": 3105
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 0.31744638085365295,
+      "learning_rate": 0.00022739156687666076,
+      "loss": 0.0359,
+      "step": 3106
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 0.35199570655822754,
+      "learning_rate": 0.00022721234560538398,
+      "loss": 0.0364,
+      "step": 3107
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 0.40483370423316956,
+      "learning_rate": 0.0002270331519320397,
+      "loss": 0.0465,
+      "step": 3108
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 0.36396750807762146,
+      "learning_rate": 0.00022685398592457045,
+      "loss": 0.0381,
+      "step": 3109
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 0.3035101294517517,
+      "learning_rate": 0.00022667484765090805,
+      "loss": 0.0267,
+      "step": 3110
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 0.33764684200286865,
+      "learning_rate": 0.00022649573717897412,
+      "loss": 0.0381,
+      "step": 3111
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 0.44471701979637146,
+      "learning_rate": 0.0002263166545766797,
+      "loss": 0.0501,
+      "step": 3112
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 0.44834375381469727,
+      "learning_rate": 0.00022613759991192494,
+      "loss": 0.0557,
+      "step": 3113
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 0.4309970736503601,
+      "learning_rate": 0.00022595857325259988,
+      "loss": 0.0422,
+      "step": 3114
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 0.3993515074253082,
+      "learning_rate": 0.00022577957466658345,
+      "loss": 0.0485,
+      "step": 3115
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 0.2582777142524719,
+      "learning_rate": 0.00022560060422174432,
+      "loss": 0.0278,
+      "step": 3116
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 0.5203733444213867,
+      "learning_rate": 0.00022542166198594033,
+      "loss": 0.0641,
+      "step": 3117
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 0.2046922892332077,
+      "learning_rate": 0.00022524274802701862,
+      "loss": 0.0206,
+      "step": 3118
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 0.2442154586315155,
+      "learning_rate": 0.00022506386241281562,
+      "loss": 0.0211,
+      "step": 3119
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 0.3358626067638397,
+      "learning_rate": 0.00022488500521115713,
+      "loss": 0.0272,
+      "step": 3120
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 0.336093544960022,
+      "learning_rate": 0.00022470617648985786,
+      "loss": 0.0311,
+      "step": 3121
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 0.4430490732192993,
+      "learning_rate": 0.00022452737631672206,
+      "loss": 0.051,
+      "step": 3122
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 0.35754460096359253,
+      "learning_rate": 0.00022434860475954314,
+      "loss": 0.017,
+      "step": 3123
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 0.4587760269641876,
+      "learning_rate": 0.0002241698618861033,
+      "loss": 0.0339,
+      "step": 3124
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 0.36623820662498474,
+      "learning_rate": 0.00022399114776417429,
+      "loss": 0.0219,
+      "step": 3125
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 0.444183349609375,
+      "learning_rate": 0.00022381246246151674,
+      "loss": 0.0375,
+      "step": 3126
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 0.46760424971580505,
+      "learning_rate": 0.0002236338060458804,
+      "loss": 0.0455,
+      "step": 3127
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 0.41369733214378357,
+      "learning_rate": 0.00022345517858500397,
+      "loss": 0.0392,
+      "step": 3128
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 0.35412994027137756,
+      "learning_rate": 0.0002232765801466155,
+      "loss": 0.0381,
+      "step": 3129
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 0.3343680202960968,
+      "learning_rate": 0.00022309801079843146,
+      "loss": 0.0365,
+      "step": 3130
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 0.38387587666511536,
+      "learning_rate": 0.0002229194706081578,
+      "loss": 0.0463,
+      "step": 3131
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 0.3985680043697357,
+      "learning_rate": 0.00022274095964348943,
+      "loss": 0.0386,
+      "step": 3132
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 0.488238126039505,
+      "learning_rate": 0.00022256247797210963,
+      "loss": 0.0751,
+      "step": 3133
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 0.31015461683273315,
+      "learning_rate": 0.00022238402566169115,
+      "loss": 0.0268,
+      "step": 3134
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 0.4037419557571411,
+      "learning_rate": 0.00022220560277989545,
+      "loss": 0.0534,
+      "step": 3135
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 0.29287636280059814,
+      "learning_rate": 0.00022202720939437258,
+      "loss": 0.0285,
+      "step": 3136
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 0.32581421732902527,
+      "learning_rate": 0.00022184884557276173,
+      "loss": 0.0278,
+      "step": 3137
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 0.40073201060295105,
+      "learning_rate": 0.00022167051138269076,
+      "loss": 0.0416,
+      "step": 3138
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 0.2873273193836212,
+      "learning_rate": 0.00022149220689177625,
+      "loss": 0.0186,
+      "step": 3139
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 0.3280405104160309,
+      "learning_rate": 0.00022131393216762355,
+      "loss": 0.0428,
+      "step": 3140
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 0.33310702443122864,
+      "learning_rate": 0.00022113568727782676,
+      "loss": 0.0297,
+      "step": 3141
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 0.40162450075149536,
+      "learning_rate": 0.00022095747228996853,
+      "loss": 0.031,
+      "step": 3142
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 0.34024202823638916,
+      "learning_rate": 0.0002207792872716205,
+      "loss": 0.0279,
+      "step": 3143
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 0.4832566976547241,
+      "learning_rate": 0.00022060113229034248,
+      "loss": 0.0394,
+      "step": 3144
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 0.29034557938575745,
+      "learning_rate": 0.00022042300741368325,
+      "loss": 0.0232,
+      "step": 3145
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 0.2678511440753937,
+      "learning_rate": 0.00022024491270918015,
+      "loss": 0.0179,
+      "step": 3146
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 0.4466453194618225,
+      "learning_rate": 0.00022006684824435883,
+      "loss": 0.054,
+      "step": 3147
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 0.398834228515625,
+      "learning_rate": 0.00021988881408673377,
+      "loss": 0.0342,
+      "step": 3148
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 0.3920031487941742,
+      "learning_rate": 0.00021971081030380774,
+      "loss": 0.0285,
+      "step": 3149
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 0.3920028805732727,
+      "learning_rate": 0.00021953283696307218,
+      "loss": 0.0367,
+      "step": 3150
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 0.5309990048408508,
+      "learning_rate": 0.00021935489413200674,
+      "loss": 0.0551,
+      "step": 3151
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 0.3222086429595947,
+      "learning_rate": 0.00021917698187807984,
+      "loss": 0.0275,
+      "step": 3152
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 0.3224433958530426,
+      "learning_rate": 0.00021899910026874786,
+      "loss": 0.0219,
+      "step": 3153
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 0.4356839656829834,
+      "learning_rate": 0.00021882124937145606,
+      "loss": 0.039,
+      "step": 3154
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 0.45426180958747864,
+      "learning_rate": 0.00021864342925363782,
+      "loss": 0.0525,
+      "step": 3155
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 0.37439417839050293,
+      "learning_rate": 0.00021846563998271454,
+      "loss": 0.0385,
+      "step": 3156
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 0.26513323187828064,
+      "learning_rate": 0.00021828788162609651,
+      "loss": 0.0192,
+      "step": 3157
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 0.4129224717617035,
+      "learning_rate": 0.00021811015425118207,
+      "loss": 0.0335,
+      "step": 3158
+    },
+    {
+      "epoch": 6.01,
+      "grad_norm": 0.21821348369121552,
+      "learning_rate": 0.00021793245792535746,
+      "loss": 0.0168,
+      "step": 3159
+    },
+    {
+      "epoch": 6.01,
+      "grad_norm": 0.3457200527191162,
+      "learning_rate": 0.0002177547927159977,
+      "loss": 0.0335,
+      "step": 3160
+    },
+    {
+      "epoch": 6.01,
+      "grad_norm": 0.3566179573535919,
+      "learning_rate": 0.00021757715869046568,
+      "loss": 0.025,
+      "step": 3161
+    },
+    {
+      "epoch": 6.01,
+      "grad_norm": 0.33264005184173584,
+      "learning_rate": 0.00021739955591611255,
+      "loss": 0.0376,
+      "step": 3162
+    },
+    {
+      "epoch": 6.01,
+      "grad_norm": 0.3586775064468384,
+      "learning_rate": 0.00021722198446027754,
+      "loss": 0.0223,
+      "step": 3163
+    },
+    {
+      "epoch": 6.02,
+      "grad_norm": 0.2978927195072174,
+      "learning_rate": 0.0002170444443902882,
+      "loss": 0.0165,
+      "step": 3164
+    },
+    {
+      "epoch": 6.02,
+      "grad_norm": 0.3191410005092621,
+      "learning_rate": 0.00021686693577345983,
+      "loss": 0.0325,
+      "step": 3165
+    },
+    {
+      "epoch": 6.02,
+      "grad_norm": 0.2880886495113373,
+      "learning_rate": 0.00021668945867709633,
+      "loss": 0.0219,
+      "step": 3166
+    },
+    {
+      "epoch": 6.02,
+      "grad_norm": 0.23689985275268555,
+      "learning_rate": 0.00021651201316848906,
+      "loss": 0.0219,
+      "step": 3167
+    },
+    {
+      "epoch": 6.02,
+      "grad_norm": 0.3292834162712097,
+      "learning_rate": 0.00021633459931491778,
+      "loss": 0.0303,
+      "step": 3168
+    },
+    {
+      "epoch": 6.02,
+      "grad_norm": 0.29908743500709534,
+      "learning_rate": 0.00021615721718365037,
+      "loss": 0.0279,
+      "step": 3169
+    },
+    {
+      "epoch": 6.03,
+      "grad_norm": 0.11661837249994278,
+      "learning_rate": 0.00021597986684194213,
+      "loss": 0.007,
+      "step": 3170
+    },
+    {
+      "epoch": 6.03,
+      "grad_norm": 0.3574976921081543,
+      "learning_rate": 0.00021580254835703685,
+      "loss": 0.036,
+      "step": 3171
+    },
+    {
+      "epoch": 6.03,
+      "grad_norm": 0.30783095955848694,
+      "learning_rate": 0.000215625261796166,
+      "loss": 0.0347,
+      "step": 3172
+    },
+    {
+      "epoch": 6.03,
+      "grad_norm": 0.2761857211589813,
+      "learning_rate": 0.00021544800722654892,
+      "loss": 0.0188,
+      "step": 3173
+    },
+    {
+      "epoch": 6.03,
+      "grad_norm": 0.37551361322402954,
+      "learning_rate": 0.0002152707847153929,
+      "loss": 0.0256,
+      "step": 3174
+    },
+    {
+      "epoch": 6.04,
+      "grad_norm": 0.31003713607788086,
+      "learning_rate": 0.00021509359432989318,
+      "loss": 0.0286,
+      "step": 3175
+    },
+    {
+      "epoch": 6.04,
+      "grad_norm": 0.16120418906211853,
+      "learning_rate": 0.00021491643613723245,
+      "loss": 0.0085,
+      "step": 3176
+    },
+    {
+      "epoch": 6.04,
+      "grad_norm": 0.4369668662548065,
+      "learning_rate": 0.00021473931020458167,
+      "loss": 0.0157,
+      "step": 3177
+    },
+    {
+      "epoch": 6.04,
+      "grad_norm": 0.24336294829845428,
+      "learning_rate": 0.0002145622165990993,
+      "loss": 0.0176,
+      "step": 3178
+    },
+    {
+      "epoch": 6.04,
+      "grad_norm": 0.3335014581680298,
+      "learning_rate": 0.00021438515538793143,
+      "loss": 0.0236,
+      "step": 3179
+    },
+    {
+      "epoch": 6.05,
+      "grad_norm": 0.2098071426153183,
+      "learning_rate": 0.00021420812663821214,
+      "loss": 0.0179,
+      "step": 3180
+    },
+    {
+      "epoch": 6.05,
+      "grad_norm": 0.26212427020072937,
+      "learning_rate": 0.00021403113041706313,
+      "loss": 0.0151,
+      "step": 3181
+    },
+    {
+      "epoch": 6.05,
+      "grad_norm": 0.26528051495552063,
+      "learning_rate": 0.00021385416679159361,
+      "loss": 0.0182,
+      "step": 3182
+    },
+    {
+      "epoch": 6.05,
+      "grad_norm": 0.2797660827636719,
+      "learning_rate": 0.00021367723582890058,
+      "loss": 0.0186,
+      "step": 3183
+    },
+    {
+      "epoch": 6.05,
+      "grad_norm": 0.23963050544261932,
+      "learning_rate": 0.00021350033759606863,
+      "loss": 0.0128,
+      "step": 3184
+    },
+    {
+      "epoch": 6.06,
+      "grad_norm": 0.2969934940338135,
+      "learning_rate": 0.00021332347216016992,
+      "loss": 0.0294,
+      "step": 3185
+    },
+    {
+      "epoch": 6.06,
+      "grad_norm": 0.25928518176078796,
+      "learning_rate": 0.00021314663958826407,
+      "loss": 0.0226,
+      "step": 3186
+    },
+    {
+      "epoch": 6.06,
+      "grad_norm": 0.26479318737983704,
+      "learning_rate": 0.00021296983994739858,
+      "loss": 0.0175,
+      "step": 3187
+    },
+    {
+      "epoch": 6.06,
+      "grad_norm": 0.40802305936813354,
+      "learning_rate": 0.00021279307330460793,
+      "loss": 0.0312,
+      "step": 3188
+    },
+    {
+      "epoch": 6.06,
+      "grad_norm": 0.22416900098323822,
+      "learning_rate": 0.00021261633972691473,
+      "loss": 0.0132,
+      "step": 3189
+    },
+    {
+      "epoch": 6.06,
+      "grad_norm": 0.2352963089942932,
+      "learning_rate": 0.0002124396392813284,
+      "loss": 0.0124,
+      "step": 3190
+    },
+    {
+      "epoch": 6.07,
+      "grad_norm": 0.36080724000930786,
+      "learning_rate": 0.0002122629720348462,
+      "loss": 0.018,
+      "step": 3191
+    },
+    {
+      "epoch": 6.07,
+      "grad_norm": 0.24065344035625458,
+      "learning_rate": 0.00021208633805445295,
+      "loss": 0.0196,
+      "step": 3192
+    },
+    {
+      "epoch": 6.07,
+      "grad_norm": 0.3321077525615692,
+      "learning_rate": 0.00021190973740712032,
+      "loss": 0.029,
+      "step": 3193
+    },
+    {
+      "epoch": 6.07,
+      "grad_norm": 0.252310186624527,
+      "learning_rate": 0.00021173317015980777,
+      "loss": 0.017,
+      "step": 3194
+    },
+    {
+      "epoch": 6.07,
+      "grad_norm": 0.17539606988430023,
+      "learning_rate": 0.00021155663637946196,
+      "loss": 0.0158,
+      "step": 3195
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 0.3417055606842041,
+      "learning_rate": 0.00021138013613301689,
+      "loss": 0.0255,
+      "step": 3196
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 0.22867339849472046,
+      "learning_rate": 0.00021120366948739372,
+      "loss": 0.0177,
+      "step": 3197
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 0.3337012529373169,
+      "learning_rate": 0.00021102723650950116,
+      "loss": 0.029,
+      "step": 3198
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 0.34088608622550964,
+      "learning_rate": 0.00021085083726623475,
+      "loss": 0.0208,
+      "step": 3199
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 0.3615238070487976,
+      "learning_rate": 0.00021067447182447754,
+      "loss": 0.0162,
+      "step": 3200
+    },
+    {
+      "epoch": 6.08,
+      "eval_blimp_filtered_avg": 0.5467164179104478,
+      "eval_blimp_filtered_std": 0.0057993625385546034,
+      "step": 3200
+    },
+    {
+      "epoch": 6.08,
+      "eval_blimp_supplement_avg": 0.5107758620689655,
+      "eval_blimp_supplement_std": 0.02311658551762217,
+      "step": 3200
+    },
+    {
+      "epoch": 6.08,
+      "eval_vqa_filtered_avg": 0.15,
+      "eval_vqa_filtered_std": 0.035887028128263734,
+      "step": 3200
+    },
+    {
+      "epoch": 6.08,
+      "eval_winoground_filtered_avg": 0.51,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 3200
+    },
+    {
+      "epoch": 6.09,
+      "grad_norm": 0.31181517243385315,
+      "learning_rate": 0.0002104981402510999,
+      "loss": 0.0239,
+      "step": 3201
+    },
+    {
+      "epoch": 6.09,
+      "grad_norm": 0.2908709943294525,
+      "learning_rate": 0.00021032184261295878,
+      "loss": 0.0134,
+      "step": 3202
+    },
+    {
+      "epoch": 6.09,
+      "grad_norm": 0.3208898603916168,
+      "learning_rate": 0.0002101455789768988,
+      "loss": 0.0261,
+      "step": 3203
+    },
+    {
+      "epoch": 6.09,
+      "grad_norm": 0.2830604314804077,
+      "learning_rate": 0.0002099693494097517,
+      "loss": 0.0147,
+      "step": 3204
+    },
+    {
+      "epoch": 6.09,
+      "grad_norm": 0.25050169229507446,
+      "learning_rate": 0.00020979315397833573,
+      "loss": 0.012,
+      "step": 3205
+    },
+    {
+      "epoch": 6.1,
+      "grad_norm": 0.2651774287223816,
+      "learning_rate": 0.00020961699274945687,
+      "loss": 0.0141,
+      "step": 3206
+    },
+    {
+      "epoch": 6.1,
+      "grad_norm": 0.295495867729187,
+      "learning_rate": 0.00020944086578990773,
+      "loss": 0.0196,
+      "step": 3207
+    },
+    {
+      "epoch": 6.1,
+      "grad_norm": 0.22694352269172668,
+      "learning_rate": 0.00020926477316646807,
+      "loss": 0.0139,
+      "step": 3208
+    },
+    {
+      "epoch": 6.1,
+      "grad_norm": 0.2760408818721771,
+      "learning_rate": 0.00020908871494590452,
+      "loss": 0.0201,
+      "step": 3209
+    },
+    {
+      "epoch": 6.1,
+      "grad_norm": 0.3508376479148865,
+      "learning_rate": 0.00020891269119497095,
+      "loss": 0.0288,
+      "step": 3210
+    },
+    {
+      "epoch": 6.1,
+      "grad_norm": 0.2661544978618622,
+      "learning_rate": 0.00020873670198040772,
+      "loss": 0.0189,
+      "step": 3211
+    },
+    {
+      "epoch": 6.11,
+      "grad_norm": 0.20288847386837006,
+      "learning_rate": 0.00020856074736894255,
+      "loss": 0.0176,
+      "step": 3212
+    },
+    {
+      "epoch": 6.11,
+      "grad_norm": 0.242544025182724,
+      "learning_rate": 0.0002083848274272896,
+      "loss": 0.0147,
+      "step": 3213
+    },
+    {
+      "epoch": 6.11,
+      "grad_norm": 0.19805963337421417,
+      "learning_rate": 0.00020820894222215022,
+      "loss": 0.0104,
+      "step": 3214
+    },
+    {
+      "epoch": 6.11,
+      "grad_norm": 0.2465982735157013,
+      "learning_rate": 0.0002080330918202126,
+      "loss": 0.0227,
+      "step": 3215
+    },
+    {
+      "epoch": 6.11,
+      "grad_norm": 0.29228171706199646,
+      "learning_rate": 0.0002078572762881514,
+      "loss": 0.0185,
+      "step": 3216
+    },
+    {
+      "epoch": 6.12,
+      "grad_norm": 0.26943814754486084,
+      "learning_rate": 0.00020768149569262836,
+      "loss": 0.0273,
+      "step": 3217
+    },
+    {
+      "epoch": 6.12,
+      "grad_norm": 0.20903509855270386,
+      "learning_rate": 0.00020750575010029196,
+      "loss": 0.012,
+      "step": 3218
+    },
+    {
+      "epoch": 6.12,
+      "grad_norm": 0.28585246205329895,
+      "learning_rate": 0.00020733003957777718,
+      "loss": 0.0152,
+      "step": 3219
+    },
+    {
+      "epoch": 6.12,
+      "grad_norm": 0.17517472803592682,
+      "learning_rate": 0.00020715436419170593,
+      "loss": 0.0116,
+      "step": 3220
+    },
+    {
+      "epoch": 6.12,
+      "grad_norm": 0.27006614208221436,
+      "learning_rate": 0.00020697872400868682,
+      "loss": 0.0169,
+      "step": 3221
+    },
+    {
+      "epoch": 6.13,
+      "grad_norm": 0.2327791452407837,
+      "learning_rate": 0.00020680311909531486,
+      "loss": 0.0131,
+      "step": 3222
+    },
+    {
+      "epoch": 6.13,
+      "grad_norm": 0.27274614572525024,
+      "learning_rate": 0.00020662754951817185,
+      "loss": 0.019,
+      "step": 3223
+    },
+    {
+      "epoch": 6.13,
+      "grad_norm": 0.1894703060388565,
+      "learning_rate": 0.00020645201534382637,
+      "loss": 0.0108,
+      "step": 3224
+    },
+    {
+      "epoch": 6.13,
+      "grad_norm": 0.2422274351119995,
+      "learning_rate": 0.00020627651663883314,
+      "loss": 0.0162,
+      "step": 3225
+    },
+    {
+      "epoch": 6.13,
+      "grad_norm": 0.3082599639892578,
+      "learning_rate": 0.00020610105346973383,
+      "loss": 0.0165,
+      "step": 3226
+    },
+    {
+      "epoch": 6.13,
+      "grad_norm": 0.18682537972927094,
+      "learning_rate": 0.00020592562590305657,
+      "loss": 0.0079,
+      "step": 3227
+    },
+    {
+      "epoch": 6.14,
+      "grad_norm": 0.23307573795318604,
+      "learning_rate": 0.0002057502340053157,
+      "loss": 0.0095,
+      "step": 3228
+    },
+    {
+      "epoch": 6.14,
+      "grad_norm": 0.21828116476535797,
+      "learning_rate": 0.00020557487784301243,
+      "loss": 0.0122,
+      "step": 3229
+    },
+    {
+      "epoch": 6.14,
+      "grad_norm": 0.24922336637973785,
+      "learning_rate": 0.00020539955748263413,
+      "loss": 0.0082,
+      "step": 3230
+    },
+    {
+      "epoch": 6.14,
+      "grad_norm": 0.24670149385929108,
+      "learning_rate": 0.00020522427299065474,
+      "loss": 0.0135,
+      "step": 3231
+    },
+    {
+      "epoch": 6.14,
+      "grad_norm": 0.21233424544334412,
+      "learning_rate": 0.0002050490244335345,
+      "loss": 0.0169,
+      "step": 3232
+    },
+    {
+      "epoch": 6.15,
+      "grad_norm": 0.25627923011779785,
+      "learning_rate": 0.00020487381187772027,
+      "loss": 0.0156,
+      "step": 3233
+    },
+    {
+      "epoch": 6.15,
+      "grad_norm": 0.19155535101890564,
+      "learning_rate": 0.00020469863538964483,
+      "loss": 0.0078,
+      "step": 3234
+    },
+    {
+      "epoch": 6.15,
+      "grad_norm": 0.10986325889825821,
+      "learning_rate": 0.00020452349503572763,
+      "loss": 0.0036,
+      "step": 3235
+    },
+    {
+      "epoch": 6.15,
+      "grad_norm": 0.27545252442359924,
+      "learning_rate": 0.0002043483908823745,
+      "loss": 0.0148,
+      "step": 3236
+    },
+    {
+      "epoch": 6.15,
+      "grad_norm": 0.08412336558103561,
+      "learning_rate": 0.0002041733229959771,
+      "loss": 0.0031,
+      "step": 3237
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 0.28829139471054077,
+      "learning_rate": 0.00020399829144291382,
+      "loss": 0.0167,
+      "step": 3238
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 0.24435383081436157,
+      "learning_rate": 0.00020382329628954886,
+      "loss": 0.02,
+      "step": 3239
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 0.3500301241874695,
+      "learning_rate": 0.00020364833760223297,
+      "loss": 0.0261,
+      "step": 3240
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 0.33243489265441895,
+      "learning_rate": 0.00020347341544730286,
+      "loss": 0.0288,
+      "step": 3241
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 0.3211993873119354,
+      "learning_rate": 0.00020329852989108148,
+      "loss": 0.0276,
+      "step": 3242
+    },
+    {
+      "epoch": 6.17,
+      "grad_norm": 0.3398200273513794,
+      "learning_rate": 0.00020312368099987776,
+      "loss": 0.0241,
+      "step": 3243
+    },
+    {
+      "epoch": 6.17,
+      "grad_norm": 0.23313555121421814,
+      "learning_rate": 0.0002029488688399871,
+      "loss": 0.0101,
+      "step": 3244
+    },
+    {
+      "epoch": 6.17,
+      "grad_norm": 0.3783957064151764,
+      "learning_rate": 0.00020277409347769043,
+      "loss": 0.0175,
+      "step": 3245
+    },
+    {
+      "epoch": 6.17,
+      "grad_norm": 0.2785942256450653,
+      "learning_rate": 0.0002025993549792551,
+      "loss": 0.0185,
+      "step": 3246
+    },
+    {
+      "epoch": 6.17,
+      "grad_norm": 0.29258930683135986,
+      "learning_rate": 0.00020242465341093458,
+      "loss": 0.0189,
+      "step": 3247
+    },
+    {
+      "epoch": 6.17,
+      "grad_norm": 0.15812307596206665,
+      "learning_rate": 0.00020224998883896787,
+      "loss": 0.0133,
+      "step": 3248
+    },
+    {
+      "epoch": 6.18,
+      "grad_norm": 0.23220491409301758,
+      "learning_rate": 0.00020207536132958034,
+      "loss": 0.016,
+      "step": 3249
+    },
+    {
+      "epoch": 6.18,
+      "grad_norm": 0.25162598490715027,
+      "learning_rate": 0.0002019007709489834,
+      "loss": 0.0126,
+      "step": 3250
+    },
+    {
+      "epoch": 6.18,
+      "grad_norm": 0.3259192109107971,
+      "learning_rate": 0.0002017262177633738,
+      "loss": 0.0199,
+      "step": 3251
+    },
+    {
+      "epoch": 6.18,
+      "grad_norm": 0.3458859920501709,
+      "learning_rate": 0.00020155170183893485,
+      "loss": 0.0259,
+      "step": 3252
+    },
+    {
+      "epoch": 6.18,
+      "grad_norm": 0.2345294952392578,
+      "learning_rate": 0.00020137722324183532,
+      "loss": 0.0107,
+      "step": 3253
+    },
+    {
+      "epoch": 6.19,
+      "grad_norm": 0.307201087474823,
+      "learning_rate": 0.00020120278203822998,
+      "loss": 0.0188,
+      "step": 3254
+    },
+    {
+      "epoch": 6.19,
+      "grad_norm": 0.22582122683525085,
+      "learning_rate": 0.0002010283782942594,
+      "loss": 0.0097,
+      "step": 3255
+    },
+    {
+      "epoch": 6.19,
+      "grad_norm": 0.21507947146892548,
+      "learning_rate": 0.00020085401207605002,
+      "loss": 0.0143,
+      "step": 3256
+    },
+    {
+      "epoch": 6.19,
+      "grad_norm": 0.27300578355789185,
+      "learning_rate": 0.00020067968344971376,
+      "loss": 0.0215,
+      "step": 3257
+    },
+    {
+      "epoch": 6.19,
+      "grad_norm": 0.3466844856739044,
+      "learning_rate": 0.00020050539248134863,
+      "loss": 0.0301,
+      "step": 3258
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 0.20451071858406067,
+      "learning_rate": 0.00020033113923703837,
+      "loss": 0.0093,
+      "step": 3259
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 0.2108183205127716,
+      "learning_rate": 0.00020015692378285205,
+      "loss": 0.0094,
+      "step": 3260
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 0.3538322150707245,
+      "learning_rate": 0.0001999827461848448,
+      "loss": 0.0149,
+      "step": 3261
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 0.17654883861541748,
+      "learning_rate": 0.00019980860650905706,
+      "loss": 0.0097,
+      "step": 3262
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 0.24415694177150726,
+      "learning_rate": 0.00019963450482151526,
+      "loss": 0.0225,
+      "step": 3263
+    },
+    {
+      "epoch": 6.21,
+      "grad_norm": 0.26709288358688354,
+      "learning_rate": 0.00019946044118823112,
+      "loss": 0.0151,
+      "step": 3264
+    },
+    {
+      "epoch": 6.21,
+      "grad_norm": 0.22450226545333862,
+      "learning_rate": 0.0001992864156752021,
+      "loss": 0.0139,
+      "step": 3265
+    },
+    {
+      "epoch": 6.21,
+      "grad_norm": 0.2667621374130249,
+      "learning_rate": 0.00019911242834841106,
+      "loss": 0.0174,
+      "step": 3266
+    },
+    {
+      "epoch": 6.21,
+      "grad_norm": 0.2584960162639618,
+      "learning_rate": 0.00019893847927382667,
+      "loss": 0.0154,
+      "step": 3267
+    },
+    {
+      "epoch": 6.21,
+      "grad_norm": 0.2402184009552002,
+      "learning_rate": 0.00019876456851740265,
+      "loss": 0.0188,
+      "step": 3268
+    },
+    {
+      "epoch": 6.21,
+      "grad_norm": 0.2993856370449066,
+      "learning_rate": 0.00019859069614507855,
+      "loss": 0.021,
+      "step": 3269
+    },
+    {
+      "epoch": 6.22,
+      "grad_norm": 0.12748394906520844,
+      "learning_rate": 0.00019841686222277943,
+      "loss": 0.0094,
+      "step": 3270
+    },
+    {
+      "epoch": 6.22,
+      "grad_norm": 0.2048962414264679,
+      "learning_rate": 0.0001982430668164153,
+      "loss": 0.0085,
+      "step": 3271
+    },
+    {
+      "epoch": 6.22,
+      "grad_norm": 0.24013325572013855,
+      "learning_rate": 0.00019806930999188203,
+      "loss": 0.0162,
+      "step": 3272
+    },
+    {
+      "epoch": 6.22,
+      "grad_norm": 0.24516181647777557,
+      "learning_rate": 0.00019789559181506076,
+      "loss": 0.0232,
+      "step": 3273
+    },
+    {
+      "epoch": 6.22,
+      "grad_norm": 0.31989994645118713,
+      "learning_rate": 0.00019772191235181775,
+      "loss": 0.0199,
+      "step": 3274
+    },
+    {
+      "epoch": 6.23,
+      "grad_norm": 0.2762034833431244,
+      "learning_rate": 0.00019754827166800488,
+      "loss": 0.0138,
+      "step": 3275
+    },
+    {
+      "epoch": 6.23,
+      "grad_norm": 0.22172538936138153,
+      "learning_rate": 0.00019737466982945915,
+      "loss": 0.0154,
+      "step": 3276
+    },
+    {
+      "epoch": 6.23,
+      "grad_norm": 0.24809440970420837,
+      "learning_rate": 0.00019720110690200286,
+      "loss": 0.0105,
+      "step": 3277
+    },
+    {
+      "epoch": 6.23,
+      "grad_norm": 0.256168007850647,
+      "learning_rate": 0.0001970275829514435,
+      "loss": 0.014,
+      "step": 3278
+    },
+    {
+      "epoch": 6.23,
+      "grad_norm": 0.18292225897312164,
+      "learning_rate": 0.00019685409804357407,
+      "loss": 0.0085,
+      "step": 3279
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 0.19679181277751923,
+      "learning_rate": 0.00019668065224417227,
+      "loss": 0.0098,
+      "step": 3280
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 0.18697421252727509,
+      "learning_rate": 0.00019650724561900138,
+      "loss": 0.0099,
+      "step": 3281
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 0.2603377103805542,
+      "learning_rate": 0.00019633387823380985,
+      "loss": 0.0279,
+      "step": 3282
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 0.20866461098194122,
+      "learning_rate": 0.00019616055015433078,
+      "loss": 0.009,
+      "step": 3283
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 0.26505246758461,
+      "learning_rate": 0.00019598726144628292,
+      "loss": 0.0196,
+      "step": 3284
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 0.3112882673740387,
+      "learning_rate": 0.00019581401217536986,
+      "loss": 0.0232,
+      "step": 3285
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 0.2654329538345337,
+      "learning_rate": 0.00019564080240728015,
+      "loss": 0.0188,
+      "step": 3286
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 0.21784302592277527,
+      "learning_rate": 0.00019546763220768746,
+      "loss": 0.0144,
+      "step": 3287
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 0.17950555682182312,
+      "learning_rate": 0.00019529450164225043,
+      "loss": 0.0144,
+      "step": 3288
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 0.2736026346683502,
+      "learning_rate": 0.00019512141077661268,
+      "loss": 0.0137,
+      "step": 3289
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 0.26154619455337524,
+      "learning_rate": 0.0001949483596764029,
+      "loss": 0.0256,
+      "step": 3290
+    },
+    {
+      "epoch": 6.26,
+      "grad_norm": 0.1946219950914383,
+      "learning_rate": 0.0001947753484072344,
+      "loss": 0.0116,
+      "step": 3291
+    },
+    {
+      "epoch": 6.26,
+      "grad_norm": 0.3421551585197449,
+      "learning_rate": 0.00019460237703470567,
+      "loss": 0.0311,
+      "step": 3292
+    },
+    {
+      "epoch": 6.26,
+      "grad_norm": 0.22935263812541962,
+      "learning_rate": 0.00019442944562440012,
+      "loss": 0.0189,
+      "step": 3293
+    },
+    {
+      "epoch": 6.26,
+      "grad_norm": 0.36268025636672974,
+      "learning_rate": 0.00019425655424188559,
+      "loss": 0.0303,
+      "step": 3294
+    },
+    {
+      "epoch": 6.26,
+      "grad_norm": 0.3655558228492737,
+      "learning_rate": 0.0001940837029527152,
+      "loss": 0.0162,
+      "step": 3295
+    },
+    {
+      "epoch": 6.27,
+      "grad_norm": 0.2442244440317154,
+      "learning_rate": 0.00019391089182242674,
+      "loss": 0.0137,
+      "step": 3296
+    },
+    {
+      "epoch": 6.27,
+      "grad_norm": 0.12407729774713516,
+      "learning_rate": 0.00019373812091654253,
+      "loss": 0.0063,
+      "step": 3297
+    },
+    {
+      "epoch": 6.27,
+      "grad_norm": 0.32682886719703674,
+      "learning_rate": 0.00019356539030056995,
+      "loss": 0.0299,
+      "step": 3298
+    },
+    {
+      "epoch": 6.27,
+      "grad_norm": 0.29414066672325134,
+      "learning_rate": 0.000193392700040001,
+      "loss": 0.0401,
+      "step": 3299
+    },
+    {
+      "epoch": 6.27,
+      "grad_norm": 0.29304441809654236,
+      "learning_rate": 0.00019322005020031233,
+      "loss": 0.0176,
+      "step": 3300
+    },
+    {
+      "epoch": 6.27,
+      "eval_blimp_filtered_avg": 0.5452238805970149,
+      "eval_blimp_filtered_std": 0.00586637150862264,
+      "step": 3300
+    },
+    {
+      "epoch": 6.27,
+      "eval_blimp_supplement_avg": 0.521551724137931,
+      "eval_blimp_supplement_std": 0.023208313835338362,
+      "step": 3300
+    },
+    {
+      "epoch": 6.27,
+      "eval_vqa_filtered_avg": 0.13,
+      "eval_vqa_filtered_std": 0.033799766898963086,
+      "step": 3300
+    },
+    {
+      "epoch": 6.27,
+      "eval_winoground_filtered_avg": 0.48,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 3300
+    },
+    {
+      "epoch": 6.28,
+      "grad_norm": 0.31623807549476624,
+      "learning_rate": 0.00019304744084696523,
+      "loss": 0.0237,
+      "step": 3301
+    },
+    {
+      "epoch": 6.28,
+      "grad_norm": 0.23027898371219635,
+      "learning_rate": 0.00019287487204540588,
+      "loss": 0.0093,
+      "step": 3302
+    },
+    {
+      "epoch": 6.28,
+      "grad_norm": 0.2817038595676422,
+      "learning_rate": 0.00019270234386106465,
+      "loss": 0.0148,
+      "step": 3303
+    },
+    {
+      "epoch": 6.28,
+      "grad_norm": 0.24087181687355042,
+      "learning_rate": 0.00019252985635935697,
+      "loss": 0.0108,
+      "step": 3304
+    },
+    {
+      "epoch": 6.28,
+      "grad_norm": 0.3114948570728302,
+      "learning_rate": 0.00019235740960568267,
+      "loss": 0.0216,
+      "step": 3305
+    },
+    {
+      "epoch": 6.29,
+      "grad_norm": 0.32051920890808105,
+      "learning_rate": 0.00019218500366542592,
+      "loss": 0.0289,
+      "step": 3306
+    },
+    {
+      "epoch": 6.29,
+      "grad_norm": 0.29504767060279846,
+      "learning_rate": 0.00019201263860395574,
+      "loss": 0.03,
+      "step": 3307
+    },
+    {
+      "epoch": 6.29,
+      "grad_norm": 0.23371849954128265,
+      "learning_rate": 0.0001918403144866256,
+      "loss": 0.0124,
+      "step": 3308
+    },
+    {
+      "epoch": 6.29,
+      "grad_norm": 0.23987899720668793,
+      "learning_rate": 0.00019166803137877315,
+      "loss": 0.0133,
+      "step": 3309
+    },
+    {
+      "epoch": 6.29,
+      "grad_norm": 0.3597557842731476,
+      "learning_rate": 0.00019149578934572083,
+      "loss": 0.0209,
+      "step": 3310
+    },
+    {
+      "epoch": 6.29,
+      "grad_norm": 0.25805458426475525,
+      "learning_rate": 0.0001913235884527753,
+      "loss": 0.018,
+      "step": 3311
+    },
+    {
+      "epoch": 6.3,
+      "grad_norm": 0.27436548471450806,
+      "learning_rate": 0.00019115142876522767,
+      "loss": 0.0122,
+      "step": 3312
+    },
+    {
+      "epoch": 6.3,
+      "grad_norm": 0.24891388416290283,
+      "learning_rate": 0.00019097931034835368,
+      "loss": 0.0126,
+      "step": 3313
+    },
+    {
+      "epoch": 6.3,
+      "grad_norm": 0.28320997953414917,
+      "learning_rate": 0.00019080723326741285,
+      "loss": 0.0257,
+      "step": 3314
+    },
+    {
+      "epoch": 6.3,
+      "grad_norm": 0.2371446192264557,
+      "learning_rate": 0.0001906351975876495,
+      "loss": 0.0119,
+      "step": 3315
+    },
+    {
+      "epoch": 6.3,
+      "grad_norm": 0.2101546674966812,
+      "learning_rate": 0.00019046320337429238,
+      "loss": 0.0177,
+      "step": 3316
+    },
+    {
+      "epoch": 6.31,
+      "grad_norm": 0.1828141063451767,
+      "learning_rate": 0.00019029125069255389,
+      "loss": 0.0154,
+      "step": 3317
+    },
+    {
+      "epoch": 6.31,
+      "grad_norm": 0.1865801066160202,
+      "learning_rate": 0.00019011933960763123,
+      "loss": 0.0109,
+      "step": 3318
+    },
+    {
+      "epoch": 6.31,
+      "grad_norm": 0.13071514666080475,
+      "learning_rate": 0.00018994747018470572,
+      "loss": 0.01,
+      "step": 3319
+    },
+    {
+      "epoch": 6.31,
+      "grad_norm": 0.2534213066101074,
+      "learning_rate": 0.00018977564248894268,
+      "loss": 0.0202,
+      "step": 3320
+    },
+    {
+      "epoch": 6.31,
+      "grad_norm": 0.23443719744682312,
+      "learning_rate": 0.00018960385658549182,
+      "loss": 0.0161,
+      "step": 3321
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 0.2567359507083893,
+      "learning_rate": 0.0001894321125394869,
+      "loss": 0.0127,
+      "step": 3322
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 0.3225496709346771,
+      "learning_rate": 0.00018926041041604586,
+      "loss": 0.019,
+      "step": 3323
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 0.28724831342697144,
+      "learning_rate": 0.00018908875028027063,
+      "loss": 0.0151,
+      "step": 3324
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 0.19402861595153809,
+      "learning_rate": 0.00018891713219724757,
+      "loss": 0.011,
+      "step": 3325
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 0.22893324494361877,
+      "learning_rate": 0.0001887455562320465,
+      "loss": 0.0201,
+      "step": 3326
+    },
+    {
+      "epoch": 6.33,
+      "grad_norm": 0.24654699862003326,
+      "learning_rate": 0.0001885740224497218,
+      "loss": 0.007,
+      "step": 3327
+    },
+    {
+      "epoch": 6.33,
+      "grad_norm": 0.23787632584571838,
+      "learning_rate": 0.00018840253091531174,
+      "loss": 0.013,
+      "step": 3328
+    },
+    {
+      "epoch": 6.33,
+      "grad_norm": 0.23256146907806396,
+      "learning_rate": 0.00018823108169383835,
+      "loss": 0.0167,
+      "step": 3329
+    },
+    {
+      "epoch": 6.33,
+      "grad_norm": 0.23074665665626526,
+      "learning_rate": 0.0001880596748503077,
+      "loss": 0.0175,
+      "step": 3330
+    },
+    {
+      "epoch": 6.33,
+      "grad_norm": 0.26432111859321594,
+      "learning_rate": 0.00018788831044971017,
+      "loss": 0.0136,
+      "step": 3331
+    },
+    {
+      "epoch": 6.33,
+      "grad_norm": 0.18909527361392975,
+      "learning_rate": 0.00018771698855701944,
+      "loss": 0.0098,
+      "step": 3332
+    },
+    {
+      "epoch": 6.34,
+      "grad_norm": 0.23469676077365875,
+      "learning_rate": 0.00018754570923719342,
+      "loss": 0.0169,
+      "step": 3333
+    },
+    {
+      "epoch": 6.34,
+      "grad_norm": 0.2103978544473648,
+      "learning_rate": 0.00018737447255517397,
+      "loss": 0.0069,
+      "step": 3334
+    },
+    {
+      "epoch": 6.34,
+      "grad_norm": 0.2565735876560211,
+      "learning_rate": 0.00018720327857588642,
+      "loss": 0.0133,
+      "step": 3335
+    },
+    {
+      "epoch": 6.34,
+      "grad_norm": 0.21231335401535034,
+      "learning_rate": 0.00018703212736424036,
+      "loss": 0.0076,
+      "step": 3336
+    },
+    {
+      "epoch": 6.34,
+      "grad_norm": 0.2387968897819519,
+      "learning_rate": 0.0001868610189851287,
+      "loss": 0.0128,
+      "step": 3337
+    },
+    {
+      "epoch": 6.35,
+      "grad_norm": 0.19674140214920044,
+      "learning_rate": 0.00018668995350342845,
+      "loss": 0.0099,
+      "step": 3338
+    },
+    {
+      "epoch": 6.35,
+      "grad_norm": 0.34821784496307373,
+      "learning_rate": 0.00018651893098400036,
+      "loss": 0.0115,
+      "step": 3339
+    },
+    {
+      "epoch": 6.35,
+      "grad_norm": 0.29434630274772644,
+      "learning_rate": 0.0001863479514916886,
+      "loss": 0.0187,
+      "step": 3340
+    },
+    {
+      "epoch": 6.35,
+      "grad_norm": 0.20638546347618103,
+      "learning_rate": 0.0001861770150913212,
+      "loss": 0.0176,
+      "step": 3341
+    },
+    {
+      "epoch": 6.35,
+      "grad_norm": 0.34810492396354675,
+      "learning_rate": 0.00018600612184771007,
+      "loss": 0.0222,
+      "step": 3342
+    },
+    {
+      "epoch": 6.36,
+      "grad_norm": 0.19939550757408142,
+      "learning_rate": 0.00018583527182565028,
+      "loss": 0.0114,
+      "step": 3343
+    },
+    {
+      "epoch": 6.36,
+      "grad_norm": 0.28078603744506836,
+      "learning_rate": 0.000185664465089921,
+      "loss": 0.0129,
+      "step": 3344
+    },
+    {
+      "epoch": 6.36,
+      "grad_norm": 0.22865884006023407,
+      "learning_rate": 0.00018549370170528457,
+      "loss": 0.0151,
+      "step": 3345
+    },
+    {
+      "epoch": 6.36,
+      "grad_norm": 0.25231456756591797,
+      "learning_rate": 0.00018532298173648721,
+      "loss": 0.0218,
+      "step": 3346
+    },
+    {
+      "epoch": 6.36,
+      "grad_norm": 0.2952040135860443,
+      "learning_rate": 0.00018515230524825845,
+      "loss": 0.0123,
+      "step": 3347
+    },
+    {
+      "epoch": 6.37,
+      "grad_norm": 0.372024267911911,
+      "learning_rate": 0.00018498167230531162,
+      "loss": 0.0307,
+      "step": 3348
+    },
+    {
+      "epoch": 6.37,
+      "grad_norm": 0.1696850061416626,
+      "learning_rate": 0.0001848110829723431,
+      "loss": 0.0144,
+      "step": 3349
+    },
+    {
+      "epoch": 6.37,
+      "grad_norm": 0.19347849488258362,
+      "learning_rate": 0.00018464053731403316,
+      "loss": 0.0123,
+      "step": 3350
+    },
+    {
+      "epoch": 6.37,
+      "grad_norm": 0.22799034416675568,
+      "learning_rate": 0.0001844700353950454,
+      "loss": 0.0076,
+      "step": 3351
+    },
+    {
+      "epoch": 6.37,
+      "grad_norm": 0.2489493489265442,
+      "learning_rate": 0.0001842995772800266,
+      "loss": 0.0147,
+      "step": 3352
+    },
+    {
+      "epoch": 6.37,
+      "grad_norm": 0.2604806423187256,
+      "learning_rate": 0.00018412916303360717,
+      "loss": 0.0142,
+      "step": 3353
+    },
+    {
+      "epoch": 6.38,
+      "grad_norm": 0.33713600039482117,
+      "learning_rate": 0.00018395879272040095,
+      "loss": 0.0382,
+      "step": 3354
+    },
+    {
+      "epoch": 6.38,
+      "grad_norm": 0.2202487736940384,
+      "learning_rate": 0.00018378846640500486,
+      "loss": 0.0098,
+      "step": 3355
+    },
+    {
+      "epoch": 6.38,
+      "grad_norm": 0.3151213824748993,
+      "learning_rate": 0.0001836181841519993,
+      "loss": 0.0226,
+      "step": 3356
+    },
+    {
+      "epoch": 6.38,
+      "grad_norm": 0.19055786728858948,
+      "learning_rate": 0.00018344794602594803,
+      "loss": 0.0081,
+      "step": 3357
+    },
+    {
+      "epoch": 6.38,
+      "grad_norm": 0.17541643977165222,
+      "learning_rate": 0.0001832777520913978,
+      "loss": 0.0076,
+      "step": 3358
+    },
+    {
+      "epoch": 6.39,
+      "grad_norm": 0.18829026818275452,
+      "learning_rate": 0.00018310760241287892,
+      "loss": 0.0129,
+      "step": 3359
+    },
+    {
+      "epoch": 6.39,
+      "grad_norm": 0.27362874150276184,
+      "learning_rate": 0.0001829374970549049,
+      "loss": 0.0238,
+      "step": 3360
+    },
+    {
+      "epoch": 6.39,
+      "grad_norm": 0.17540180683135986,
+      "learning_rate": 0.00018276743608197215,
+      "loss": 0.0076,
+      "step": 3361
+    },
+    {
+      "epoch": 6.39,
+      "grad_norm": 0.2125861495733261,
+      "learning_rate": 0.0001825974195585606,
+      "loss": 0.0152,
+      "step": 3362
+    },
+    {
+      "epoch": 6.39,
+      "grad_norm": 0.23881226778030396,
+      "learning_rate": 0.00018242744754913293,
+      "loss": 0.0198,
+      "step": 3363
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 0.28282785415649414,
+      "learning_rate": 0.00018225752011813537,
+      "loss": 0.0131,
+      "step": 3364
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 0.227904811501503,
+      "learning_rate": 0.00018208763732999715,
+      "loss": 0.0161,
+      "step": 3365
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 0.2396094650030136,
+      "learning_rate": 0.0001819177992491302,
+      "loss": 0.0104,
+      "step": 3366
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 0.153622105717659,
+      "learning_rate": 0.00018174800593993002,
+      "loss": 0.0099,
+      "step": 3367
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 0.17526863515377045,
+      "learning_rate": 0.0001815782574667748,
+      "loss": 0.0117,
+      "step": 3368
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 0.3508951961994171,
+      "learning_rate": 0.0001814085538940258,
+      "loss": 0.0173,
+      "step": 3369
+    },
+    {
+      "epoch": 6.41,
+      "grad_norm": 0.16141445934772491,
+      "learning_rate": 0.00018123889528602724,
+      "loss": 0.0088,
+      "step": 3370
+    },
+    {
+      "epoch": 6.41,
+      "grad_norm": 0.2465951144695282,
+      "learning_rate": 0.00018106928170710652,
+      "loss": 0.0169,
+      "step": 3371
+    },
+    {
+      "epoch": 6.41,
+      "grad_norm": 0.28537940979003906,
+      "learning_rate": 0.00018089971322157356,
+      "loss": 0.0117,
+      "step": 3372
+    },
+    {
+      "epoch": 6.41,
+      "grad_norm": 0.2741108536720276,
+      "learning_rate": 0.00018073018989372147,
+      "loss": 0.0198,
+      "step": 3373
+    },
+    {
+      "epoch": 6.41,
+      "grad_norm": 0.1757652312517166,
+      "learning_rate": 0.00018056071178782637,
+      "loss": 0.0121,
+      "step": 3374
+    },
+    {
+      "epoch": 6.42,
+      "grad_norm": 0.2184172421693802,
+      "learning_rate": 0.00018039127896814667,
+      "loss": 0.0117,
+      "step": 3375
+    },
+    {
+      "epoch": 6.42,
+      "grad_norm": 0.14818522334098816,
+      "learning_rate": 0.00018022189149892418,
+      "loss": 0.006,
+      "step": 3376
+    },
+    {
+      "epoch": 6.42,
+      "grad_norm": 0.25795283913612366,
+      "learning_rate": 0.0001800525494443834,
+      "loss": 0.0174,
+      "step": 3377
+    },
+    {
+      "epoch": 6.42,
+      "grad_norm": 0.17726556956768036,
+      "learning_rate": 0.00017988325286873136,
+      "loss": 0.0094,
+      "step": 3378
+    },
+    {
+      "epoch": 6.42,
+      "grad_norm": 0.14405611157417297,
+      "learning_rate": 0.00017971400183615797,
+      "loss": 0.0063,
+      "step": 3379
+    },
+    {
+      "epoch": 6.43,
+      "grad_norm": 0.22868730127811432,
+      "learning_rate": 0.0001795447964108361,
+      "loss": 0.0079,
+      "step": 3380
+    },
+    {
+      "epoch": 6.43,
+      "grad_norm": 0.2788814902305603,
+      "learning_rate": 0.00017937563665692093,
+      "loss": 0.0237,
+      "step": 3381
+    },
+    {
+      "epoch": 6.43,
+      "grad_norm": 0.329814076423645,
+      "learning_rate": 0.00017920652263855058,
+      "loss": 0.0207,
+      "step": 3382
+    },
+    {
+      "epoch": 6.43,
+      "grad_norm": 0.2854446768760681,
+      "learning_rate": 0.000179037454419846,
+      "loss": 0.0126,
+      "step": 3383
+    },
+    {
+      "epoch": 6.43,
+      "grad_norm": 0.15086723864078522,
+      "learning_rate": 0.00017886843206491019,
+      "loss": 0.0067,
+      "step": 3384
+    },
+    {
+      "epoch": 6.44,
+      "grad_norm": 0.15878674387931824,
+      "learning_rate": 0.00017869945563782935,
+      "loss": 0.011,
+      "step": 3385
+    },
+    {
+      "epoch": 6.44,
+      "grad_norm": 0.2881024181842804,
+      "learning_rate": 0.00017853052520267192,
+      "loss": 0.0192,
+      "step": 3386
+    },
+    {
+      "epoch": 6.44,
+      "grad_norm": 0.2749207317829132,
+      "learning_rate": 0.00017836164082348906,
+      "loss": 0.0178,
+      "step": 3387
+    },
+    {
+      "epoch": 6.44,
+      "grad_norm": 0.19062606990337372,
+      "learning_rate": 0.00017819280256431462,
+      "loss": 0.0082,
+      "step": 3388
+    },
+    {
+      "epoch": 6.44,
+      "grad_norm": 0.2202426642179489,
+      "learning_rate": 0.00017802401048916446,
+      "loss": 0.0144,
+      "step": 3389
+    },
+    {
+      "epoch": 6.44,
+      "grad_norm": 0.24107229709625244,
+      "learning_rate": 0.0001778552646620374,
+      "loss": 0.0151,
+      "step": 3390
+    },
+    {
+      "epoch": 6.45,
+      "grad_norm": 0.22425568103790283,
+      "learning_rate": 0.00017768656514691457,
+      "loss": 0.0089,
+      "step": 3391
+    },
+    {
+      "epoch": 6.45,
+      "grad_norm": 0.25728046894073486,
+      "learning_rate": 0.0001775179120077595,
+      "loss": 0.0166,
+      "step": 3392
+    },
+    {
+      "epoch": 6.45,
+      "grad_norm": 0.23864100873470306,
+      "learning_rate": 0.0001773493053085181,
+      "loss": 0.0141,
+      "step": 3393
+    },
+    {
+      "epoch": 6.45,
+      "grad_norm": 0.42108678817749023,
+      "learning_rate": 0.00017718074511311892,
+      "loss": 0.0351,
+      "step": 3394
+    },
+    {
+      "epoch": 6.45,
+      "grad_norm": 0.2102043628692627,
+      "learning_rate": 0.00017701223148547252,
+      "loss": 0.0143,
+      "step": 3395
+    },
+    {
+      "epoch": 6.46,
+      "grad_norm": 0.18634220957756042,
+      "learning_rate": 0.00017684376448947204,
+      "loss": 0.0095,
+      "step": 3396
+    },
+    {
+      "epoch": 6.46,
+      "grad_norm": 0.3426073491573334,
+      "learning_rate": 0.000176675344188993,
+      "loss": 0.0245,
+      "step": 3397
+    },
+    {
+      "epoch": 6.46,
+      "grad_norm": 0.20593495666980743,
+      "learning_rate": 0.00017650697064789292,
+      "loss": 0.0128,
+      "step": 3398
+    },
+    {
+      "epoch": 6.46,
+      "grad_norm": 0.21779927611351013,
+      "learning_rate": 0.00017633864393001186,
+      "loss": 0.0196,
+      "step": 3399
+    },
+    {
+      "epoch": 6.46,
+      "grad_norm": 0.1486906260251999,
+      "learning_rate": 0.0001761703640991721,
+      "loss": 0.0041,
+      "step": 3400
+    },
+    {
+      "epoch": 6.46,
+      "eval_blimp_filtered_avg": 0.546865671641791,
+      "eval_blimp_filtered_std": 0.005853988277721553,
+      "step": 3400
+    },
+    {
+      "epoch": 6.46,
+      "eval_blimp_supplement_avg": 0.5495689655172413,
+      "eval_blimp_supplement_std": 0.023073891583442255,
+      "step": 3400
+    },
+    {
+      "epoch": 6.46,
+      "eval_vqa_filtered_avg": 0.13,
+      "eval_vqa_filtered_std": 0.03379976689896309,
+      "step": 3400
+    },
+    {
+      "epoch": 6.46,
+      "eval_winoground_filtered_avg": 0.49,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 3400
+    },
+    {
+      "epoch": 6.47,
+      "grad_norm": 0.26170337200164795,
+      "learning_rate": 0.00017600213121917802,
+      "loss": 0.0212,
+      "step": 3401
+    },
+    {
+      "epoch": 6.47,
+      "grad_norm": 0.16890594363212585,
+      "learning_rate": 0.0001758339453538161,
+      "loss": 0.0081,
+      "step": 3402
+    },
+    {
+      "epoch": 6.47,
+      "grad_norm": 0.21899373829364777,
+      "learning_rate": 0.0001756658065668555,
+      "loss": 0.0093,
+      "step": 3403
+    },
+    {
+      "epoch": 6.47,
+      "grad_norm": 0.23664015531539917,
+      "learning_rate": 0.00017549771492204682,
+      "loss": 0.013,
+      "step": 3404
+    },
+    {
+      "epoch": 6.47,
+      "grad_norm": 0.22118347883224487,
+      "learning_rate": 0.0001753296704831233,
+      "loss": 0.0232,
+      "step": 3405
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 0.16482095420360565,
+      "learning_rate": 0.00017516167331380023,
+      "loss": 0.0091,
+      "step": 3406
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 0.2763455808162689,
+      "learning_rate": 0.00017499372347777473,
+      "loss": 0.0148,
+      "step": 3407
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 0.25034385919570923,
+      "learning_rate": 0.00017482582103872605,
+      "loss": 0.0157,
+      "step": 3408
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 0.24113920331001282,
+      "learning_rate": 0.00017465796606031588,
+      "loss": 0.018,
+      "step": 3409
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 0.19313886761665344,
+      "learning_rate": 0.00017449015860618715,
+      "loss": 0.0087,
+      "step": 3410
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 0.14758391678333282,
+      "learning_rate": 0.00017432239873996555,
+      "loss": 0.0064,
+      "step": 3411
+    },
+    {
+      "epoch": 6.49,
+      "grad_norm": 0.3145643174648285,
+      "learning_rate": 0.000174154686525258,
+      "loss": 0.0222,
+      "step": 3412
+    },
+    {
+      "epoch": 6.49,
+      "grad_norm": 0.261694073677063,
+      "learning_rate": 0.00017398702202565405,
+      "loss": 0.0233,
+      "step": 3413
+    },
+    {
+      "epoch": 6.49,
+      "grad_norm": 0.1400567591190338,
+      "learning_rate": 0.00017381940530472473,
+      "loss": 0.0099,
+      "step": 3414
+    },
+    {
+      "epoch": 6.49,
+      "grad_norm": 0.21918372809886932,
+      "learning_rate": 0.000173651836426023,
+      "loss": 0.0121,
+      "step": 3415
+    },
+    {
+      "epoch": 6.49,
+      "grad_norm": 0.17025597393512726,
+      "learning_rate": 0.00017348431545308367,
+      "loss": 0.0129,
+      "step": 3416
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 0.2272828072309494,
+      "learning_rate": 0.0001733168424494238,
+      "loss": 0.011,
+      "step": 3417
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 0.24448055028915405,
+      "learning_rate": 0.00017314941747854142,
+      "loss": 0.0142,
+      "step": 3418
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 0.17332744598388672,
+      "learning_rate": 0.00017298204060391722,
+      "loss": 0.0097,
+      "step": 3419
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 0.20733897387981415,
+      "learning_rate": 0.00017281471188901326,
+      "loss": 0.0075,
+      "step": 3420
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 0.26947131752967834,
+      "learning_rate": 0.00017264743139727305,
+      "loss": 0.014,
+      "step": 3421
+    },
+    {
+      "epoch": 6.51,
+      "grad_norm": 0.21431225538253784,
+      "learning_rate": 0.00017248019919212238,
+      "loss": 0.0091,
+      "step": 3422
+    },
+    {
+      "epoch": 6.51,
+      "grad_norm": 0.2708381116390228,
+      "learning_rate": 0.0001723130153369684,
+      "loss": 0.0165,
+      "step": 3423
+    },
+    {
+      "epoch": 6.51,
+      "grad_norm": 0.13499288260936737,
+      "learning_rate": 0.0001721458798952001,
+      "loss": 0.0043,
+      "step": 3424
+    },
+    {
+      "epoch": 6.51,
+      "grad_norm": 0.37263721227645874,
+      "learning_rate": 0.00017197879293018772,
+      "loss": 0.0252,
+      "step": 3425
+    },
+    {
+      "epoch": 6.51,
+      "grad_norm": 0.3715760111808777,
+      "learning_rate": 0.00017181175450528387,
+      "loss": 0.0341,
+      "step": 3426
+    },
+    {
+      "epoch": 6.52,
+      "grad_norm": 0.34372109174728394,
+      "learning_rate": 0.0001716447646838218,
+      "loss": 0.016,
+      "step": 3427
+    },
+    {
+      "epoch": 6.52,
+      "grad_norm": 0.24543045461177826,
+      "learning_rate": 0.0001714778235291173,
+      "loss": 0.0125,
+      "step": 3428
+    },
+    {
+      "epoch": 6.52,
+      "grad_norm": 0.17087380588054657,
+      "learning_rate": 0.00017131093110446698,
+      "loss": 0.0102,
+      "step": 3429
+    },
+    {
+      "epoch": 6.52,
+      "grad_norm": 0.17404741048812866,
+      "learning_rate": 0.00017114408747314925,
+      "loss": 0.0121,
+      "step": 3430
+    },
+    {
+      "epoch": 6.52,
+      "grad_norm": 0.27733221650123596,
+      "learning_rate": 0.000170977292698424,
+      "loss": 0.0167,
+      "step": 3431
+    },
+    {
+      "epoch": 6.52,
+      "grad_norm": 0.2070723921060562,
+      "learning_rate": 0.00017081054684353272,
+      "loss": 0.0078,
+      "step": 3432
+    },
+    {
+      "epoch": 6.53,
+      "grad_norm": 0.16392472386360168,
+      "learning_rate": 0.00017064384997169809,
+      "loss": 0.0064,
+      "step": 3433
+    },
+    {
+      "epoch": 6.53,
+      "grad_norm": 0.18133734166622162,
+      "learning_rate": 0.00017047720214612445,
+      "loss": 0.0099,
+      "step": 3434
+    },
+    {
+      "epoch": 6.53,
+      "grad_norm": 0.2104812115430832,
+      "learning_rate": 0.0001703106034299974,
+      "loss": 0.0086,
+      "step": 3435
+    },
+    {
+      "epoch": 6.53,
+      "grad_norm": 0.37937429547309875,
+      "learning_rate": 0.00017014405388648387,
+      "loss": 0.0216,
+      "step": 3436
+    },
+    {
+      "epoch": 6.53,
+      "grad_norm": 0.2932267487049103,
+      "learning_rate": 0.00016997755357873258,
+      "loss": 0.0242,
+      "step": 3437
+    },
+    {
+      "epoch": 6.54,
+      "grad_norm": 0.21098953485488892,
+      "learning_rate": 0.0001698111025698728,
+      "loss": 0.0107,
+      "step": 3438
+    },
+    {
+      "epoch": 6.54,
+      "grad_norm": 0.19323748350143433,
+      "learning_rate": 0.00016964470092301594,
+      "loss": 0.011,
+      "step": 3439
+    },
+    {
+      "epoch": 6.54,
+      "grad_norm": 0.2553883194923401,
+      "learning_rate": 0.00016947834870125416,
+      "loss": 0.0147,
+      "step": 3440
+    },
+    {
+      "epoch": 6.54,
+      "grad_norm": 0.2581605613231659,
+      "learning_rate": 0.00016931204596766101,
+      "loss": 0.0135,
+      "step": 3441
+    },
+    {
+      "epoch": 6.54,
+      "grad_norm": 0.2675668001174927,
+      "learning_rate": 0.0001691457927852914,
+      "loss": 0.0098,
+      "step": 3442
+    },
+    {
+      "epoch": 6.55,
+      "grad_norm": 0.17402218282222748,
+      "learning_rate": 0.00016897958921718119,
+      "loss": 0.0135,
+      "step": 3443
+    },
+    {
+      "epoch": 6.55,
+      "grad_norm": 0.24789339303970337,
+      "learning_rate": 0.0001688134353263477,
+      "loss": 0.0187,
+      "step": 3444
+    },
+    {
+      "epoch": 6.55,
+      "grad_norm": 0.16742469370365143,
+      "learning_rate": 0.0001686473311757892,
+      "loss": 0.0078,
+      "step": 3445
+    },
+    {
+      "epoch": 6.55,
+      "grad_norm": 0.4003869891166687,
+      "learning_rate": 0.00016848127682848548,
+      "loss": 0.0264,
+      "step": 3446
+    },
+    {
+      "epoch": 6.55,
+      "grad_norm": 0.2547103762626648,
+      "learning_rate": 0.00016831527234739677,
+      "loss": 0.0078,
+      "step": 3447
+    },
+    {
+      "epoch": 6.56,
+      "grad_norm": 0.2579076290130615,
+      "learning_rate": 0.00016814931779546513,
+      "loss": 0.0179,
+      "step": 3448
+    },
+    {
+      "epoch": 6.56,
+      "grad_norm": 0.21441352367401123,
+      "learning_rate": 0.00016798341323561318,
+      "loss": 0.0144,
+      "step": 3449
+    },
+    {
+      "epoch": 6.56,
+      "grad_norm": 0.29387399554252625,
+      "learning_rate": 0.0001678175587307448,
+      "loss": 0.0296,
+      "step": 3450
+    },
+    {
+      "epoch": 6.56,
+      "grad_norm": 0.3117658495903015,
+      "learning_rate": 0.00016765175434374488,
+      "loss": 0.0242,
+      "step": 3451
+    },
+    {
+      "epoch": 6.56,
+      "grad_norm": 0.22564835846424103,
+      "learning_rate": 0.00016748600013747925,
+      "loss": 0.0098,
+      "step": 3452
+    },
+    {
+      "epoch": 6.56,
+      "grad_norm": 0.2685282826423645,
+      "learning_rate": 0.00016732029617479472,
+      "loss": 0.0101,
+      "step": 3453
+    },
+    {
+      "epoch": 6.57,
+      "grad_norm": 0.30162855982780457,
+      "learning_rate": 0.00016715464251851897,
+      "loss": 0.026,
+      "step": 3454
+    },
+    {
+      "epoch": 6.57,
+      "grad_norm": 0.16462038457393646,
+      "learning_rate": 0.00016698903923146108,
+      "loss": 0.011,
+      "step": 3455
+    },
+    {
+      "epoch": 6.57,
+      "grad_norm": 0.1699143350124359,
+      "learning_rate": 0.00016682348637641017,
+      "loss": 0.0055,
+      "step": 3456
+    },
+    {
+      "epoch": 6.57,
+      "grad_norm": 0.2699480354785919,
+      "learning_rate": 0.00016665798401613703,
+      "loss": 0.0167,
+      "step": 3457
+    },
+    {
+      "epoch": 6.57,
+      "grad_norm": 0.24445657432079315,
+      "learning_rate": 0.000166492532213393,
+      "loss": 0.0216,
+      "step": 3458
+    },
+    {
+      "epoch": 6.58,
+      "grad_norm": 0.2933023273944855,
+      "learning_rate": 0.00016632713103091016,
+      "loss": 0.0194,
+      "step": 3459
+    },
+    {
+      "epoch": 6.58,
+      "grad_norm": 0.4899539351463318,
+      "learning_rate": 0.0001661617805314015,
+      "loss": 0.0178,
+      "step": 3460
+    },
+    {
+      "epoch": 6.58,
+      "grad_norm": 0.2240944653749466,
+      "learning_rate": 0.00016599648077756082,
+      "loss": 0.0116,
+      "step": 3461
+    },
+    {
+      "epoch": 6.58,
+      "grad_norm": 0.21454301476478577,
+      "learning_rate": 0.00016583123183206246,
+      "loss": 0.0161,
+      "step": 3462
+    },
+    {
+      "epoch": 6.58,
+      "grad_norm": 0.2061445116996765,
+      "learning_rate": 0.00016566603375756212,
+      "loss": 0.0128,
+      "step": 3463
+    },
+    {
+      "epoch": 6.59,
+      "grad_norm": 0.24813877046108246,
+      "learning_rate": 0.0001655008866166952,
+      "loss": 0.0112,
+      "step": 3464
+    },
+    {
+      "epoch": 6.59,
+      "grad_norm": 0.18625973165035248,
+      "learning_rate": 0.0001653357904720788,
+      "loss": 0.0108,
+      "step": 3465
+    },
+    {
+      "epoch": 6.59,
+      "grad_norm": 0.24800622463226318,
+      "learning_rate": 0.00016517074538631018,
+      "loss": 0.0112,
+      "step": 3466
+    },
+    {
+      "epoch": 6.59,
+      "grad_norm": 0.1923033893108368,
+      "learning_rate": 0.00016500575142196702,
+      "loss": 0.0103,
+      "step": 3467
+    },
+    {
+      "epoch": 6.59,
+      "grad_norm": 0.22186177968978882,
+      "learning_rate": 0.00016484080864160816,
+      "loss": 0.0129,
+      "step": 3468
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 0.17639334499835968,
+      "learning_rate": 0.00016467591710777273,
+      "loss": 0.009,
+      "step": 3469
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 0.32647404074668884,
+      "learning_rate": 0.00016451107688298045,
+      "loss": 0.0272,
+      "step": 3470
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 0.1447795331478119,
+      "learning_rate": 0.00016434628802973144,
+      "loss": 0.0056,
+      "step": 3471
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 0.2861032485961914,
+      "learning_rate": 0.00016418155061050693,
+      "loss": 0.0123,
+      "step": 3472
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 0.14975285530090332,
+      "learning_rate": 0.00016401686468776774,
+      "loss": 0.0043,
+      "step": 3473
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 0.22194044291973114,
+      "learning_rate": 0.00016385223032395598,
+      "loss": 0.0098,
+      "step": 3474
+    },
+    {
+      "epoch": 6.61,
+      "grad_norm": 0.2705504596233368,
+      "learning_rate": 0.0001636876475814938,
+      "loss": 0.016,
+      "step": 3475
+    },
+    {
+      "epoch": 6.61,
+      "grad_norm": 0.3080645203590393,
+      "learning_rate": 0.00016352311652278388,
+      "loss": 0.0179,
+      "step": 3476
+    },
+    {
+      "epoch": 6.61,
+      "grad_norm": 0.24837519228458405,
+      "learning_rate": 0.00016335863721020924,
+      "loss": 0.0134,
+      "step": 3477
+    },
+    {
+      "epoch": 6.61,
+      "grad_norm": 0.29112738370895386,
+      "learning_rate": 0.00016319420970613334,
+      "loss": 0.0162,
+      "step": 3478
+    },
+    {
+      "epoch": 6.61,
+      "grad_norm": 0.22857633233070374,
+      "learning_rate": 0.00016302983407290002,
+      "loss": 0.0202,
+      "step": 3479
+    },
+    {
+      "epoch": 6.62,
+      "grad_norm": 0.18444441258907318,
+      "learning_rate": 0.0001628655103728332,
+      "loss": 0.011,
+      "step": 3480
+    },
+    {
+      "epoch": 6.62,
+      "grad_norm": 0.15074364840984344,
+      "learning_rate": 0.0001627012386682378,
+      "loss": 0.0052,
+      "step": 3481
+    },
+    {
+      "epoch": 6.62,
+      "grad_norm": 0.2177027016878128,
+      "learning_rate": 0.00016253701902139796,
+      "loss": 0.0094,
+      "step": 3482
+    },
+    {
+      "epoch": 6.62,
+      "grad_norm": 0.24838688969612122,
+      "learning_rate": 0.00016237285149457928,
+      "loss": 0.0219,
+      "step": 3483
+    },
+    {
+      "epoch": 6.62,
+      "grad_norm": 0.2829641103744507,
+      "learning_rate": 0.00016220873615002644,
+      "loss": 0.0151,
+      "step": 3484
+    },
+    {
+      "epoch": 6.63,
+      "grad_norm": 0.23266643285751343,
+      "learning_rate": 0.00016204467304996518,
+      "loss": 0.0167,
+      "step": 3485
+    },
+    {
+      "epoch": 6.63,
+      "grad_norm": 0.25052645802497864,
+      "learning_rate": 0.0001618806622566011,
+      "loss": 0.0138,
+      "step": 3486
+    },
+    {
+      "epoch": 6.63,
+      "grad_norm": 0.24766546487808228,
+      "learning_rate": 0.00016171670383211998,
+      "loss": 0.0125,
+      "step": 3487
+    },
+    {
+      "epoch": 6.63,
+      "grad_norm": 0.23045948147773743,
+      "learning_rate": 0.0001615527978386878,
+      "loss": 0.0105,
+      "step": 3488
+    },
+    {
+      "epoch": 6.63,
+      "grad_norm": 0.29384782910346985,
+      "learning_rate": 0.0001613889443384506,
+      "loss": 0.0192,
+      "step": 3489
+    },
+    {
+      "epoch": 6.63,
+      "grad_norm": 0.29214125871658325,
+      "learning_rate": 0.00016122514339353457,
+      "loss": 0.0131,
+      "step": 3490
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 0.28599414229393005,
+      "learning_rate": 0.0001610613950660457,
+      "loss": 0.0207,
+      "step": 3491
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 0.25773048400878906,
+      "learning_rate": 0.00016089769941807084,
+      "loss": 0.0123,
+      "step": 3492
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 0.18247129023075104,
+      "learning_rate": 0.0001607340565116757,
+      "loss": 0.0142,
+      "step": 3493
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 0.3562219440937042,
+      "learning_rate": 0.00016057046640890698,
+      "loss": 0.0218,
+      "step": 3494
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 0.42792537808418274,
+      "learning_rate": 0.00016040692917179084,
+      "loss": 0.0177,
+      "step": 3495
+    },
+    {
+      "epoch": 6.65,
+      "grad_norm": 0.27011194825172424,
+      "learning_rate": 0.0001602434448623336,
+      "loss": 0.0132,
+      "step": 3496
+    },
+    {
+      "epoch": 6.65,
+      "grad_norm": 0.21817974746227264,
+      "learning_rate": 0.00016008001354252138,
+      "loss": 0.0137,
+      "step": 3497
+    },
+    {
+      "epoch": 6.65,
+      "grad_norm": 0.23021112382411957,
+      "learning_rate": 0.00015991663527432028,
+      "loss": 0.013,
+      "step": 3498
+    },
+    {
+      "epoch": 6.65,
+      "grad_norm": 0.37814390659332275,
+      "learning_rate": 0.00015975331011967628,
+      "loss": 0.0246,
+      "step": 3499
+    },
+    {
+      "epoch": 6.65,
+      "grad_norm": 0.30496376752853394,
+      "learning_rate": 0.0001595900381405151,
+      "loss": 0.0164,
+      "step": 3500
+    },
+    {
+      "epoch": 6.65,
+      "eval_blimp_filtered_avg": 0.5425373134328358,
+      "eval_blimp_filtered_std": 0.005848277415337074,
+      "step": 3500
+    },
+    {
+      "epoch": 6.65,
+      "eval_blimp_supplement_avg": 0.5538793103448276,
+      "eval_blimp_supplement_std": 0.023036724260967142,
+      "step": 3500
+    },
+    {
+      "epoch": 6.65,
+      "eval_vqa_filtered_avg": 0.14,
+      "eval_vqa_filtered_std": 0.034873508801977704,
+      "step": 3500
+    },
+    {
+      "epoch": 6.65,
+      "eval_winoground_filtered_avg": 0.47,
+      "eval_winoground_filtered_std": 0.0501613558046592,
+      "step": 3500
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 0.22132186591625214,
+      "learning_rate": 0.0001594268193987428,
+      "loss": 0.0161,
+      "step": 3501
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 0.20204704999923706,
+      "learning_rate": 0.00015926365395624434,
+      "loss": 0.0204,
+      "step": 3502
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 0.2163960337638855,
+      "learning_rate": 0.00015910054187488542,
+      "loss": 0.0076,
+      "step": 3503
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 0.24517259001731873,
+      "learning_rate": 0.00015893748321651094,
+      "loss": 0.0132,
+      "step": 3504
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 0.3865545094013214,
+      "learning_rate": 0.0001587744780429457,
+      "loss": 0.0257,
+      "step": 3505
+    },
+    {
+      "epoch": 6.67,
+      "grad_norm": 0.27735504508018494,
+      "learning_rate": 0.00015861152641599412,
+      "loss": 0.0203,
+      "step": 3506
+    },
+    {
+      "epoch": 6.67,
+      "grad_norm": 0.11114240437746048,
+      "learning_rate": 0.00015844862839744077,
+      "loss": 0.0038,
+      "step": 3507
+    },
+    {
+      "epoch": 6.67,
+      "grad_norm": 0.21170379221439362,
+      "learning_rate": 0.000158285784049049,
+      "loss": 0.0096,
+      "step": 3508
+    },
+    {
+      "epoch": 6.67,
+      "grad_norm": 0.21150603890419006,
+      "learning_rate": 0.00015812299343256287,
+      "loss": 0.0073,
+      "step": 3509
+    },
+    {
+      "epoch": 6.67,
+      "grad_norm": 0.20711956918239594,
+      "learning_rate": 0.00015796025660970512,
+      "loss": 0.0136,
+      "step": 3510
+    },
+    {
+      "epoch": 6.67,
+      "grad_norm": 0.24845798313617706,
+      "learning_rate": 0.0001577975736421788,
+      "loss": 0.0136,
+      "step": 3511
+    },
+    {
+      "epoch": 6.68,
+      "grad_norm": 0.3459714949131012,
+      "learning_rate": 0.00015763494459166625,
+      "loss": 0.0167,
+      "step": 3512
+    },
+    {
+      "epoch": 6.68,
+      "grad_norm": 0.2271159440279007,
+      "learning_rate": 0.00015747236951982907,
+      "loss": 0.0138,
+      "step": 3513
+    },
+    {
+      "epoch": 6.68,
+      "grad_norm": 0.22579340636730194,
+      "learning_rate": 0.00015730984848830905,
+      "loss": 0.0094,
+      "step": 3514
+    },
+    {
+      "epoch": 6.68,
+      "grad_norm": 0.29341647028923035,
+      "learning_rate": 0.000157147381558727,
+      "loss": 0.0229,
+      "step": 3515
+    },
+    {
+      "epoch": 6.68,
+      "grad_norm": 0.2175038754940033,
+      "learning_rate": 0.00015698496879268338,
+      "loss": 0.014,
+      "step": 3516
+    },
+    {
+      "epoch": 6.69,
+      "grad_norm": 0.153773233294487,
+      "learning_rate": 0.00015682261025175796,
+      "loss": 0.0044,
+      "step": 3517
+    },
+    {
+      "epoch": 6.69,
+      "grad_norm": 0.2568443715572357,
+      "learning_rate": 0.00015666030599751046,
+      "loss": 0.0194,
+      "step": 3518
+    },
+    {
+      "epoch": 6.69,
+      "grad_norm": 0.14005787670612335,
+      "learning_rate": 0.00015649805609147912,
+      "loss": 0.0042,
+      "step": 3519
+    },
+    {
+      "epoch": 6.69,
+      "grad_norm": 0.17617319524288177,
+      "learning_rate": 0.00015633586059518245,
+      "loss": 0.0065,
+      "step": 3520
+    },
+    {
+      "epoch": 6.69,
+      "grad_norm": 0.21150831878185272,
+      "learning_rate": 0.00015617371957011788,
+      "loss": 0.0127,
+      "step": 3521
+    },
+    {
+      "epoch": 6.7,
+      "grad_norm": 0.2956348657608032,
+      "learning_rate": 0.00015601163307776229,
+      "loss": 0.0085,
+      "step": 3522
+    },
+    {
+      "epoch": 6.7,
+      "grad_norm": 0.2579307556152344,
+      "learning_rate": 0.00015584960117957187,
+      "loss": 0.0146,
+      "step": 3523
+    },
+    {
+      "epoch": 6.7,
+      "grad_norm": 0.33557552099227905,
+      "learning_rate": 0.00015568762393698205,
+      "loss": 0.0252,
+      "step": 3524
+    },
+    {
+      "epoch": 6.7,
+      "grad_norm": 0.40421515703201294,
+      "learning_rate": 0.00015552570141140766,
+      "loss": 0.0278,
+      "step": 3525
+    },
+    {
+      "epoch": 6.7,
+      "grad_norm": 0.1881566345691681,
+      "learning_rate": 0.0001553638336642426,
+      "loss": 0.0068,
+      "step": 3526
+    },
+    {
+      "epoch": 6.71,
+      "grad_norm": 0.3436504900455475,
+      "learning_rate": 0.00015520202075686046,
+      "loss": 0.019,
+      "step": 3527
+    },
+    {
+      "epoch": 6.71,
+      "grad_norm": 0.17568156123161316,
+      "learning_rate": 0.00015504026275061324,
+      "loss": 0.0066,
+      "step": 3528
+    },
+    {
+      "epoch": 6.71,
+      "grad_norm": 0.24178263545036316,
+      "learning_rate": 0.00015487855970683291,
+      "loss": 0.0092,
+      "step": 3529
+    },
+    {
+      "epoch": 6.71,
+      "grad_norm": 0.29775843024253845,
+      "learning_rate": 0.00015471691168683022,
+      "loss": 0.0168,
+      "step": 3530
+    },
+    {
+      "epoch": 6.71,
+      "grad_norm": 0.24221335351467133,
+      "learning_rate": 0.00015455531875189506,
+      "loss": 0.012,
+      "step": 3531
+    },
+    {
+      "epoch": 6.71,
+      "grad_norm": 0.19742120802402496,
+      "learning_rate": 0.00015439378096329648,
+      "loss": 0.011,
+      "step": 3532
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 0.291246235370636,
+      "learning_rate": 0.0001542322983822827,
+      "loss": 0.0199,
+      "step": 3533
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 0.21611259877681732,
+      "learning_rate": 0.00015407087107008083,
+      "loss": 0.0123,
+      "step": 3534
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 0.2187098264694214,
+      "learning_rate": 0.00015390949908789718,
+      "loss": 0.015,
+      "step": 3535
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 0.18280333280563354,
+      "learning_rate": 0.00015374818249691707,
+      "loss": 0.0074,
+      "step": 3536
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 0.22524289786815643,
+      "learning_rate": 0.0001535869213583046,
+      "loss": 0.013,
+      "step": 3537
+    },
+    {
+      "epoch": 6.73,
+      "grad_norm": 0.20895539224147797,
+      "learning_rate": 0.00015342571573320342,
+      "loss": 0.0091,
+      "step": 3538
+    },
+    {
+      "epoch": 6.73,
+      "grad_norm": 0.31643471121788025,
+      "learning_rate": 0.00015326456568273518,
+      "loss": 0.0221,
+      "step": 3539
+    },
+    {
+      "epoch": 6.73,
+      "grad_norm": 0.4021128714084625,
+      "learning_rate": 0.00015310347126800148,
+      "loss": 0.05,
+      "step": 3540
+    },
+    {
+      "epoch": 6.73,
+      "grad_norm": 0.4130493104457855,
+      "learning_rate": 0.0001529424325500822,
+      "loss": 0.0368,
+      "step": 3541
+    },
+    {
+      "epoch": 6.73,
+      "grad_norm": 0.268380731344223,
+      "learning_rate": 0.0001527814495900362,
+      "loss": 0.0165,
+      "step": 3542
+    },
+    {
+      "epoch": 6.74,
+      "grad_norm": 0.18179921805858612,
+      "learning_rate": 0.0001526205224489014,
+      "loss": 0.0078,
+      "step": 3543
+    },
+    {
+      "epoch": 6.74,
+      "grad_norm": 0.2605191469192505,
+      "learning_rate": 0.0001524596511876943,
+      "loss": 0.0138,
+      "step": 3544
+    },
+    {
+      "epoch": 6.74,
+      "grad_norm": 0.2293798327445984,
+      "learning_rate": 0.00015229883586741035,
+      "loss": 0.0108,
+      "step": 3545
+    },
+    {
+      "epoch": 6.74,
+      "grad_norm": 0.1982216238975525,
+      "learning_rate": 0.00015213807654902367,
+      "loss": 0.0108,
+      "step": 3546
+    },
+    {
+      "epoch": 6.74,
+      "grad_norm": 0.1339135617017746,
+      "learning_rate": 0.00015197737329348754,
+      "loss": 0.0059,
+      "step": 3547
+    },
+    {
+      "epoch": 6.75,
+      "grad_norm": 0.21493743360042572,
+      "learning_rate": 0.0001518167261617333,
+      "loss": 0.0159,
+      "step": 3548
+    },
+    {
+      "epoch": 6.75,
+      "grad_norm": 0.2949875593185425,
+      "learning_rate": 0.0001516561352146717,
+      "loss": 0.0126,
+      "step": 3549
+    },
+    {
+      "epoch": 6.75,
+      "grad_norm": 0.21591828763484955,
+      "learning_rate": 0.00015149560051319182,
+      "loss": 0.0106,
+      "step": 3550
+    },
+    {
+      "epoch": 6.75,
+      "grad_norm": 0.15970522165298462,
+      "learning_rate": 0.00015133512211816142,
+      "loss": 0.0104,
+      "step": 3551
+    },
+    {
+      "epoch": 6.75,
+      "grad_norm": 0.269972026348114,
+      "learning_rate": 0.00015117470009042691,
+      "loss": 0.0212,
+      "step": 3552
+    },
+    {
+      "epoch": 6.75,
+      "grad_norm": 0.1543676257133484,
+      "learning_rate": 0.00015101433449081367,
+      "loss": 0.0076,
+      "step": 3553
+    },
+    {
+      "epoch": 6.76,
+      "grad_norm": 0.17760495841503143,
+      "learning_rate": 0.000150854025380125,
+      "loss": 0.0098,
+      "step": 3554
+    },
+    {
+      "epoch": 6.76,
+      "grad_norm": 0.23355263471603394,
+      "learning_rate": 0.0001506937728191435,
+      "loss": 0.0137,
+      "step": 3555
+    },
+    {
+      "epoch": 6.76,
+      "grad_norm": 0.23258772492408752,
+      "learning_rate": 0.00015053357686862994,
+      "loss": 0.0167,
+      "step": 3556
+    },
+    {
+      "epoch": 6.76,
+      "grad_norm": 0.18455778062343597,
+      "learning_rate": 0.00015037343758932364,
+      "loss": 0.0159,
+      "step": 3557
+    },
+    {
+      "epoch": 6.76,
+      "grad_norm": 0.19678185880184174,
+      "learning_rate": 0.00015021335504194263,
+      "loss": 0.0143,
+      "step": 3558
+    },
+    {
+      "epoch": 6.77,
+      "grad_norm": 0.37779760360717773,
+      "learning_rate": 0.00015005332928718298,
+      "loss": 0.0224,
+      "step": 3559
+    },
+    {
+      "epoch": 6.77,
+      "grad_norm": 0.1682596355676651,
+      "learning_rate": 0.00014989336038571984,
+      "loss": 0.0082,
+      "step": 3560
+    },
+    {
+      "epoch": 6.77,
+      "grad_norm": 0.19949524104595184,
+      "learning_rate": 0.00014973344839820643,
+      "loss": 0.0108,
+      "step": 3561
+    },
+    {
+      "epoch": 6.77,
+      "grad_norm": 0.16154934465885162,
+      "learning_rate": 0.00014957359338527444,
+      "loss": 0.0154,
+      "step": 3562
+    },
+    {
+      "epoch": 6.77,
+      "grad_norm": 0.1221017837524414,
+      "learning_rate": 0.00014941379540753388,
+      "loss": 0.009,
+      "step": 3563
+    },
+    {
+      "epoch": 6.78,
+      "grad_norm": 0.19914448261260986,
+      "learning_rate": 0.00014925405452557354,
+      "loss": 0.0127,
+      "step": 3564
+    },
+    {
+      "epoch": 6.78,
+      "grad_norm": 0.1338150054216385,
+      "learning_rate": 0.00014909437079995986,
+      "loss": 0.0048,
+      "step": 3565
+    },
+    {
+      "epoch": 6.78,
+      "grad_norm": 0.23834961652755737,
+      "learning_rate": 0.0001489347442912383,
+      "loss": 0.0164,
+      "step": 3566
+    },
+    {
+      "epoch": 6.78,
+      "grad_norm": 0.21305963397026062,
+      "learning_rate": 0.00014877517505993223,
+      "loss": 0.0093,
+      "step": 3567
+    },
+    {
+      "epoch": 6.78,
+      "grad_norm": 0.3246478736400604,
+      "learning_rate": 0.00014861566316654346,
+      "loss": 0.0165,
+      "step": 3568
+    },
+    {
+      "epoch": 6.79,
+      "grad_norm": 0.25084373354911804,
+      "learning_rate": 0.0001484562086715519,
+      "loss": 0.0141,
+      "step": 3569
+    },
+    {
+      "epoch": 6.79,
+      "grad_norm": 0.21717716753482819,
+      "learning_rate": 0.0001482968116354159,
+      "loss": 0.0152,
+      "step": 3570
+    },
+    {
+      "epoch": 6.79,
+      "grad_norm": 0.18900321424007416,
+      "learning_rate": 0.00014813747211857187,
+      "loss": 0.0092,
+      "step": 3571
+    },
+    {
+      "epoch": 6.79,
+      "grad_norm": 0.26227426528930664,
+      "learning_rate": 0.00014797819018143437,
+      "loss": 0.0151,
+      "step": 3572
+    },
+    {
+      "epoch": 6.79,
+      "grad_norm": 0.3139238953590393,
+      "learning_rate": 0.00014781896588439658,
+      "loss": 0.0225,
+      "step": 3573
+    },
+    {
+      "epoch": 6.79,
+      "grad_norm": 0.1994101107120514,
+      "learning_rate": 0.000147659799287829,
+      "loss": 0.0076,
+      "step": 3574
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 0.16667909920215607,
+      "learning_rate": 0.00014750069045208107,
+      "loss": 0.0106,
+      "step": 3575
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 0.193839430809021,
+      "learning_rate": 0.00014734163943747985,
+      "loss": 0.0107,
+      "step": 3576
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 0.17453965544700623,
+      "learning_rate": 0.00014718264630433065,
+      "loss": 0.0149,
+      "step": 3577
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 0.2228081077337265,
+      "learning_rate": 0.0001470237111129168,
+      "loss": 0.0087,
+      "step": 3578
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 0.32927340269088745,
+      "learning_rate": 0.00014686483392349959,
+      "loss": 0.0177,
+      "step": 3579
+    },
+    {
+      "epoch": 6.81,
+      "grad_norm": 0.227274551987648,
+      "learning_rate": 0.00014670601479631841,
+      "loss": 0.0101,
+      "step": 3580
+    },
+    {
+      "epoch": 6.81,
+      "grad_norm": 0.21727076172828674,
+      "learning_rate": 0.00014654725379159065,
+      "loss": 0.0155,
+      "step": 3581
+    },
+    {
+      "epoch": 6.81,
+      "grad_norm": 0.3101034462451935,
+      "learning_rate": 0.0001463885509695116,
+      "loss": 0.0148,
+      "step": 3582
+    },
+    {
+      "epoch": 6.81,
+      "grad_norm": 0.3826908767223358,
+      "learning_rate": 0.00014622990639025438,
+      "loss": 0.0229,
+      "step": 3583
+    },
+    {
+      "epoch": 6.81,
+      "grad_norm": 0.18059301376342773,
+      "learning_rate": 0.0001460713201139705,
+      "loss": 0.0108,
+      "step": 3584
+    },
+    {
+      "epoch": 6.82,
+      "grad_norm": 0.27148863673210144,
+      "learning_rate": 0.00014591279220078855,
+      "loss": 0.0241,
+      "step": 3585
+    },
+    {
+      "epoch": 6.82,
+      "grad_norm": 0.20044052600860596,
+      "learning_rate": 0.00014575432271081586,
+      "loss": 0.0113,
+      "step": 3586
+    },
+    {
+      "epoch": 6.82,
+      "grad_norm": 0.1551116406917572,
+      "learning_rate": 0.00014559591170413706,
+      "loss": 0.0084,
+      "step": 3587
+    },
+    {
+      "epoch": 6.82,
+      "grad_norm": 0.21452130377292633,
+      "learning_rate": 0.00014543755924081476,
+      "loss": 0.0094,
+      "step": 3588
+    },
+    {
+      "epoch": 6.82,
+      "grad_norm": 0.14290033280849457,
+      "learning_rate": 0.00014527926538088936,
+      "loss": 0.0077,
+      "step": 3589
+    },
+    {
+      "epoch": 6.83,
+      "grad_norm": 0.287442684173584,
+      "learning_rate": 0.00014512103018437907,
+      "loss": 0.0195,
+      "step": 3590
+    },
+    {
+      "epoch": 6.83,
+      "grad_norm": 0.22354090213775635,
+      "learning_rate": 0.00014496285371127984,
+      "loss": 0.015,
+      "step": 3591
+    },
+    {
+      "epoch": 6.83,
+      "grad_norm": 0.13138145208358765,
+      "learning_rate": 0.0001448047360215652,
+      "loss": 0.0054,
+      "step": 3592
+    },
+    {
+      "epoch": 6.83,
+      "grad_norm": 0.26583728194236755,
+      "learning_rate": 0.00014464667717518692,
+      "loss": 0.0131,
+      "step": 3593
+    },
+    {
+      "epoch": 6.83,
+      "grad_norm": 0.26246827840805054,
+      "learning_rate": 0.00014448867723207363,
+      "loss": 0.0114,
+      "step": 3594
+    },
+    {
+      "epoch": 6.83,
+      "grad_norm": 0.25939589738845825,
+      "learning_rate": 0.00014433073625213242,
+      "loss": 0.0158,
+      "step": 3595
+    },
+    {
+      "epoch": 6.84,
+      "grad_norm": 0.27061042189598083,
+      "learning_rate": 0.00014417285429524753,
+      "loss": 0.0148,
+      "step": 3596
+    },
+    {
+      "epoch": 6.84,
+      "grad_norm": 0.1431720107793808,
+      "learning_rate": 0.000144015031421281,
+      "loss": 0.0072,
+      "step": 3597
+    },
+    {
+      "epoch": 6.84,
+      "grad_norm": 0.24964430928230286,
+      "learning_rate": 0.0001438572676900723,
+      "loss": 0.0102,
+      "step": 3598
+    },
+    {
+      "epoch": 6.84,
+      "grad_norm": 0.14568409323692322,
+      "learning_rate": 0.00014369956316143907,
+      "loss": 0.0069,
+      "step": 3599
+    },
+    {
+      "epoch": 6.84,
+      "grad_norm": 0.2597874104976654,
+      "learning_rate": 0.00014354191789517546,
+      "loss": 0.0111,
+      "step": 3600
+    },
+    {
+      "epoch": 6.84,
+      "eval_blimp_filtered_avg": 0.5408955223880597,
+      "eval_blimp_filtered_std": 0.005847142270179747,
+      "step": 3600
+    },
+    {
+      "epoch": 6.84,
+      "eval_blimp_supplement_avg": 0.5387931034482759,
+      "eval_blimp_supplement_std": 0.02314297562598383,
+      "step": 3600
+    },
+    {
+      "epoch": 6.84,
+      "eval_vqa_filtered_avg": 0.12,
+      "eval_vqa_filtered_std": 0.03265986323710906,
+      "step": 3600
+    },
+    {
+      "epoch": 6.84,
+      "eval_winoground_filtered_avg": 0.51,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 3600
+    },
+    {
+      "epoch": 6.85,
+      "grad_norm": 0.30834251642227173,
+      "learning_rate": 0.00014338433195105416,
+      "loss": 0.0255,
+      "step": 3601
+    },
+    {
+      "epoch": 6.85,
+      "grad_norm": 0.2665657103061676,
+      "learning_rate": 0.00014322680538882482,
+      "loss": 0.011,
+      "step": 3602
+    },
+    {
+      "epoch": 6.85,
+      "grad_norm": 0.27100345492362976,
+      "learning_rate": 0.00014306933826821466,
+      "loss": 0.0175,
+      "step": 3603
+    },
+    {
+      "epoch": 6.85,
+      "grad_norm": 0.24496594071388245,
+      "learning_rate": 0.00014291193064892846,
+      "loss": 0.0185,
+      "step": 3604
+    },
+    {
+      "epoch": 6.85,
+      "grad_norm": 0.20313118398189545,
+      "learning_rate": 0.00014275458259064837,
+      "loss": 0.0122,
+      "step": 3605
+    },
+    {
+      "epoch": 6.86,
+      "grad_norm": 0.2646270990371704,
+      "learning_rate": 0.000142597294153034,
+      "loss": 0.0135,
+      "step": 3606
+    },
+    {
+      "epoch": 6.86,
+      "grad_norm": 0.17286130785942078,
+      "learning_rate": 0.00014244006539572226,
+      "loss": 0.0084,
+      "step": 3607
+    },
+    {
+      "epoch": 6.86,
+      "grad_norm": 0.20812182128429413,
+      "learning_rate": 0.00014228289637832755,
+      "loss": 0.0096,
+      "step": 3608
+    },
+    {
+      "epoch": 6.86,
+      "grad_norm": 0.1718285083770752,
+      "learning_rate": 0.00014212578716044143,
+      "loss": 0.0111,
+      "step": 3609
+    },
+    {
+      "epoch": 6.86,
+      "grad_norm": 0.16319935023784637,
+      "learning_rate": 0.00014196873780163329,
+      "loss": 0.0076,
+      "step": 3610
+    },
+    {
+      "epoch": 6.87,
+      "grad_norm": 0.22734007239341736,
+      "learning_rate": 0.00014181174836144897,
+      "loss": 0.0067,
+      "step": 3611
+    },
+    {
+      "epoch": 6.87,
+      "grad_norm": 0.19892044365406036,
+      "learning_rate": 0.00014165481889941248,
+      "loss": 0.0125,
+      "step": 3612
+    },
+    {
+      "epoch": 6.87,
+      "grad_norm": 0.18760593235492706,
+      "learning_rate": 0.0001414979494750245,
+      "loss": 0.0103,
+      "step": 3613
+    },
+    {
+      "epoch": 6.87,
+      "grad_norm": 0.20384031534194946,
+      "learning_rate": 0.0001413411401477632,
+      "loss": 0.0122,
+      "step": 3614
+    },
+    {
+      "epoch": 6.87,
+      "grad_norm": 0.2395206093788147,
+      "learning_rate": 0.00014118439097708386,
+      "loss": 0.0112,
+      "step": 3615
+    },
+    {
+      "epoch": 6.87,
+      "grad_norm": 0.1963171809911728,
+      "learning_rate": 0.00014102770202241901,
+      "loss": 0.016,
+      "step": 3616
+    },
+    {
+      "epoch": 6.88,
+      "grad_norm": 0.27020585536956787,
+      "learning_rate": 0.0001408710733431783,
+      "loss": 0.0135,
+      "step": 3617
+    },
+    {
+      "epoch": 6.88,
+      "grad_norm": 0.27207937836647034,
+      "learning_rate": 0.0001407145049987485,
+      "loss": 0.0105,
+      "step": 3618
+    },
+    {
+      "epoch": 6.88,
+      "grad_norm": 0.2537366449832916,
+      "learning_rate": 0.00014055799704849385,
+      "loss": 0.0156,
+      "step": 3619
+    },
+    {
+      "epoch": 6.88,
+      "grad_norm": 0.2007504254579544,
+      "learning_rate": 0.00014040154955175492,
+      "loss": 0.012,
+      "step": 3620
+    },
+    {
+      "epoch": 6.88,
+      "grad_norm": 0.22894182801246643,
+      "learning_rate": 0.00014024516256785024,
+      "loss": 0.0188,
+      "step": 3621
+    },
+    {
+      "epoch": 6.89,
+      "grad_norm": 0.2475810945034027,
+      "learning_rate": 0.00014008883615607487,
+      "loss": 0.0098,
+      "step": 3622
+    },
+    {
+      "epoch": 6.89,
+      "grad_norm": 0.31347689032554626,
+      "learning_rate": 0.00013993257037570102,
+      "loss": 0.0262,
+      "step": 3623
+    },
+    {
+      "epoch": 6.89,
+      "grad_norm": 0.21904762089252472,
+      "learning_rate": 0.0001397763652859779,
+      "loss": 0.0196,
+      "step": 3624
+    },
+    {
+      "epoch": 6.89,
+      "grad_norm": 0.2842084467411041,
+      "learning_rate": 0.0001396202209461318,
+      "loss": 0.0133,
+      "step": 3625
+    },
+    {
+      "epoch": 6.89,
+      "grad_norm": 0.34036341309547424,
+      "learning_rate": 0.0001394641374153658,
+      "loss": 0.0213,
+      "step": 3626
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 0.2091987282037735,
+      "learning_rate": 0.00013930811475285998,
+      "loss": 0.0157,
+      "step": 3627
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 0.27661463618278503,
+      "learning_rate": 0.00013915215301777168,
+      "loss": 0.0175,
+      "step": 3628
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 0.2327796071767807,
+      "learning_rate": 0.00013899625226923448,
+      "loss": 0.0148,
+      "step": 3629
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 0.10735620558261871,
+      "learning_rate": 0.0001388404125663596,
+      "loss": 0.0052,
+      "step": 3630
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 0.22661687433719635,
+      "learning_rate": 0.00013868463396823425,
+      "loss": 0.0123,
+      "step": 3631
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 0.10397245734930038,
+      "learning_rate": 0.00013852891653392334,
+      "loss": 0.0105,
+      "step": 3632
+    },
+    {
+      "epoch": 6.91,
+      "grad_norm": 0.21884872019290924,
+      "learning_rate": 0.000138373260322468,
+      "loss": 0.0068,
+      "step": 3633
+    },
+    {
+      "epoch": 6.91,
+      "grad_norm": 0.3188261091709137,
+      "learning_rate": 0.00013821766539288643,
+      "loss": 0.0226,
+      "step": 3634
+    },
+    {
+      "epoch": 6.91,
+      "grad_norm": 0.20242483913898468,
+      "learning_rate": 0.00013806213180417342,
+      "loss": 0.0078,
+      "step": 3635
+    },
+    {
+      "epoch": 6.91,
+      "grad_norm": 0.3234082758426666,
+      "learning_rate": 0.00013790665961530062,
+      "loss": 0.0104,
+      "step": 3636
+    },
+    {
+      "epoch": 6.91,
+      "grad_norm": 0.2117815762758255,
+      "learning_rate": 0.0001377512488852164,
+      "loss": 0.008,
+      "step": 3637
+    },
+    {
+      "epoch": 6.92,
+      "grad_norm": 0.22797289490699768,
+      "learning_rate": 0.00013759589967284557,
+      "loss": 0.0159,
+      "step": 3638
+    },
+    {
+      "epoch": 6.92,
+      "grad_norm": 0.2831760048866272,
+      "learning_rate": 0.00013744061203709032,
+      "loss": 0.0088,
+      "step": 3639
+    },
+    {
+      "epoch": 6.92,
+      "grad_norm": 0.2702014148235321,
+      "learning_rate": 0.00013728538603682844,
+      "loss": 0.0241,
+      "step": 3640
+    },
+    {
+      "epoch": 6.92,
+      "grad_norm": 0.29624152183532715,
+      "learning_rate": 0.0001371302217309153,
+      "loss": 0.0212,
+      "step": 3641
+    },
+    {
+      "epoch": 6.92,
+      "grad_norm": 0.18596014380455017,
+      "learning_rate": 0.00013697511917818241,
+      "loss": 0.0154,
+      "step": 3642
+    },
+    {
+      "epoch": 6.93,
+      "grad_norm": 0.22594869136810303,
+      "learning_rate": 0.00013682007843743792,
+      "loss": 0.011,
+      "step": 3643
+    },
+    {
+      "epoch": 6.93,
+      "grad_norm": 0.2714582681655884,
+      "learning_rate": 0.0001366650995674665,
+      "loss": 0.0162,
+      "step": 3644
+    },
+    {
+      "epoch": 6.93,
+      "grad_norm": 0.2908836305141449,
+      "learning_rate": 0.00013651018262702974,
+      "loss": 0.0134,
+      "step": 3645
+    },
+    {
+      "epoch": 6.93,
+      "grad_norm": 0.1818009614944458,
+      "learning_rate": 0.000136355327674865,
+      "loss": 0.0092,
+      "step": 3646
+    },
+    {
+      "epoch": 6.93,
+      "grad_norm": 0.20754975080490112,
+      "learning_rate": 0.00013620053476968695,
+      "loss": 0.0101,
+      "step": 3647
+    },
+    {
+      "epoch": 6.94,
+      "grad_norm": 0.16264256834983826,
+      "learning_rate": 0.00013604580397018622,
+      "loss": 0.0085,
+      "step": 3648
+    },
+    {
+      "epoch": 6.94,
+      "grad_norm": 0.21996024250984192,
+      "learning_rate": 0.00013589113533503007,
+      "loss": 0.0163,
+      "step": 3649
+    },
+    {
+      "epoch": 6.94,
+      "grad_norm": 0.18879708647727966,
+      "learning_rate": 0.00013573652892286215,
+      "loss": 0.0123,
+      "step": 3650
+    },
+    {
+      "epoch": 6.94,
+      "grad_norm": 0.3309245705604553,
+      "learning_rate": 0.00013558198479230254,
+      "loss": 0.024,
+      "step": 3651
+    },
+    {
+      "epoch": 6.94,
+      "grad_norm": 0.17644555866718292,
+      "learning_rate": 0.00013542750300194771,
+      "loss": 0.0097,
+      "step": 3652
+    },
+    {
+      "epoch": 6.94,
+      "grad_norm": 0.19607970118522644,
+      "learning_rate": 0.0001352730836103703,
+      "loss": 0.0153,
+      "step": 3653
+    },
+    {
+      "epoch": 6.95,
+      "grad_norm": 0.1683328002691269,
+      "learning_rate": 0.00013511872667611985,
+      "loss": 0.0126,
+      "step": 3654
+    },
+    {
+      "epoch": 6.95,
+      "grad_norm": 0.13821370899677277,
+      "learning_rate": 0.0001349644322577214,
+      "loss": 0.0045,
+      "step": 3655
+    },
+    {
+      "epoch": 6.95,
+      "grad_norm": 0.13564372062683105,
+      "learning_rate": 0.00013481020041367715,
+      "loss": 0.0077,
+      "step": 3656
+    },
+    {
+      "epoch": 6.95,
+      "grad_norm": 0.16568143665790558,
+      "learning_rate": 0.00013465603120246474,
+      "loss": 0.0075,
+      "step": 3657
+    },
+    {
+      "epoch": 6.95,
+      "grad_norm": 0.24049104750156403,
+      "learning_rate": 0.00013450192468253874,
+      "loss": 0.0079,
+      "step": 3658
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 0.2175944298505783,
+      "learning_rate": 0.00013434788091232962,
+      "loss": 0.0165,
+      "step": 3659
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 0.30983033776283264,
+      "learning_rate": 0.00013419389995024408,
+      "loss": 0.0176,
+      "step": 3660
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 0.1811114102602005,
+      "learning_rate": 0.0001340399818546651,
+      "loss": 0.007,
+      "step": 3661
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 0.14656805992126465,
+      "learning_rate": 0.00013388612668395174,
+      "loss": 0.0093,
+      "step": 3662
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 0.1849527657032013,
+      "learning_rate": 0.00013373233449643921,
+      "loss": 0.0046,
+      "step": 3663
+    },
+    {
+      "epoch": 6.97,
+      "grad_norm": 0.2996760308742523,
+      "learning_rate": 0.0001335786053504388,
+      "loss": 0.0144,
+      "step": 3664
+    },
+    {
+      "epoch": 6.97,
+      "grad_norm": 0.2320442795753479,
+      "learning_rate": 0.00013342493930423822,
+      "loss": 0.016,
+      "step": 3665
+    },
+    {
+      "epoch": 6.97,
+      "grad_norm": 0.17990632355213165,
+      "learning_rate": 0.00013327133641610064,
+      "loss": 0.0163,
+      "step": 3666
+    },
+    {
+      "epoch": 6.97,
+      "grad_norm": 0.27979937195777893,
+      "learning_rate": 0.0001331177967442659,
+      "loss": 0.0152,
+      "step": 3667
+    },
+    {
+      "epoch": 6.97,
+      "grad_norm": 0.17194388806819916,
+      "learning_rate": 0.0001329643203469495,
+      "loss": 0.0069,
+      "step": 3668
+    },
+    {
+      "epoch": 6.98,
+      "grad_norm": 0.33162721991539,
+      "learning_rate": 0.00013281090728234308,
+      "loss": 0.0189,
+      "step": 3669
+    },
+    {
+      "epoch": 6.98,
+      "grad_norm": 0.2851185202598572,
+      "learning_rate": 0.00013265755760861423,
+      "loss": 0.0102,
+      "step": 3670
+    },
+    {
+      "epoch": 6.98,
+      "grad_norm": 0.2643820345401764,
+      "learning_rate": 0.00013250427138390653,
+      "loss": 0.0097,
+      "step": 3671
+    },
+    {
+      "epoch": 6.98,
+      "grad_norm": 0.23040655255317688,
+      "learning_rate": 0.00013235104866633947,
+      "loss": 0.0109,
+      "step": 3672
+    },
+    {
+      "epoch": 6.98,
+      "grad_norm": 0.18906722962856293,
+      "learning_rate": 0.00013219788951400843,
+      "loss": 0.0051,
+      "step": 3673
+    },
+    {
+      "epoch": 6.98,
+      "grad_norm": 0.3312613070011139,
+      "learning_rate": 0.00013204479398498496,
+      "loss": 0.0193,
+      "step": 3674
+    },
+    {
+      "epoch": 6.99,
+      "grad_norm": 0.22128500044345856,
+      "learning_rate": 0.0001318917621373159,
+      "loss": 0.0137,
+      "step": 3675
+    },
+    {
+      "epoch": 6.99,
+      "grad_norm": 0.19628086686134338,
+      "learning_rate": 0.0001317387940290247,
+      "loss": 0.0102,
+      "step": 3676
+    },
+    {
+      "epoch": 6.99,
+      "grad_norm": 0.10708604753017426,
+      "learning_rate": 0.00013158588971811,
+      "loss": 0.0029,
+      "step": 3677
+    },
+    {
+      "epoch": 6.99,
+      "grad_norm": 0.16760829091072083,
+      "learning_rate": 0.00013143304926254662,
+      "loss": 0.007,
+      "step": 3678
+    },
+    {
+      "epoch": 6.99,
+      "grad_norm": 0.23991824686527252,
+      "learning_rate": 0.00013128027272028505,
+      "loss": 0.013,
+      "step": 3679
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 0.24200288951396942,
+      "learning_rate": 0.00013112756014925145,
+      "loss": 0.0127,
+      "step": 3680
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 0.30327215790748596,
+      "learning_rate": 0.00013097491160734794,
+      "loss": 0.0118,
+      "step": 3681
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 0.24100050330162048,
+      "learning_rate": 0.00013082232715245224,
+      "loss": 0.0121,
+      "step": 3682
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 0.22211432456970215,
+      "learning_rate": 0.0001306698068424177,
+      "loss": 0.0089,
+      "step": 3683
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 0.0948360413312912,
+      "learning_rate": 0.0001305173507350734,
+      "loss": 0.0034,
+      "step": 3684
+    },
+    {
+      "epoch": 7.01,
+      "grad_norm": 0.17018717527389526,
+      "learning_rate": 0.00013036495888822436,
+      "loss": 0.008,
+      "step": 3685
+    },
+    {
+      "epoch": 7.01,
+      "grad_norm": 0.08899276703596115,
+      "learning_rate": 0.00013021263135965062,
+      "loss": 0.0032,
+      "step": 3686
+    },
+    {
+      "epoch": 7.01,
+      "grad_norm": 0.14672411978244781,
+      "learning_rate": 0.00013006036820710845,
+      "loss": 0.0072,
+      "step": 3687
+    },
+    {
+      "epoch": 7.01,
+      "grad_norm": 0.26800772547721863,
+      "learning_rate": 0.00012990816948832944,
+      "loss": 0.0109,
+      "step": 3688
+    },
+    {
+      "epoch": 7.01,
+      "grad_norm": 0.10023466497659683,
+      "learning_rate": 0.00012975603526102066,
+      "loss": 0.0022,
+      "step": 3689
+    },
+    {
+      "epoch": 7.02,
+      "grad_norm": 0.25695231556892395,
+      "learning_rate": 0.00012960396558286483,
+      "loss": 0.0106,
+      "step": 3690
+    },
+    {
+      "epoch": 7.02,
+      "grad_norm": 0.19227997958660126,
+      "learning_rate": 0.0001294519605115204,
+      "loss": 0.0077,
+      "step": 3691
+    },
+    {
+      "epoch": 7.02,
+      "grad_norm": 0.20034730434417725,
+      "learning_rate": 0.0001293000201046208,
+      "loss": 0.0062,
+      "step": 3692
+    },
+    {
+      "epoch": 7.02,
+      "grad_norm": 0.1226910874247551,
+      "learning_rate": 0.00012914814441977558,
+      "loss": 0.0036,
+      "step": 3693
+    },
+    {
+      "epoch": 7.02,
+      "grad_norm": 0.09757458418607712,
+      "learning_rate": 0.00012899633351456924,
+      "loss": 0.0044,
+      "step": 3694
+    },
+    {
+      "epoch": 7.02,
+      "grad_norm": 0.12404948472976685,
+      "learning_rate": 0.00012884458744656202,
+      "loss": 0.0064,
+      "step": 3695
+    },
+    {
+      "epoch": 7.03,
+      "grad_norm": 0.22362014651298523,
+      "learning_rate": 0.00012869290627328937,
+      "loss": 0.0074,
+      "step": 3696
+    },
+    {
+      "epoch": 7.03,
+      "grad_norm": 0.1148075982928276,
+      "learning_rate": 0.0001285412900522623,
+      "loss": 0.0064,
+      "step": 3697
+    },
+    {
+      "epoch": 7.03,
+      "grad_norm": 0.14441058039665222,
+      "learning_rate": 0.00012838973884096713,
+      "loss": 0.0024,
+      "step": 3698
+    },
+    {
+      "epoch": 7.03,
+      "grad_norm": 0.027441008016467094,
+      "learning_rate": 0.00012823825269686541,
+      "loss": 0.001,
+      "step": 3699
+    },
+    {
+      "epoch": 7.03,
+      "grad_norm": 0.18166694045066833,
+      "learning_rate": 0.00012808683167739443,
+      "loss": 0.0027,
+      "step": 3700
+    },
+    {
+      "epoch": 7.03,
+      "eval_blimp_filtered_avg": 0.5401492537313433,
+      "eval_blimp_filtered_std": 0.0058629567338664675,
+      "step": 3700
+    },
+    {
+      "epoch": 7.03,
+      "eval_blimp_supplement_avg": 0.5517241379310345,
+      "eval_blimp_supplement_std": 0.023168322382271546,
+      "step": 3700
+    },
+    {
+      "epoch": 7.03,
+      "eval_vqa_filtered_avg": 0.15,
+      "eval_vqa_filtered_std": 0.03588702812826371,
+      "step": 3700
+    },
+    {
+      "epoch": 7.03,
+      "eval_winoground_filtered_avg": 0.52,
+      "eval_winoground_filtered_std": 0.05021167315686779,
+      "step": 3700
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 0.1965988278388977,
+      "learning_rate": 0.00012793547583996612,
+      "loss": 0.0131,
+      "step": 3701
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 0.18323926627635956,
+      "learning_rate": 0.00012778418524196834,
+      "loss": 0.0077,
+      "step": 3702
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 0.04890342801809311,
+      "learning_rate": 0.00012763295994076384,
+      "loss": 0.0011,
+      "step": 3703
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 0.09236554056406021,
+      "learning_rate": 0.00012748179999369077,
+      "loss": 0.0037,
+      "step": 3704
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 0.1728556752204895,
+      "learning_rate": 0.00012733070545806243,
+      "loss": 0.0071,
+      "step": 3705
+    },
+    {
+      "epoch": 7.05,
+      "grad_norm": 0.1999249905347824,
+      "learning_rate": 0.0001271796763911673,
+      "loss": 0.0055,
+      "step": 3706
+    },
+    {
+      "epoch": 7.05,
+      "grad_norm": 0.24007059633731842,
+      "learning_rate": 0.00012702871285026913,
+      "loss": 0.0111,
+      "step": 3707
+    },
+    {
+      "epoch": 7.05,
+      "grad_norm": 0.22470150887966156,
+      "learning_rate": 0.00012687781489260672,
+      "loss": 0.0066,
+      "step": 3708
+    },
+    {
+      "epoch": 7.05,
+      "grad_norm": 0.13741044700145721,
+      "learning_rate": 0.0001267269825753941,
+      "loss": 0.0031,
+      "step": 3709
+    },
+    {
+      "epoch": 7.05,
+      "grad_norm": 0.047964438796043396,
+      "learning_rate": 0.00012657621595582016,
+      "loss": 0.0013,
+      "step": 3710
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 0.2642070949077606,
+      "learning_rate": 0.00012642551509104954,
+      "loss": 0.0082,
+      "step": 3711
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 0.1917313188314438,
+      "learning_rate": 0.00012627488003822098,
+      "loss": 0.0154,
+      "step": 3712
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 0.12957455217838287,
+      "learning_rate": 0.00012612431085444912,
+      "loss": 0.0067,
+      "step": 3713
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 0.2129407525062561,
+      "learning_rate": 0.00012597380759682325,
+      "loss": 0.0081,
+      "step": 3714
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 0.16917093098163605,
+      "learning_rate": 0.00012582337032240754,
+      "loss": 0.0145,
+      "step": 3715
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 0.1693619191646576,
+      "learning_rate": 0.00012567299908824146,
+      "loss": 0.0057,
+      "step": 3716
+    },
+    {
+      "epoch": 7.07,
+      "grad_norm": 0.070194773375988,
+      "learning_rate": 0.00012552269395133922,
+      "loss": 0.0023,
+      "step": 3717
+    },
+    {
+      "epoch": 7.07,
+      "grad_norm": 0.3151678144931793,
+      "learning_rate": 0.00012537245496869,
+      "loss": 0.0203,
+      "step": 3718
+    },
+    {
+      "epoch": 7.07,
+      "grad_norm": 0.06825432926416397,
+      "learning_rate": 0.00012522228219725789,
+      "loss": 0.002,
+      "step": 3719
+    },
+    {
+      "epoch": 7.07,
+      "grad_norm": 0.27343693375587463,
+      "learning_rate": 0.00012507217569398217,
+      "loss": 0.0132,
+      "step": 3720
+    },
+    {
+      "epoch": 7.07,
+      "grad_norm": 0.1460888683795929,
+      "learning_rate": 0.00012492213551577633,
+      "loss": 0.0081,
+      "step": 3721
+    },
+    {
+      "epoch": 7.08,
+      "grad_norm": 0.07731921225786209,
+      "learning_rate": 0.0001247721617195295,
+      "loss": 0.0027,
+      "step": 3722
+    },
+    {
+      "epoch": 7.08,
+      "grad_norm": 0.15644392371177673,
+      "learning_rate": 0.00012462225436210504,
+      "loss": 0.0101,
+      "step": 3723
+    },
+    {
+      "epoch": 7.08,
+      "grad_norm": 0.23197180032730103,
+      "learning_rate": 0.00012447241350034145,
+      "loss": 0.0118,
+      "step": 3724
+    },
+    {
+      "epoch": 7.08,
+      "grad_norm": 0.2023295760154724,
+      "learning_rate": 0.00012432263919105185,
+      "loss": 0.0078,
+      "step": 3725
+    },
+    {
+      "epoch": 7.08,
+      "grad_norm": 0.12722396850585938,
+      "learning_rate": 0.0001241729314910242,
+      "loss": 0.005,
+      "step": 3726
+    },
+    {
+      "epoch": 7.09,
+      "grad_norm": 0.19655947387218475,
+      "learning_rate": 0.00012402329045702116,
+      "loss": 0.0037,
+      "step": 3727
+    },
+    {
+      "epoch": 7.09,
+      "grad_norm": 0.19688691198825836,
+      "learning_rate": 0.0001238737161457802,
+      "loss": 0.016,
+      "step": 3728
+    },
+    {
+      "epoch": 7.09,
+      "grad_norm": 0.11575788259506226,
+      "learning_rate": 0.0001237242086140134,
+      "loss": 0.0054,
+      "step": 3729
+    },
+    {
+      "epoch": 7.09,
+      "grad_norm": 0.09897477924823761,
+      "learning_rate": 0.00012357476791840746,
+      "loss": 0.0021,
+      "step": 3730
+    },
+    {
+      "epoch": 7.09,
+      "grad_norm": 0.20128971338272095,
+      "learning_rate": 0.00012342539411562412,
+      "loss": 0.0144,
+      "step": 3731
+    },
+    {
+      "epoch": 7.1,
+      "grad_norm": 0.13159237802028656,
+      "learning_rate": 0.00012327608726229907,
+      "loss": 0.0079,
+      "step": 3732
+    },
+    {
+      "epoch": 7.1,
+      "grad_norm": 0.18053287267684937,
+      "learning_rate": 0.00012312684741504332,
+      "loss": 0.0137,
+      "step": 3733
+    },
+    {
+      "epoch": 7.1,
+      "grad_norm": 0.041263386607170105,
+      "learning_rate": 0.00012297767463044208,
+      "loss": 0.0015,
+      "step": 3734
+    },
+    {
+      "epoch": 7.1,
+      "grad_norm": 0.08285720646381378,
+      "learning_rate": 0.0001228285689650552,
+      "loss": 0.0028,
+      "step": 3735
+    },
+    {
+      "epoch": 7.1,
+      "grad_norm": 0.1426668018102646,
+      "learning_rate": 0.00012267953047541706,
+      "loss": 0.0042,
+      "step": 3736
+    },
+    {
+      "epoch": 7.1,
+      "grad_norm": 0.11019343137741089,
+      "learning_rate": 0.00012253055921803682,
+      "loss": 0.0021,
+      "step": 3737
+    },
+    {
+      "epoch": 7.11,
+      "grad_norm": 0.14096446335315704,
+      "learning_rate": 0.00012238165524939765,
+      "loss": 0.0101,
+      "step": 3738
+    },
+    {
+      "epoch": 7.11,
+      "grad_norm": 0.10174082964658737,
+      "learning_rate": 0.00012223281862595772,
+      "loss": 0.0038,
+      "step": 3739
+    },
+    {
+      "epoch": 7.11,
+      "grad_norm": 0.10789841413497925,
+      "learning_rate": 0.00012208404940414937,
+      "loss": 0.0039,
+      "step": 3740
+    },
+    {
+      "epoch": 7.11,
+      "grad_norm": 0.17103004455566406,
+      "learning_rate": 0.00012193534764037943,
+      "loss": 0.0093,
+      "step": 3741
+    },
+    {
+      "epoch": 7.11,
+      "grad_norm": 0.13981063663959503,
+      "learning_rate": 0.00012178671339102922,
+      "loss": 0.0109,
+      "step": 3742
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 0.09751623123884201,
+      "learning_rate": 0.00012163814671245442,
+      "loss": 0.0033,
+      "step": 3743
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 0.10560385882854462,
+      "learning_rate": 0.0001214896476609851,
+      "loss": 0.0045,
+      "step": 3744
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 0.08571907132863998,
+      "learning_rate": 0.00012134121629292552,
+      "loss": 0.0032,
+      "step": 3745
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 0.05475155636668205,
+      "learning_rate": 0.00012119285266455484,
+      "loss": 0.0016,
+      "step": 3746
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 0.04877667501568794,
+      "learning_rate": 0.00012104455683212564,
+      "loss": 0.0015,
+      "step": 3747
+    },
+    {
+      "epoch": 7.13,
+      "grad_norm": 0.09516564756631851,
+      "learning_rate": 0.00012089632885186564,
+      "loss": 0.0038,
+      "step": 3748
+    },
+    {
+      "epoch": 7.13,
+      "grad_norm": 0.22852498292922974,
+      "learning_rate": 0.00012074816877997637,
+      "loss": 0.0066,
+      "step": 3749
+    },
+    {
+      "epoch": 7.13,
+      "grad_norm": 0.11852756142616272,
+      "learning_rate": 0.00012060007667263374,
+      "loss": 0.0026,
+      "step": 3750
+    },
+    {
+      "epoch": 7.13,
+      "grad_norm": 0.13732858002185822,
+      "learning_rate": 0.00012045205258598786,
+      "loss": 0.0069,
+      "step": 3751
+    },
+    {
+      "epoch": 7.13,
+      "grad_norm": 0.20002444088459015,
+      "learning_rate": 0.00012030409657616306,
+      "loss": 0.0106,
+      "step": 3752
+    },
+    {
+      "epoch": 7.13,
+      "grad_norm": 0.1415424942970276,
+      "learning_rate": 0.0001201562086992579,
+      "loss": 0.0025,
+      "step": 3753
+    },
+    {
+      "epoch": 7.14,
+      "grad_norm": 0.16294683516025543,
+      "learning_rate": 0.00012000838901134502,
+      "loss": 0.0042,
+      "step": 3754
+    },
+    {
+      "epoch": 7.14,
+      "grad_norm": 0.27604150772094727,
+      "learning_rate": 0.00011986063756847129,
+      "loss": 0.0081,
+      "step": 3755
+    },
+    {
+      "epoch": 7.14,
+      "grad_norm": 0.19468402862548828,
+      "learning_rate": 0.00011971295442665755,
+      "loss": 0.0085,
+      "step": 3756
+    },
+    {
+      "epoch": 7.14,
+      "grad_norm": 0.2314595878124237,
+      "learning_rate": 0.00011956533964189918,
+      "loss": 0.0097,
+      "step": 3757
+    },
+    {
+      "epoch": 7.14,
+      "grad_norm": 0.16542653739452362,
+      "learning_rate": 0.00011941779327016493,
+      "loss": 0.0049,
+      "step": 3758
+    },
+    {
+      "epoch": 7.15,
+      "grad_norm": 0.07420577108860016,
+      "learning_rate": 0.00011927031536739831,
+      "loss": 0.0032,
+      "step": 3759
+    },
+    {
+      "epoch": 7.15,
+      "grad_norm": 0.1154956966638565,
+      "learning_rate": 0.00011912290598951644,
+      "loss": 0.0042,
+      "step": 3760
+    },
+    {
+      "epoch": 7.15,
+      "grad_norm": 0.16177032887935638,
+      "learning_rate": 0.00011897556519241062,
+      "loss": 0.0067,
+      "step": 3761
+    },
+    {
+      "epoch": 7.15,
+      "grad_norm": 0.0334647111594677,
+      "learning_rate": 0.00011882829303194605,
+      "loss": 0.0013,
+      "step": 3762
+    },
+    {
+      "epoch": 7.15,
+      "grad_norm": 0.14544470608234406,
+      "learning_rate": 0.00011868108956396207,
+      "loss": 0.0082,
+      "step": 3763
+    },
+    {
+      "epoch": 7.16,
+      "grad_norm": 0.15135306119918823,
+      "learning_rate": 0.00011853395484427176,
+      "loss": 0.0067,
+      "step": 3764
+    },
+    {
+      "epoch": 7.16,
+      "grad_norm": 0.13548751175403595,
+      "learning_rate": 0.0001183868889286622,
+      "loss": 0.0066,
+      "step": 3765
+    },
+    {
+      "epoch": 7.16,
+      "grad_norm": 0.08642897009849548,
+      "learning_rate": 0.00011823989187289476,
+      "loss": 0.0021,
+      "step": 3766
+    },
+    {
+      "epoch": 7.16,
+      "grad_norm": 0.05003029480576515,
+      "learning_rate": 0.00011809296373270389,
+      "loss": 0.0015,
+      "step": 3767
+    },
+    {
+      "epoch": 7.16,
+      "grad_norm": 0.09651600569486618,
+      "learning_rate": 0.00011794610456379878,
+      "loss": 0.0035,
+      "step": 3768
+    },
+    {
+      "epoch": 7.17,
+      "grad_norm": 0.06292024999856949,
+      "learning_rate": 0.00011779931442186191,
+      "loss": 0.0021,
+      "step": 3769
+    },
+    {
+      "epoch": 7.17,
+      "grad_norm": 0.10055705159902573,
+      "learning_rate": 0.00011765259336254983,
+      "loss": 0.0025,
+      "step": 3770
+    },
+    {
+      "epoch": 7.17,
+      "grad_norm": 0.08155447244644165,
+      "learning_rate": 0.00011750594144149274,
+      "loss": 0.0019,
+      "step": 3771
+    },
+    {
+      "epoch": 7.17,
+      "grad_norm": 0.09257533401250839,
+      "learning_rate": 0.00011735935871429474,
+      "loss": 0.0021,
+      "step": 3772
+    },
+    {
+      "epoch": 7.17,
+      "grad_norm": 0.05412004515528679,
+      "learning_rate": 0.00011721284523653368,
+      "loss": 0.0013,
+      "step": 3773
+    },
+    {
+      "epoch": 7.17,
+      "grad_norm": 0.13410142064094543,
+      "learning_rate": 0.00011706640106376103,
+      "loss": 0.0024,
+      "step": 3774
+    },
+    {
+      "epoch": 7.18,
+      "grad_norm": 0.07152893394231796,
+      "learning_rate": 0.00011692002625150238,
+      "loss": 0.0015,
+      "step": 3775
+    },
+    {
+      "epoch": 7.18,
+      "grad_norm": 0.04963964968919754,
+      "learning_rate": 0.00011677372085525636,
+      "loss": 0.0022,
+      "step": 3776
+    },
+    {
+      "epoch": 7.18,
+      "grad_norm": 0.18686196208000183,
+      "learning_rate": 0.00011662748493049603,
+      "loss": 0.0065,
+      "step": 3777
+    },
+    {
+      "epoch": 7.18,
+      "grad_norm": 0.09778320044279099,
+      "learning_rate": 0.00011648131853266734,
+      "loss": 0.0039,
+      "step": 3778
+    },
+    {
+      "epoch": 7.18,
+      "grad_norm": 0.11319936066865921,
+      "learning_rate": 0.00011633522171719054,
+      "loss": 0.0025,
+      "step": 3779
+    },
+    {
+      "epoch": 7.19,
+      "grad_norm": 0.19281665980815887,
+      "learning_rate": 0.00011618919453945919,
+      "loss": 0.0076,
+      "step": 3780
+    },
+    {
+      "epoch": 7.19,
+      "grad_norm": 0.15041005611419678,
+      "learning_rate": 0.00011604323705484041,
+      "loss": 0.0126,
+      "step": 3781
+    },
+    {
+      "epoch": 7.19,
+      "grad_norm": 0.23571383953094482,
+      "learning_rate": 0.00011589734931867494,
+      "loss": 0.0113,
+      "step": 3782
+    },
+    {
+      "epoch": 7.19,
+      "grad_norm": 0.09244998544454575,
+      "learning_rate": 0.00011575153138627738,
+      "loss": 0.0025,
+      "step": 3783
+    },
+    {
+      "epoch": 7.19,
+      "grad_norm": 0.08689813315868378,
+      "learning_rate": 0.00011560578331293516,
+      "loss": 0.0028,
+      "step": 3784
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 0.12533530592918396,
+      "learning_rate": 0.00011546010515390998,
+      "loss": 0.0105,
+      "step": 3785
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 0.2308184653520584,
+      "learning_rate": 0.00011531449696443658,
+      "loss": 0.0106,
+      "step": 3786
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 0.1814446747303009,
+      "learning_rate": 0.0001151689587997233,
+      "loss": 0.024,
+      "step": 3787
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 0.1285831779241562,
+      "learning_rate": 0.00011502349071495191,
+      "loss": 0.0038,
+      "step": 3788
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 0.1639883667230606,
+      "learning_rate": 0.00011487809276527767,
+      "loss": 0.0077,
+      "step": 3789
+    },
+    {
+      "epoch": 7.21,
+      "grad_norm": 0.1021813452243805,
+      "learning_rate": 0.00011473276500582914,
+      "loss": 0.0064,
+      "step": 3790
+    },
+    {
+      "epoch": 7.21,
+      "grad_norm": 0.201278954744339,
+      "learning_rate": 0.00011458750749170824,
+      "loss": 0.0043,
+      "step": 3791
+    },
+    {
+      "epoch": 7.21,
+      "grad_norm": 0.3452257812023163,
+      "learning_rate": 0.00011444232027799065,
+      "loss": 0.0223,
+      "step": 3792
+    },
+    {
+      "epoch": 7.21,
+      "grad_norm": 0.18365712463855743,
+      "learning_rate": 0.0001142972034197247,
+      "loss": 0.0053,
+      "step": 3793
+    },
+    {
+      "epoch": 7.21,
+      "grad_norm": 0.10285883396863937,
+      "learning_rate": 0.00011415215697193271,
+      "loss": 0.0019,
+      "step": 3794
+    },
+    {
+      "epoch": 7.21,
+      "grad_norm": 0.13758790493011475,
+      "learning_rate": 0.00011400718098960993,
+      "loss": 0.0041,
+      "step": 3795
+    },
+    {
+      "epoch": 7.22,
+      "grad_norm": 0.15340152382850647,
+      "learning_rate": 0.00011386227552772499,
+      "loss": 0.0082,
+      "step": 3796
+    },
+    {
+      "epoch": 7.22,
+      "grad_norm": 0.09982304275035858,
+      "learning_rate": 0.00011371744064121974,
+      "loss": 0.0022,
+      "step": 3797
+    },
+    {
+      "epoch": 7.22,
+      "grad_norm": 0.17092829942703247,
+      "learning_rate": 0.00011357267638500935,
+      "loss": 0.0045,
+      "step": 3798
+    },
+    {
+      "epoch": 7.22,
+      "grad_norm": 0.12320729345083237,
+      "learning_rate": 0.0001134279828139822,
+      "loss": 0.0025,
+      "step": 3799
+    },
+    {
+      "epoch": 7.22,
+      "grad_norm": 0.14494533836841583,
+      "learning_rate": 0.00011328335998299963,
+      "loss": 0.0076,
+      "step": 3800
+    },
+    {
+      "epoch": 7.22,
+      "eval_blimp_filtered_avg": 0.5461194029850747,
+      "eval_blimp_filtered_std": 0.005840885397322259,
+      "step": 3800
+    },
+    {
+      "epoch": 7.22,
+      "eval_blimp_supplement_avg": 0.5431034482758621,
+      "eval_blimp_supplement_std": 0.023123693509581343,
+      "step": 3800
+    },
+    {
+      "epoch": 7.22,
+      "eval_vqa_filtered_avg": 0.12,
+      "eval_vqa_filtered_std": 0.03265986323710906,
+      "step": 3800
+    },
+    {
+      "epoch": 7.22,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 3800
+    },
+    {
+      "epoch": 7.23,
+      "grad_norm": 0.074303537607193,
+      "learning_rate": 0.00011313880794689679,
+      "loss": 0.0017,
+      "step": 3801
+    },
+    {
+      "epoch": 7.23,
+      "grad_norm": 0.13375483453273773,
+      "learning_rate": 0.0001129943267604811,
+      "loss": 0.0059,
+      "step": 3802
+    },
+    {
+      "epoch": 7.23,
+      "grad_norm": 0.11006172001361847,
+      "learning_rate": 0.00011284991647853398,
+      "loss": 0.0027,
+      "step": 3803
+    },
+    {
+      "epoch": 7.23,
+      "grad_norm": 0.20781584084033966,
+      "learning_rate": 0.00011270557715580917,
+      "loss": 0.0057,
+      "step": 3804
+    },
+    {
+      "epoch": 7.23,
+      "grad_norm": 0.17031773924827576,
+      "learning_rate": 0.0001125613088470342,
+      "loss": 0.0044,
+      "step": 3805
+    },
+    {
+      "epoch": 7.24,
+      "grad_norm": 0.10886078327894211,
+      "learning_rate": 0.00011241711160690933,
+      "loss": 0.0049,
+      "step": 3806
+    },
+    {
+      "epoch": 7.24,
+      "grad_norm": 0.10179899632930756,
+      "learning_rate": 0.00011227298549010781,
+      "loss": 0.0044,
+      "step": 3807
+    },
+    {
+      "epoch": 7.24,
+      "grad_norm": 0.1269051432609558,
+      "learning_rate": 0.00011212893055127615,
+      "loss": 0.0024,
+      "step": 3808
+    },
+    {
+      "epoch": 7.24,
+      "grad_norm": 0.12605036795139313,
+      "learning_rate": 0.00011198494684503368,
+      "loss": 0.0058,
+      "step": 3809
+    },
+    {
+      "epoch": 7.24,
+      "grad_norm": 0.16540399193763733,
+      "learning_rate": 0.00011184103442597287,
+      "loss": 0.0039,
+      "step": 3810
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 0.1720535308122635,
+      "learning_rate": 0.00011169719334865891,
+      "loss": 0.005,
+      "step": 3811
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 0.16166216135025024,
+      "learning_rate": 0.00011155342366763046,
+      "loss": 0.0057,
+      "step": 3812
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 0.13763488829135895,
+      "learning_rate": 0.00011140972543739836,
+      "loss": 0.0061,
+      "step": 3813
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 0.14980502426624298,
+      "learning_rate": 0.00011126609871244705,
+      "loss": 0.0055,
+      "step": 3814
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 0.08673214167356491,
+      "learning_rate": 0.0001111225435472335,
+      "loss": 0.0028,
+      "step": 3815
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 0.02181820385158062,
+      "learning_rate": 0.00011097905999618763,
+      "loss": 0.0006,
+      "step": 3816
+    },
+    {
+      "epoch": 7.26,
+      "grad_norm": 0.08254721760749817,
+      "learning_rate": 0.0001108356481137122,
+      "loss": 0.0038,
+      "step": 3817
+    },
+    {
+      "epoch": 7.26,
+      "grad_norm": 0.13598065078258514,
+      "learning_rate": 0.00011069230795418279,
+      "loss": 0.0031,
+      "step": 3818
+    },
+    {
+      "epoch": 7.26,
+      "grad_norm": 0.1120840534567833,
+      "learning_rate": 0.00011054903957194778,
+      "loss": 0.0054,
+      "step": 3819
+    },
+    {
+      "epoch": 7.26,
+      "grad_norm": 0.11079658567905426,
+      "learning_rate": 0.00011040584302132827,
+      "loss": 0.0061,
+      "step": 3820
+    },
+    {
+      "epoch": 7.26,
+      "grad_norm": 0.11221101880073547,
+      "learning_rate": 0.0001102627183566185,
+      "loss": 0.0039,
+      "step": 3821
+    },
+    {
+      "epoch": 7.27,
+      "grad_norm": 0.12699200212955475,
+      "learning_rate": 0.00011011966563208474,
+      "loss": 0.0033,
+      "step": 3822
+    },
+    {
+      "epoch": 7.27,
+      "grad_norm": 0.1776549369096756,
+      "learning_rate": 0.00010997668490196669,
+      "loss": 0.0073,
+      "step": 3823
+    },
+    {
+      "epoch": 7.27,
+      "grad_norm": 0.2175322324037552,
+      "learning_rate": 0.0001098337762204764,
+      "loss": 0.0118,
+      "step": 3824
+    },
+    {
+      "epoch": 7.27,
+      "grad_norm": 0.10095300525426865,
+      "learning_rate": 0.00010969093964179863,
+      "loss": 0.0017,
+      "step": 3825
+    },
+    {
+      "epoch": 7.27,
+      "grad_norm": 0.10306946933269501,
+      "learning_rate": 0.00010954817522009084,
+      "loss": 0.0022,
+      "step": 3826
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 0.12816134095191956,
+      "learning_rate": 0.00010940548300948317,
+      "loss": 0.004,
+      "step": 3827
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 0.16520491242408752,
+      "learning_rate": 0.0001092628630640782,
+      "loss": 0.0076,
+      "step": 3828
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 0.17636096477508545,
+      "learning_rate": 0.00010912031543795156,
+      "loss": 0.0041,
+      "step": 3829
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 0.1191505640745163,
+      "learning_rate": 0.00010897784018515074,
+      "loss": 0.005,
+      "step": 3830
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 0.04538269713521004,
+      "learning_rate": 0.00010883543735969652,
+      "loss": 0.0015,
+      "step": 3831
+    },
+    {
+      "epoch": 7.29,
+      "grad_norm": 0.14366485178470612,
+      "learning_rate": 0.00010869310701558179,
+      "loss": 0.0034,
+      "step": 3832
+    },
+    {
+      "epoch": 7.29,
+      "grad_norm": 0.13167054951190948,
+      "learning_rate": 0.00010855084920677202,
+      "loss": 0.0047,
+      "step": 3833
+    },
+    {
+      "epoch": 7.29,
+      "grad_norm": 0.09079673141241074,
+      "learning_rate": 0.00010840866398720528,
+      "loss": 0.0048,
+      "step": 3834
+    },
+    {
+      "epoch": 7.29,
+      "grad_norm": 0.05846530571579933,
+      "learning_rate": 0.00010826655141079202,
+      "loss": 0.0012,
+      "step": 3835
+    },
+    {
+      "epoch": 7.29,
+      "grad_norm": 0.1979508101940155,
+      "learning_rate": 0.00010812451153141523,
+      "loss": 0.0119,
+      "step": 3836
+    },
+    {
+      "epoch": 7.29,
+      "grad_norm": 0.1391127109527588,
+      "learning_rate": 0.00010798254440293018,
+      "loss": 0.0024,
+      "step": 3837
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 0.14590077102184296,
+      "learning_rate": 0.00010784065007916491,
+      "loss": 0.0035,
+      "step": 3838
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 0.0589585117995739,
+      "learning_rate": 0.00010769882861391928,
+      "loss": 0.0014,
+      "step": 3839
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 0.07086075842380524,
+      "learning_rate": 0.00010755708006096614,
+      "loss": 0.002,
+      "step": 3840
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 0.14349332451820374,
+      "learning_rate": 0.00010741540447405031,
+      "loss": 0.0051,
+      "step": 3841
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 0.06697641313076019,
+      "learning_rate": 0.00010727380190688908,
+      "loss": 0.002,
+      "step": 3842
+    },
+    {
+      "epoch": 7.31,
+      "grad_norm": 0.11811905354261398,
+      "learning_rate": 0.00010713227241317195,
+      "loss": 0.0028,
+      "step": 3843
+    },
+    {
+      "epoch": 7.31,
+      "grad_norm": 0.09835381805896759,
+      "learning_rate": 0.00010699081604656085,
+      "loss": 0.004,
+      "step": 3844
+    },
+    {
+      "epoch": 7.31,
+      "grad_norm": 0.233662948012352,
+      "learning_rate": 0.00010684943286068993,
+      "loss": 0.0082,
+      "step": 3845
+    },
+    {
+      "epoch": 7.31,
+      "grad_norm": 0.10110045969486237,
+      "learning_rate": 0.00010670812290916543,
+      "loss": 0.0055,
+      "step": 3846
+    },
+    {
+      "epoch": 7.31,
+      "grad_norm": 0.2191174030303955,
+      "learning_rate": 0.00010656688624556632,
+      "loss": 0.0086,
+      "step": 3847
+    },
+    {
+      "epoch": 7.32,
+      "grad_norm": 0.10092798620462418,
+      "learning_rate": 0.000106425722923443,
+      "loss": 0.0052,
+      "step": 3848
+    },
+    {
+      "epoch": 7.32,
+      "grad_norm": 0.10454420000314713,
+      "learning_rate": 0.00010628463299631883,
+      "loss": 0.0049,
+      "step": 3849
+    },
+    {
+      "epoch": 7.32,
+      "grad_norm": 0.0883038192987442,
+      "learning_rate": 0.00010614361651768891,
+      "loss": 0.0045,
+      "step": 3850
+    },
+    {
+      "epoch": 7.32,
+      "grad_norm": 0.10077980905771255,
+      "learning_rate": 0.00010600267354102058,
+      "loss": 0.0022,
+      "step": 3851
+    },
+    {
+      "epoch": 7.32,
+      "grad_norm": 0.1326596438884735,
+      "learning_rate": 0.00010586180411975335,
+      "loss": 0.0044,
+      "step": 3852
+    },
+    {
+      "epoch": 7.33,
+      "grad_norm": 0.10058380663394928,
+      "learning_rate": 0.00010572100830729876,
+      "loss": 0.0035,
+      "step": 3853
+    },
+    {
+      "epoch": 7.33,
+      "grad_norm": 0.13691933453083038,
+      "learning_rate": 0.00010558028615704056,
+      "loss": 0.02,
+      "step": 3854
+    },
+    {
+      "epoch": 7.33,
+      "grad_norm": 0.17982117831707,
+      "learning_rate": 0.00010543963772233447,
+      "loss": 0.0106,
+      "step": 3855
+    },
+    {
+      "epoch": 7.33,
+      "grad_norm": 0.19012513756752014,
+      "learning_rate": 0.00010529906305650832,
+      "loss": 0.0056,
+      "step": 3856
+    },
+    {
+      "epoch": 7.33,
+      "grad_norm": 0.21838918328285217,
+      "learning_rate": 0.00010515856221286182,
+      "loss": 0.0056,
+      "step": 3857
+    },
+    {
+      "epoch": 7.33,
+      "grad_norm": 0.2982530891895294,
+      "learning_rate": 0.00010501813524466714,
+      "loss": 0.0163,
+      "step": 3858
+    },
+    {
+      "epoch": 7.34,
+      "grad_norm": 0.11210102587938309,
+      "learning_rate": 0.00010487778220516768,
+      "loss": 0.0031,
+      "step": 3859
+    },
+    {
+      "epoch": 7.34,
+      "grad_norm": 0.12145328521728516,
+      "learning_rate": 0.00010473750314757963,
+      "loss": 0.0054,
+      "step": 3860
+    },
+    {
+      "epoch": 7.34,
+      "grad_norm": 0.17167417705059052,
+      "learning_rate": 0.00010459729812509056,
+      "loss": 0.0066,
+      "step": 3861
+    },
+    {
+      "epoch": 7.34,
+      "grad_norm": 0.17210602760314941,
+      "learning_rate": 0.00010445716719086018,
+      "loss": 0.0128,
+      "step": 3862
+    },
+    {
+      "epoch": 7.34,
+      "grad_norm": 0.13199707865715027,
+      "learning_rate": 0.0001043171103980201,
+      "loss": 0.0053,
+      "step": 3863
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 0.03279474750161171,
+      "learning_rate": 0.00010417712779967376,
+      "loss": 0.0008,
+      "step": 3864
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 0.04349233955144882,
+      "learning_rate": 0.00010403721944889656,
+      "loss": 0.0011,
+      "step": 3865
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 0.13710527122020721,
+      "learning_rate": 0.00010389738539873557,
+      "loss": 0.0053,
+      "step": 3866
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 0.13334079086780548,
+      "learning_rate": 0.00010375762570221014,
+      "loss": 0.0035,
+      "step": 3867
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 0.1720355749130249,
+      "learning_rate": 0.00010361794041231071,
+      "loss": 0.0044,
+      "step": 3868
+    },
+    {
+      "epoch": 7.36,
+      "grad_norm": 0.12716442346572876,
+      "learning_rate": 0.0001034783295820002,
+      "loss": 0.0027,
+      "step": 3869
+    },
+    {
+      "epoch": 7.36,
+      "grad_norm": 0.08692093938589096,
+      "learning_rate": 0.00010333879326421284,
+      "loss": 0.0052,
+      "step": 3870
+    },
+    {
+      "epoch": 7.36,
+      "grad_norm": 0.08804266899824142,
+      "learning_rate": 0.00010319933151185485,
+      "loss": 0.0021,
+      "step": 3871
+    },
+    {
+      "epoch": 7.36,
+      "grad_norm": 0.14963935315608978,
+      "learning_rate": 0.00010305994437780399,
+      "loss": 0.0037,
+      "step": 3872
+    },
+    {
+      "epoch": 7.36,
+      "grad_norm": 0.12284870445728302,
+      "learning_rate": 0.00010292063191491008,
+      "loss": 0.005,
+      "step": 3873
+    },
+    {
+      "epoch": 7.37,
+      "grad_norm": 0.24227400124073029,
+      "learning_rate": 0.00010278139417599401,
+      "loss": 0.0064,
+      "step": 3874
+    },
+    {
+      "epoch": 7.37,
+      "grad_norm": 0.24972118437290192,
+      "learning_rate": 0.00010264223121384916,
+      "loss": 0.0119,
+      "step": 3875
+    },
+    {
+      "epoch": 7.37,
+      "grad_norm": 0.15173201262950897,
+      "learning_rate": 0.00010250314308123966,
+      "loss": 0.0079,
+      "step": 3876
+    },
+    {
+      "epoch": 7.37,
+      "grad_norm": 0.07276877760887146,
+      "learning_rate": 0.00010236412983090204,
+      "loss": 0.002,
+      "step": 3877
+    },
+    {
+      "epoch": 7.37,
+      "grad_norm": 0.10876510292291641,
+      "learning_rate": 0.00010222519151554402,
+      "loss": 0.0021,
+      "step": 3878
+    },
+    {
+      "epoch": 7.37,
+      "grad_norm": 0.14648006856441498,
+      "learning_rate": 0.00010208632818784494,
+      "loss": 0.0052,
+      "step": 3879
+    },
+    {
+      "epoch": 7.38,
+      "grad_norm": 0.20102213323116302,
+      "learning_rate": 0.00010194753990045588,
+      "loss": 0.0093,
+      "step": 3880
+    },
+    {
+      "epoch": 7.38,
+      "grad_norm": 0.1571972668170929,
+      "learning_rate": 0.00010180882670599928,
+      "loss": 0.0043,
+      "step": 3881
+    },
+    {
+      "epoch": 7.38,
+      "grad_norm": 0.20462632179260254,
+      "learning_rate": 0.00010167018865706922,
+      "loss": 0.0049,
+      "step": 3882
+    },
+    {
+      "epoch": 7.38,
+      "grad_norm": 0.12511806190013885,
+      "learning_rate": 0.00010153162580623111,
+      "loss": 0.0049,
+      "step": 3883
+    },
+    {
+      "epoch": 7.38,
+      "grad_norm": 0.1337905377149582,
+      "learning_rate": 0.00010139313820602236,
+      "loss": 0.0039,
+      "step": 3884
+    },
+    {
+      "epoch": 7.39,
+      "grad_norm": 0.13089951872825623,
+      "learning_rate": 0.00010125472590895108,
+      "loss": 0.0045,
+      "step": 3885
+    },
+    {
+      "epoch": 7.39,
+      "grad_norm": 0.14334940910339355,
+      "learning_rate": 0.00010111638896749747,
+      "loss": 0.0074,
+      "step": 3886
+    },
+    {
+      "epoch": 7.39,
+      "grad_norm": 0.05526348575949669,
+      "learning_rate": 0.00010097812743411284,
+      "loss": 0.0013,
+      "step": 3887
+    },
+    {
+      "epoch": 7.39,
+      "grad_norm": 0.11799751222133636,
+      "learning_rate": 0.00010083994136122001,
+      "loss": 0.0043,
+      "step": 3888
+    },
+    {
+      "epoch": 7.39,
+      "grad_norm": 0.11252140998840332,
+      "learning_rate": 0.0001007018308012132,
+      "loss": 0.0054,
+      "step": 3889
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 0.0837024375796318,
+      "learning_rate": 0.00010056379580645786,
+      "loss": 0.0021,
+      "step": 3890
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 0.09408940374851227,
+      "learning_rate": 0.00010042583642929098,
+      "loss": 0.0024,
+      "step": 3891
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 0.09226690977811813,
+      "learning_rate": 0.00010028795272202066,
+      "loss": 0.0024,
+      "step": 3892
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 0.04648909717798233,
+      "learning_rate": 0.00010015014473692676,
+      "loss": 0.0013,
+      "step": 3893
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 0.0862855538725853,
+      "learning_rate": 0.0001000124125262597,
+      "loss": 0.0021,
+      "step": 3894
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 0.13353952765464783,
+      "learning_rate": 9.987475614224186e-05,
+      "loss": 0.0023,
+      "step": 3895
+    },
+    {
+      "epoch": 7.41,
+      "grad_norm": 0.02874172292649746,
+      "learning_rate": 9.973717563706653e-05,
+      "loss": 0.0011,
+      "step": 3896
+    },
+    {
+      "epoch": 7.41,
+      "grad_norm": 0.079706571996212,
+      "learning_rate": 9.959967106289829e-05,
+      "loss": 0.0014,
+      "step": 3897
+    },
+    {
+      "epoch": 7.41,
+      "grad_norm": 0.08855663239955902,
+      "learning_rate": 9.946224247187289e-05,
+      "loss": 0.0028,
+      "step": 3898
+    },
+    {
+      "epoch": 7.41,
+      "grad_norm": 0.17291395366191864,
+      "learning_rate": 9.932488991609736e-05,
+      "loss": 0.0036,
+      "step": 3899
+    },
+    {
+      "epoch": 7.41,
+      "grad_norm": 0.08305663615465164,
+      "learning_rate": 9.918761344764979e-05,
+      "loss": 0.0047,
+      "step": 3900
+    },
+    {
+      "epoch": 7.41,
+      "eval_blimp_filtered_avg": 0.5425373134328358,
+      "eval_blimp_filtered_std": 0.00585143157728736,
+      "step": 3900
+    },
+    {
+      "epoch": 7.41,
+      "eval_blimp_supplement_avg": 0.5366379310344828,
+      "eval_blimp_supplement_std": 0.02311436382187363,
+      "step": 3900
+    },
+    {
+      "epoch": 7.41,
+      "eval_vqa_filtered_avg": 0.11,
+      "eval_vqa_filtered_std": 0.031446603773522035,
+      "step": 3900
+    },
+    {
+      "epoch": 7.41,
+      "eval_winoground_filtered_avg": 0.51,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 3900
+    },
+    {
+      "epoch": 7.42,
+      "grad_norm": 0.12025030702352524,
+      "learning_rate": 9.905041311857956e-05,
+      "loss": 0.0079,
+      "step": 3901
+    },
+    {
+      "epoch": 7.42,
+      "grad_norm": 0.0774415135383606,
+      "learning_rate": 9.891328898090704e-05,
+      "loss": 0.0038,
+      "step": 3902
+    },
+    {
+      "epoch": 7.42,
+      "grad_norm": 0.1087917611002922,
+      "learning_rate": 9.87762410866237e-05,
+      "loss": 0.0044,
+      "step": 3903
+    },
+    {
+      "epoch": 7.42,
+      "grad_norm": 0.06169675663113594,
+      "learning_rate": 9.863926948769251e-05,
+      "loss": 0.0024,
+      "step": 3904
+    },
+    {
+      "epoch": 7.42,
+      "grad_norm": 0.06911018490791321,
+      "learning_rate": 9.850237423604676e-05,
+      "loss": 0.0016,
+      "step": 3905
+    },
+    {
+      "epoch": 7.43,
+      "grad_norm": 0.06891800463199615,
+      "learning_rate": 9.836555538359146e-05,
+      "loss": 0.0018,
+      "step": 3906
+    },
+    {
+      "epoch": 7.43,
+      "grad_norm": 0.10246939212083817,
+      "learning_rate": 9.822881298220237e-05,
+      "loss": 0.0025,
+      "step": 3907
+    },
+    {
+      "epoch": 7.43,
+      "grad_norm": 0.07779446989297867,
+      "learning_rate": 9.80921470837263e-05,
+      "loss": 0.0026,
+      "step": 3908
+    },
+    {
+      "epoch": 7.43,
+      "grad_norm": 0.15437725186347961,
+      "learning_rate": 9.7955557739981e-05,
+      "loss": 0.0028,
+      "step": 3909
+    },
+    {
+      "epoch": 7.43,
+      "grad_norm": 0.10927714407444,
+      "learning_rate": 9.781904500275526e-05,
+      "loss": 0.0039,
+      "step": 3910
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 0.10706699639558792,
+      "learning_rate": 9.768260892380884e-05,
+      "loss": 0.0033,
+      "step": 3911
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 0.12842115759849548,
+      "learning_rate": 9.754624955487231e-05,
+      "loss": 0.0061,
+      "step": 3912
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 0.09449569880962372,
+      "learning_rate": 9.74099669476475e-05,
+      "loss": 0.0025,
+      "step": 3913
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 0.08930085599422455,
+      "learning_rate": 9.72737611538065e-05,
+      "loss": 0.0021,
+      "step": 3914
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 0.1557353436946869,
+      "learning_rate": 9.7137632224993e-05,
+      "loss": 0.0064,
+      "step": 3915
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 0.18276040256023407,
+      "learning_rate": 9.700158021282107e-05,
+      "loss": 0.0046,
+      "step": 3916
+    },
+    {
+      "epoch": 7.45,
+      "grad_norm": 0.14849288761615753,
+      "learning_rate": 9.686560516887576e-05,
+      "loss": 0.0036,
+      "step": 3917
+    },
+    {
+      "epoch": 7.45,
+      "grad_norm": 0.15910960733890533,
+      "learning_rate": 9.672970714471281e-05,
+      "loss": 0.004,
+      "step": 3918
+    },
+    {
+      "epoch": 7.45,
+      "grad_norm": 0.07738698273897171,
+      "learning_rate": 9.65938861918592e-05,
+      "loss": 0.003,
+      "step": 3919
+    },
+    {
+      "epoch": 7.45,
+      "grad_norm": 0.16044344007968903,
+      "learning_rate": 9.645814236181196e-05,
+      "loss": 0.0046,
+      "step": 3920
+    },
+    {
+      "epoch": 7.45,
+      "grad_norm": 0.17430368065834045,
+      "learning_rate": 9.632247570603959e-05,
+      "loss": 0.0121,
+      "step": 3921
+    },
+    {
+      "epoch": 7.46,
+      "grad_norm": 0.1101655513048172,
+      "learning_rate": 9.61868862759809e-05,
+      "loss": 0.0036,
+      "step": 3922
+    },
+    {
+      "epoch": 7.46,
+      "grad_norm": 0.1174202710390091,
+      "learning_rate": 9.605137412304555e-05,
+      "loss": 0.0056,
+      "step": 3923
+    },
+    {
+      "epoch": 7.46,
+      "grad_norm": 0.05459747835993767,
+      "learning_rate": 9.591593929861405e-05,
+      "loss": 0.0009,
+      "step": 3924
+    },
+    {
+      "epoch": 7.46,
+      "grad_norm": 0.059509698301553726,
+      "learning_rate": 9.578058185403702e-05,
+      "loss": 0.0018,
+      "step": 3925
+    },
+    {
+      "epoch": 7.46,
+      "grad_norm": 0.07616113871335983,
+      "learning_rate": 9.564530184063653e-05,
+      "loss": 0.0028,
+      "step": 3926
+    },
+    {
+      "epoch": 7.47,
+      "grad_norm": 0.12724652886390686,
+      "learning_rate": 9.551009930970479e-05,
+      "loss": 0.0021,
+      "step": 3927
+    },
+    {
+      "epoch": 7.47,
+      "grad_norm": 0.0856863409280777,
+      "learning_rate": 9.537497431250478e-05,
+      "loss": 0.0036,
+      "step": 3928
+    },
+    {
+      "epoch": 7.47,
+      "grad_norm": 0.07111367583274841,
+      "learning_rate": 9.523992690026993e-05,
+      "loss": 0.0015,
+      "step": 3929
+    },
+    {
+      "epoch": 7.47,
+      "grad_norm": 0.12479518353939056,
+      "learning_rate": 9.51049571242047e-05,
+      "loss": 0.0065,
+      "step": 3930
+    },
+    {
+      "epoch": 7.47,
+      "grad_norm": 0.19258439540863037,
+      "learning_rate": 9.497006503548335e-05,
+      "loss": 0.0033,
+      "step": 3931
+    },
+    {
+      "epoch": 7.48,
+      "grad_norm": 0.10836288332939148,
+      "learning_rate": 9.48352506852515e-05,
+      "loss": 0.0037,
+      "step": 3932
+    },
+    {
+      "epoch": 7.48,
+      "grad_norm": 0.15684980154037476,
+      "learning_rate": 9.470051412462479e-05,
+      "loss": 0.0042,
+      "step": 3933
+    },
+    {
+      "epoch": 7.48,
+      "grad_norm": 0.10312049835920334,
+      "learning_rate": 9.456585540468948e-05,
+      "loss": 0.003,
+      "step": 3934
+    },
+    {
+      "epoch": 7.48,
+      "grad_norm": 0.08166796714067459,
+      "learning_rate": 9.44312745765024e-05,
+      "loss": 0.0016,
+      "step": 3935
+    },
+    {
+      "epoch": 7.48,
+      "grad_norm": 0.11367364972829819,
+      "learning_rate": 9.42967716910907e-05,
+      "loss": 0.0023,
+      "step": 3936
+    },
+    {
+      "epoch": 7.48,
+      "grad_norm": 0.2158229798078537,
+      "learning_rate": 9.416234679945208e-05,
+      "loss": 0.0081,
+      "step": 3937
+    },
+    {
+      "epoch": 7.49,
+      "grad_norm": 0.15637458860874176,
+      "learning_rate": 9.40279999525546e-05,
+      "loss": 0.0064,
+      "step": 3938
+    },
+    {
+      "epoch": 7.49,
+      "grad_norm": 0.09727032482624054,
+      "learning_rate": 9.389373120133704e-05,
+      "loss": 0.0052,
+      "step": 3939
+    },
+    {
+      "epoch": 7.49,
+      "grad_norm": 0.12521329522132874,
+      "learning_rate": 9.375954059670795e-05,
+      "loss": 0.0024,
+      "step": 3940
+    },
+    {
+      "epoch": 7.49,
+      "grad_norm": 0.046644095331430435,
+      "learning_rate": 9.362542818954682e-05,
+      "loss": 0.0012,
+      "step": 3941
+    },
+    {
+      "epoch": 7.49,
+      "grad_norm": 0.14109084010124207,
+      "learning_rate": 9.349139403070326e-05,
+      "loss": 0.0039,
+      "step": 3942
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 0.08029679208993912,
+      "learning_rate": 9.335743817099717e-05,
+      "loss": 0.0024,
+      "step": 3943
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 0.11416181921958923,
+      "learning_rate": 9.322356066121889e-05,
+      "loss": 0.0016,
+      "step": 3944
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 0.07085923850536346,
+      "learning_rate": 9.308976155212891e-05,
+      "loss": 0.0026,
+      "step": 3945
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 0.1116473600268364,
+      "learning_rate": 9.29560408944581e-05,
+      "loss": 0.0019,
+      "step": 3946
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 0.08090844750404358,
+      "learning_rate": 9.28223987389075e-05,
+      "loss": 0.0037,
+      "step": 3947
+    },
+    {
+      "epoch": 7.51,
+      "grad_norm": 0.07270245254039764,
+      "learning_rate": 9.268883513614866e-05,
+      "loss": 0.0037,
+      "step": 3948
+    },
+    {
+      "epoch": 7.51,
+      "grad_norm": 0.13530170917510986,
+      "learning_rate": 9.25553501368228e-05,
+      "loss": 0.0085,
+      "step": 3949
+    },
+    {
+      "epoch": 7.51,
+      "grad_norm": 0.14133824408054352,
+      "learning_rate": 9.242194379154199e-05,
+      "loss": 0.0038,
+      "step": 3950
+    },
+    {
+      "epoch": 7.51,
+      "grad_norm": 0.17420002818107605,
+      "learning_rate": 9.228861615088778e-05,
+      "loss": 0.0037,
+      "step": 3951
+    },
+    {
+      "epoch": 7.51,
+      "grad_norm": 0.06362462043762207,
+      "learning_rate": 9.215536726541258e-05,
+      "loss": 0.0016,
+      "step": 3952
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 0.15312916040420532,
+      "learning_rate": 9.202219718563848e-05,
+      "loss": 0.0023,
+      "step": 3953
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 0.10911372303962708,
+      "learning_rate": 9.188910596205779e-05,
+      "loss": 0.0032,
+      "step": 3954
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 0.16678935289382935,
+      "learning_rate": 9.1756093645133e-05,
+      "loss": 0.0049,
+      "step": 3955
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 0.08540314435958862,
+      "learning_rate": 9.162316028529662e-05,
+      "loss": 0.0027,
+      "step": 3956
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 0.04763156920671463,
+      "learning_rate": 9.14903059329512e-05,
+      "loss": 0.001,
+      "step": 3957
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 0.14146476984024048,
+      "learning_rate": 9.13575306384693e-05,
+      "loss": 0.0092,
+      "step": 3958
+    },
+    {
+      "epoch": 7.53,
+      "grad_norm": 0.059753693640232086,
+      "learning_rate": 9.122483445219387e-05,
+      "loss": 0.0016,
+      "step": 3959
+    },
+    {
+      "epoch": 7.53,
+      "grad_norm": 0.07820072025060654,
+      "learning_rate": 9.109221742443718e-05,
+      "loss": 0.0017,
+      "step": 3960
+    },
+    {
+      "epoch": 7.53,
+      "grad_norm": 0.0743987113237381,
+      "learning_rate": 9.095967960548215e-05,
+      "loss": 0.0029,
+      "step": 3961
+    },
+    {
+      "epoch": 7.53,
+      "grad_norm": 0.15836217999458313,
+      "learning_rate": 9.082722104558127e-05,
+      "loss": 0.0055,
+      "step": 3962
+    },
+    {
+      "epoch": 7.53,
+      "grad_norm": 0.13664257526397705,
+      "learning_rate": 9.069484179495717e-05,
+      "loss": 0.0051,
+      "step": 3963
+    },
+    {
+      "epoch": 7.54,
+      "grad_norm": 0.05727314203977585,
+      "learning_rate": 9.056254190380219e-05,
+      "loss": 0.0013,
+      "step": 3964
+    },
+    {
+      "epoch": 7.54,
+      "grad_norm": 0.09864210337400436,
+      "learning_rate": 9.043032142227901e-05,
+      "loss": 0.0022,
+      "step": 3965
+    },
+    {
+      "epoch": 7.54,
+      "grad_norm": 0.23010645806789398,
+      "learning_rate": 9.029818040051958e-05,
+      "loss": 0.0039,
+      "step": 3966
+    },
+    {
+      "epoch": 7.54,
+      "grad_norm": 0.07779252529144287,
+      "learning_rate": 9.016611888862636e-05,
+      "loss": 0.0018,
+      "step": 3967
+    },
+    {
+      "epoch": 7.54,
+      "grad_norm": 0.08042765408754349,
+      "learning_rate": 9.003413693667123e-05,
+      "loss": 0.0033,
+      "step": 3968
+    },
+    {
+      "epoch": 7.55,
+      "grad_norm": 0.09166476130485535,
+      "learning_rate": 8.990223459469603e-05,
+      "loss": 0.0032,
+      "step": 3969
+    },
+    {
+      "epoch": 7.55,
+      "grad_norm": 0.06879884004592896,
+      "learning_rate": 8.977041191271247e-05,
+      "loss": 0.0027,
+      "step": 3970
+    },
+    {
+      "epoch": 7.55,
+      "grad_norm": 0.1686389148235321,
+      "learning_rate": 8.9638668940702e-05,
+      "loss": 0.0031,
+      "step": 3971
+    },
+    {
+      "epoch": 7.55,
+      "grad_norm": 0.11059785634279251,
+      "learning_rate": 8.950700572861586e-05,
+      "loss": 0.007,
+      "step": 3972
+    },
+    {
+      "epoch": 7.55,
+      "grad_norm": 0.03746294975280762,
+      "learning_rate": 8.937542232637505e-05,
+      "loss": 0.0011,
+      "step": 3973
+    },
+    {
+      "epoch": 7.56,
+      "grad_norm": 0.24719147384166718,
+      "learning_rate": 8.92439187838703e-05,
+      "loss": 0.0047,
+      "step": 3974
+    },
+    {
+      "epoch": 7.56,
+      "grad_norm": 0.1051805168390274,
+      "learning_rate": 8.911249515096195e-05,
+      "loss": 0.0021,
+      "step": 3975
+    },
+    {
+      "epoch": 7.56,
+      "grad_norm": 0.13369788229465485,
+      "learning_rate": 8.898115147748052e-05,
+      "loss": 0.0074,
+      "step": 3976
+    },
+    {
+      "epoch": 7.56,
+      "grad_norm": 0.16706682741641998,
+      "learning_rate": 8.884988781322539e-05,
+      "loss": 0.0104,
+      "step": 3977
+    },
+    {
+      "epoch": 7.56,
+      "grad_norm": 0.1426948457956314,
+      "learning_rate": 8.871870420796644e-05,
+      "loss": 0.0041,
+      "step": 3978
+    },
+    {
+      "epoch": 7.56,
+      "grad_norm": 0.25022417306900024,
+      "learning_rate": 8.858760071144264e-05,
+      "loss": 0.0079,
+      "step": 3979
+    },
+    {
+      "epoch": 7.57,
+      "grad_norm": 0.1330529898405075,
+      "learning_rate": 8.845657737336283e-05,
+      "loss": 0.0094,
+      "step": 3980
+    },
+    {
+      "epoch": 7.57,
+      "grad_norm": 0.16625908017158508,
+      "learning_rate": 8.832563424340535e-05,
+      "loss": 0.0049,
+      "step": 3981
+    },
+    {
+      "epoch": 7.57,
+      "grad_norm": 0.21519093215465546,
+      "learning_rate": 8.819477137121823e-05,
+      "loss": 0.008,
+      "step": 3982
+    },
+    {
+      "epoch": 7.57,
+      "grad_norm": 0.10989359021186829,
+      "learning_rate": 8.806398880641899e-05,
+      "loss": 0.0042,
+      "step": 3983
+    },
+    {
+      "epoch": 7.57,
+      "grad_norm": 0.12254755198955536,
+      "learning_rate": 8.793328659859462e-05,
+      "loss": 0.0031,
+      "step": 3984
+    },
+    {
+      "epoch": 7.58,
+      "grad_norm": 0.1623019129037857,
+      "learning_rate": 8.7802664797302e-05,
+      "loss": 0.0085,
+      "step": 3985
+    },
+    {
+      "epoch": 7.58,
+      "grad_norm": 0.07945939153432846,
+      "learning_rate": 8.767212345206698e-05,
+      "loss": 0.0024,
+      "step": 3986
+    },
+    {
+      "epoch": 7.58,
+      "grad_norm": 0.12242913991212845,
+      "learning_rate": 8.75416626123854e-05,
+      "loss": 0.0048,
+      "step": 3987
+    },
+    {
+      "epoch": 7.58,
+      "grad_norm": 0.13323494791984558,
+      "learning_rate": 8.741128232772234e-05,
+      "loss": 0.0061,
+      "step": 3988
+    },
+    {
+      "epoch": 7.58,
+      "grad_norm": 0.18731307983398438,
+      "learning_rate": 8.72809826475123e-05,
+      "loss": 0.008,
+      "step": 3989
+    },
+    {
+      "epoch": 7.59,
+      "grad_norm": 0.07786751538515091,
+      "learning_rate": 8.715076362115937e-05,
+      "loss": 0.0018,
+      "step": 3990
+    },
+    {
+      "epoch": 7.59,
+      "grad_norm": 0.09151764959096909,
+      "learning_rate": 8.702062529803688e-05,
+      "loss": 0.0044,
+      "step": 3991
+    },
+    {
+      "epoch": 7.59,
+      "grad_norm": 0.08909397572278976,
+      "learning_rate": 8.689056772748773e-05,
+      "loss": 0.0038,
+      "step": 3992
+    },
+    {
+      "epoch": 7.59,
+      "grad_norm": 0.05358659848570824,
+      "learning_rate": 8.6760590958824e-05,
+      "loss": 0.0011,
+      "step": 3993
+    },
+    {
+      "epoch": 7.59,
+      "grad_norm": 0.05639692768454552,
+      "learning_rate": 8.663069504132757e-05,
+      "loss": 0.0014,
+      "step": 3994
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 0.0958288311958313,
+      "learning_rate": 8.6500880024249e-05,
+      "loss": 0.0037,
+      "step": 3995
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 0.04297510161995888,
+      "learning_rate": 8.637114595680875e-05,
+      "loss": 0.0008,
+      "step": 3996
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 0.028716009110212326,
+      "learning_rate": 8.624149288819628e-05,
+      "loss": 0.0006,
+      "step": 3997
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 0.16010837256908417,
+      "learning_rate": 8.61119208675705e-05,
+      "loss": 0.0102,
+      "step": 3998
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 0.10298864543437958,
+      "learning_rate": 8.598242994405949e-05,
+      "loss": 0.0089,
+      "step": 3999
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 0.14167311787605286,
+      "learning_rate": 8.585302016676057e-05,
+      "loss": 0.0033,
+      "step": 4000
+    },
+    {
+      "epoch": 7.6,
+      "eval_blimp_filtered_avg": 0.5458208955223881,
+      "eval_blimp_filtered_std": 0.005847892645055243,
+      "step": 4000
+    },
+    {
+      "epoch": 7.6,
+      "eval_blimp_supplement_avg": 0.5431034482758621,
+      "eval_blimp_supplement_std": 0.02318756738661242,
+      "step": 4000
+    },
+    {
+      "epoch": 7.6,
+      "eval_vqa_filtered_avg": 0.14,
+      "eval_vqa_filtered_std": 0.034873508801977704,
+      "step": 4000
+    },
+    {
+      "epoch": 7.6,
+      "eval_winoground_filtered_avg": 0.53,
+      "eval_winoground_filtered_std": 0.05016135580465919,
+      "step": 4000
+    },
+    {
+      "epoch": 7.61,
+      "grad_norm": 0.1303245723247528,
+      "learning_rate": 8.572369158474041e-05,
+      "loss": 0.0043,
+      "step": 4001
+    },
+    {
+      "epoch": 7.61,
+      "grad_norm": 0.14664055407047272,
+      "learning_rate": 8.559444424703471e-05,
+      "loss": 0.0103,
+      "step": 4002
+    },
+    {
+      "epoch": 7.61,
+      "grad_norm": 0.1769479662179947,
+      "learning_rate": 8.546527820264859e-05,
+      "loss": 0.0093,
+      "step": 4003
+    },
+    {
+      "epoch": 7.61,
+      "grad_norm": 0.08099735528230667,
+      "learning_rate": 8.533619350055607e-05,
+      "loss": 0.003,
+      "step": 4004
+    },
+    {
+      "epoch": 7.61,
+      "grad_norm": 0.15764102339744568,
+      "learning_rate": 8.520719018970077e-05,
+      "loss": 0.0069,
+      "step": 4005
+    },
+    {
+      "epoch": 7.62,
+      "grad_norm": 0.048241473734378815,
+      "learning_rate": 8.507826831899482e-05,
+      "loss": 0.0012,
+      "step": 4006
+    },
+    {
+      "epoch": 7.62,
+      "grad_norm": 0.19723178446292877,
+      "learning_rate": 8.49494279373201e-05,
+      "loss": 0.0067,
+      "step": 4007
+    },
+    {
+      "epoch": 7.62,
+      "grad_norm": 0.11620783805847168,
+      "learning_rate": 8.482066909352722e-05,
+      "loss": 0.0026,
+      "step": 4008
+    },
+    {
+      "epoch": 7.62,
+      "grad_norm": 0.09193775057792664,
+      "learning_rate": 8.469199183643599e-05,
+      "loss": 0.002,
+      "step": 4009
+    },
+    {
+      "epoch": 7.62,
+      "grad_norm": 0.12230918556451797,
+      "learning_rate": 8.456339621483507e-05,
+      "loss": 0.0043,
+      "step": 4010
+    },
+    {
+      "epoch": 7.63,
+      "grad_norm": 0.18952202796936035,
+      "learning_rate": 8.443488227748275e-05,
+      "loss": 0.0127,
+      "step": 4011
+    },
+    {
+      "epoch": 7.63,
+      "grad_norm": 0.10513338446617126,
+      "learning_rate": 8.430645007310556e-05,
+      "loss": 0.0048,
+      "step": 4012
+    },
+    {
+      "epoch": 7.63,
+      "grad_norm": 0.1264156997203827,
+      "learning_rate": 8.41780996503997e-05,
+      "loss": 0.0032,
+      "step": 4013
+    },
+    {
+      "epoch": 7.63,
+      "grad_norm": 0.10326673090457916,
+      "learning_rate": 8.404983105803002e-05,
+      "loss": 0.0026,
+      "step": 4014
+    },
+    {
+      "epoch": 7.63,
+      "grad_norm": 0.1269594430923462,
+      "learning_rate": 8.392164434463044e-05,
+      "loss": 0.0061,
+      "step": 4015
+    },
+    {
+      "epoch": 7.63,
+      "grad_norm": 0.23270970582962036,
+      "learning_rate": 8.379353955880378e-05,
+      "loss": 0.0073,
+      "step": 4016
+    },
+    {
+      "epoch": 7.64,
+      "grad_norm": 0.05319410562515259,
+      "learning_rate": 8.366551674912189e-05,
+      "loss": 0.0014,
+      "step": 4017
+    },
+    {
+      "epoch": 7.64,
+      "grad_norm": 0.1398974508047104,
+      "learning_rate": 8.353757596412546e-05,
+      "loss": 0.005,
+      "step": 4018
+    },
+    {
+      "epoch": 7.64,
+      "grad_norm": 0.13132092356681824,
+      "learning_rate": 8.3409717252324e-05,
+      "loss": 0.0065,
+      "step": 4019
+    },
+    {
+      "epoch": 7.64,
+      "grad_norm": 0.098666250705719,
+      "learning_rate": 8.328194066219632e-05,
+      "loss": 0.009,
+      "step": 4020
+    },
+    {
+      "epoch": 7.64,
+      "grad_norm": 0.32968997955322266,
+      "learning_rate": 8.315424624218936e-05,
+      "loss": 0.0307,
+      "step": 4021
+    },
+    {
+      "epoch": 7.65,
+      "grad_norm": 0.0939842239022255,
+      "learning_rate": 8.302663404071974e-05,
+      "loss": 0.0026,
+      "step": 4022
+    },
+    {
+      "epoch": 7.65,
+      "grad_norm": 0.11019507795572281,
+      "learning_rate": 8.289910410617214e-05,
+      "loss": 0.0034,
+      "step": 4023
+    },
+    {
+      "epoch": 7.65,
+      "grad_norm": 0.08741119503974915,
+      "learning_rate": 8.277165648690058e-05,
+      "loss": 0.0028,
+      "step": 4024
+    },
+    {
+      "epoch": 7.65,
+      "grad_norm": 0.09374744445085526,
+      "learning_rate": 8.264429123122771e-05,
+      "loss": 0.0042,
+      "step": 4025
+    },
+    {
+      "epoch": 7.65,
+      "grad_norm": 0.0767391100525856,
+      "learning_rate": 8.251700838744486e-05,
+      "loss": 0.0017,
+      "step": 4026
+    },
+    {
+      "epoch": 7.66,
+      "grad_norm": 0.07770463079214096,
+      "learning_rate": 8.238980800381216e-05,
+      "loss": 0.0027,
+      "step": 4027
+    },
+    {
+      "epoch": 7.66,
+      "grad_norm": 0.1753794550895691,
+      "learning_rate": 8.226269012855861e-05,
+      "loss": 0.0063,
+      "step": 4028
+    },
+    {
+      "epoch": 7.66,
+      "grad_norm": 0.13963709771633148,
+      "learning_rate": 8.213565480988169e-05,
+      "loss": 0.0053,
+      "step": 4029
+    },
+    {
+      "epoch": 7.66,
+      "grad_norm": 0.1402224600315094,
+      "learning_rate": 8.200870209594768e-05,
+      "loss": 0.0038,
+      "step": 4030
+    },
+    {
+      "epoch": 7.66,
+      "grad_norm": 0.09487905353307724,
+      "learning_rate": 8.188183203489183e-05,
+      "loss": 0.0023,
+      "step": 4031
+    },
+    {
+      "epoch": 7.67,
+      "grad_norm": 0.041400544345378876,
+      "learning_rate": 8.175504467481736e-05,
+      "loss": 0.0015,
+      "step": 4032
+    },
+    {
+      "epoch": 7.67,
+      "grad_norm": 0.11547128111124039,
+      "learning_rate": 8.162834006379693e-05,
+      "loss": 0.0036,
+      "step": 4033
+    },
+    {
+      "epoch": 7.67,
+      "grad_norm": 0.14916066825389862,
+      "learning_rate": 8.15017182498713e-05,
+      "loss": 0.0031,
+      "step": 4034
+    },
+    {
+      "epoch": 7.67,
+      "grad_norm": 0.09352637082338333,
+      "learning_rate": 8.137517928104997e-05,
+      "loss": 0.0035,
+      "step": 4035
+    },
+    {
+      "epoch": 7.67,
+      "grad_norm": 0.14974668622016907,
+      "learning_rate": 8.12487232053111e-05,
+      "loss": 0.0087,
+      "step": 4036
+    },
+    {
+      "epoch": 7.67,
+      "grad_norm": 0.11398351192474365,
+      "learning_rate": 8.11223500706013e-05,
+      "loss": 0.0026,
+      "step": 4037
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 0.21017543971538544,
+      "learning_rate": 8.09960599248359e-05,
+      "loss": 0.0103,
+      "step": 4038
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 0.05676844343543053,
+      "learning_rate": 8.086985281589851e-05,
+      "loss": 0.0016,
+      "step": 4039
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 0.14155696332454681,
+      "learning_rate": 8.074372879164166e-05,
+      "loss": 0.003,
+      "step": 4040
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 0.047536782920360565,
+      "learning_rate": 8.061768789988581e-05,
+      "loss": 0.0011,
+      "step": 4041
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 0.0430448018014431,
+      "learning_rate": 8.049173018842046e-05,
+      "loss": 0.0012,
+      "step": 4042
+    },
+    {
+      "epoch": 7.69,
+      "grad_norm": 0.25413843989372253,
+      "learning_rate": 8.03658557050033e-05,
+      "loss": 0.005,
+      "step": 4043
+    },
+    {
+      "epoch": 7.69,
+      "grad_norm": 0.05802446976304054,
+      "learning_rate": 8.02400644973604e-05,
+      "loss": 0.002,
+      "step": 4044
+    },
+    {
+      "epoch": 7.69,
+      "grad_norm": 0.296644002199173,
+      "learning_rate": 8.011435661318634e-05,
+      "loss": 0.0077,
+      "step": 4045
+    },
+    {
+      "epoch": 7.69,
+      "grad_norm": 0.11945470422506332,
+      "learning_rate": 7.998873210014434e-05,
+      "loss": 0.003,
+      "step": 4046
+    },
+    {
+      "epoch": 7.69,
+      "grad_norm": 0.18291762471199036,
+      "learning_rate": 7.98631910058655e-05,
+      "loss": 0.0049,
+      "step": 4047
+    },
+    {
+      "epoch": 7.7,
+      "grad_norm": 0.11856916546821594,
+      "learning_rate": 7.973773337794976e-05,
+      "loss": 0.0064,
+      "step": 4048
+    },
+    {
+      "epoch": 7.7,
+      "grad_norm": 0.07038857042789459,
+      "learning_rate": 7.96123592639651e-05,
+      "loss": 0.0013,
+      "step": 4049
+    },
+    {
+      "epoch": 7.7,
+      "grad_norm": 0.20491763949394226,
+      "learning_rate": 7.948706871144798e-05,
+      "loss": 0.009,
+      "step": 4050
+    },
+    {
+      "epoch": 7.7,
+      "grad_norm": 0.0901680439710617,
+      "learning_rate": 7.936186176790339e-05,
+      "loss": 0.0018,
+      "step": 4051
+    },
+    {
+      "epoch": 7.7,
+      "grad_norm": 0.1236966997385025,
+      "learning_rate": 7.923673848080404e-05,
+      "loss": 0.0037,
+      "step": 4052
+    },
+    {
+      "epoch": 7.71,
+      "grad_norm": 0.05191247910261154,
+      "learning_rate": 7.911169889759152e-05,
+      "loss": 0.0016,
+      "step": 4053
+    },
+    {
+      "epoch": 7.71,
+      "grad_norm": 0.16157817840576172,
+      "learning_rate": 7.898674306567539e-05,
+      "loss": 0.0096,
+      "step": 4054
+    },
+    {
+      "epoch": 7.71,
+      "grad_norm": 0.21498946845531464,
+      "learning_rate": 7.886187103243343e-05,
+      "loss": 0.0049,
+      "step": 4055
+    },
+    {
+      "epoch": 7.71,
+      "grad_norm": 0.07445069402456284,
+      "learning_rate": 7.873708284521169e-05,
+      "loss": 0.0015,
+      "step": 4056
+    },
+    {
+      "epoch": 7.71,
+      "grad_norm": 0.10784732550382614,
+      "learning_rate": 7.861237855132473e-05,
+      "loss": 0.004,
+      "step": 4057
+    },
+    {
+      "epoch": 7.71,
+      "grad_norm": 0.14395251870155334,
+      "learning_rate": 7.848775819805463e-05,
+      "loss": 0.0023,
+      "step": 4058
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 0.17760717868804932,
+      "learning_rate": 7.836322183265229e-05,
+      "loss": 0.0061,
+      "step": 4059
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 0.08418362587690353,
+      "learning_rate": 7.82387695023365e-05,
+      "loss": 0.0014,
+      "step": 4060
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 0.0792212188243866,
+      "learning_rate": 7.811440125429417e-05,
+      "loss": 0.004,
+      "step": 4061
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 0.07627598196268082,
+      "learning_rate": 7.799011713568035e-05,
+      "loss": 0.0026,
+      "step": 4062
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 0.060418397188186646,
+      "learning_rate": 7.786591719361823e-05,
+      "loss": 0.0017,
+      "step": 4063
+    },
+    {
+      "epoch": 7.73,
+      "grad_norm": 0.09671566635370255,
+      "learning_rate": 7.774180147519903e-05,
+      "loss": 0.0034,
+      "step": 4064
+    },
+    {
+      "epoch": 7.73,
+      "grad_norm": 0.09250207990407944,
+      "learning_rate": 7.761777002748203e-05,
+      "loss": 0.0019,
+      "step": 4065
+    },
+    {
+      "epoch": 7.73,
+      "grad_norm": 0.11556510627269745,
+      "learning_rate": 7.749382289749479e-05,
+      "loss": 0.0033,
+      "step": 4066
+    },
+    {
+      "epoch": 7.73,
+      "grad_norm": 0.11957184970378876,
+      "learning_rate": 7.736996013223241e-05,
+      "loss": 0.0064,
+      "step": 4067
+    },
+    {
+      "epoch": 7.73,
+      "grad_norm": 0.08345303684473038,
+      "learning_rate": 7.724618177865852e-05,
+      "loss": 0.001,
+      "step": 4068
+    },
+    {
+      "epoch": 7.74,
+      "grad_norm": 0.05402961000800133,
+      "learning_rate": 7.712248788370444e-05,
+      "loss": 0.0022,
+      "step": 4069
+    },
+    {
+      "epoch": 7.74,
+      "grad_norm": 0.14943891763687134,
+      "learning_rate": 7.69988784942696e-05,
+      "loss": 0.0044,
+      "step": 4070
+    },
+    {
+      "epoch": 7.74,
+      "grad_norm": 0.14743800461292267,
+      "learning_rate": 7.687535365722126e-05,
+      "loss": 0.0128,
+      "step": 4071
+    },
+    {
+      "epoch": 7.74,
+      "grad_norm": 0.10255265235900879,
+      "learning_rate": 7.675191341939472e-05,
+      "loss": 0.0018,
+      "step": 4072
+    },
+    {
+      "epoch": 7.74,
+      "grad_norm": 0.11189456284046173,
+      "learning_rate": 7.662855782759325e-05,
+      "loss": 0.004,
+      "step": 4073
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 0.14537198841571808,
+      "learning_rate": 7.650528692858788e-05,
+      "loss": 0.0021,
+      "step": 4074
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 0.1996055692434311,
+      "learning_rate": 7.638210076911763e-05,
+      "loss": 0.0123,
+      "step": 4075
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 0.16970914602279663,
+      "learning_rate": 7.625899939588929e-05,
+      "loss": 0.0029,
+      "step": 4076
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 0.13972023129463196,
+      "learning_rate": 7.613598285557786e-05,
+      "loss": 0.0057,
+      "step": 4077
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 0.04111985117197037,
+      "learning_rate": 7.601305119482548e-05,
+      "loss": 0.001,
+      "step": 4078
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 0.20507654547691345,
+      "learning_rate": 7.589020446024286e-05,
+      "loss": 0.009,
+      "step": 4079
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 0.09638693928718567,
+      "learning_rate": 7.576744269840802e-05,
+      "loss": 0.0042,
+      "step": 4080
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 0.10455556958913803,
+      "learning_rate": 7.564476595586698e-05,
+      "loss": 0.0039,
+      "step": 4081
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 0.06798138469457626,
+      "learning_rate": 7.552217427913337e-05,
+      "loss": 0.0015,
+      "step": 4082
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 0.35759738087654114,
+      "learning_rate": 7.539966771468874e-05,
+      "loss": 0.0197,
+      "step": 4083
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 0.01813567243516445,
+      "learning_rate": 7.527724630898229e-05,
+      "loss": 0.0005,
+      "step": 4084
+    },
+    {
+      "epoch": 7.77,
+      "grad_norm": 0.09170147031545639,
+      "learning_rate": 7.515491010843075e-05,
+      "loss": 0.0042,
+      "step": 4085
+    },
+    {
+      "epoch": 7.77,
+      "grad_norm": 0.03410191088914871,
+      "learning_rate": 7.503265915941905e-05,
+      "loss": 0.0009,
+      "step": 4086
+    },
+    {
+      "epoch": 7.77,
+      "grad_norm": 0.23523762822151184,
+      "learning_rate": 7.49104935082991e-05,
+      "loss": 0.0115,
+      "step": 4087
+    },
+    {
+      "epoch": 7.77,
+      "grad_norm": 0.10170857608318329,
+      "learning_rate": 7.478841320139107e-05,
+      "loss": 0.0049,
+      "step": 4088
+    },
+    {
+      "epoch": 7.77,
+      "grad_norm": 0.13393162190914154,
+      "learning_rate": 7.466641828498249e-05,
+      "loss": 0.0048,
+      "step": 4089
+    },
+    {
+      "epoch": 7.78,
+      "grad_norm": 0.039297182112932205,
+      "learning_rate": 7.454450880532854e-05,
+      "loss": 0.0008,
+      "step": 4090
+    },
+    {
+      "epoch": 7.78,
+      "grad_norm": 0.06274878233671188,
+      "learning_rate": 7.442268480865206e-05,
+      "loss": 0.0021,
+      "step": 4091
+    },
+    {
+      "epoch": 7.78,
+      "grad_norm": 0.20076295733451843,
+      "learning_rate": 7.430094634114335e-05,
+      "loss": 0.0147,
+      "step": 4092
+    },
+    {
+      "epoch": 7.78,
+      "grad_norm": 0.03032536618411541,
+      "learning_rate": 7.417929344896048e-05,
+      "loss": 0.0007,
+      "step": 4093
+    },
+    {
+      "epoch": 7.78,
+      "grad_norm": 0.14117053151130676,
+      "learning_rate": 7.405772617822882e-05,
+      "loss": 0.0039,
+      "step": 4094
+    },
+    {
+      "epoch": 7.79,
+      "grad_norm": 0.1175849512219429,
+      "learning_rate": 7.393624457504174e-05,
+      "loss": 0.0019,
+      "step": 4095
+    },
+    {
+      "epoch": 7.79,
+      "grad_norm": 0.0354345440864563,
+      "learning_rate": 7.381484868545939e-05,
+      "loss": 0.0013,
+      "step": 4096
+    },
+    {
+      "epoch": 7.79,
+      "grad_norm": 0.07249219715595245,
+      "learning_rate": 7.369353855551025e-05,
+      "loss": 0.0031,
+      "step": 4097
+    },
+    {
+      "epoch": 7.79,
+      "grad_norm": 0.09213780611753464,
+      "learning_rate": 7.35723142311895e-05,
+      "loss": 0.0042,
+      "step": 4098
+    },
+    {
+      "epoch": 7.79,
+      "grad_norm": 0.1743892878293991,
+      "learning_rate": 7.345117575846045e-05,
+      "loss": 0.0074,
+      "step": 4099
+    },
+    {
+      "epoch": 7.79,
+      "grad_norm": 0.12464078515768051,
+      "learning_rate": 7.333012318325348e-05,
+      "loss": 0.0093,
+      "step": 4100
+    },
+    {
+      "epoch": 7.79,
+      "eval_blimp_filtered_avg": 0.5471641791044776,
+      "eval_blimp_filtered_std": 0.005839324948409529,
+      "step": 4100
+    },
+    {
+      "epoch": 7.79,
+      "eval_blimp_supplement_avg": 0.5366379310344828,
+      "eval_blimp_supplement_std": 0.02320685965854663,
+      "step": 4100
+    },
+    {
+      "epoch": 7.79,
+      "eval_vqa_filtered_avg": 0.16,
+      "eval_vqa_filtered_std": 0.03684529491774708,
+      "step": 4100
+    },
+    {
+      "epoch": 7.79,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 4100
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 0.09942837059497833,
+      "learning_rate": 7.320915655146648e-05,
+      "loss": 0.0049,
+      "step": 4101
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 0.15560336410999298,
+      "learning_rate": 7.308827590896466e-05,
+      "loss": 0.0039,
+      "step": 4102
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 0.08754675090312958,
+      "learning_rate": 7.296748130158105e-05,
+      "loss": 0.003,
+      "step": 4103
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 0.05453642085194588,
+      "learning_rate": 7.284677277511537e-05,
+      "loss": 0.0015,
+      "step": 4104
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 0.16894111037254333,
+      "learning_rate": 7.272615037533533e-05,
+      "loss": 0.0055,
+      "step": 4105
+    },
+    {
+      "epoch": 7.81,
+      "grad_norm": 0.21784016489982605,
+      "learning_rate": 7.260561414797567e-05,
+      "loss": 0.0112,
+      "step": 4106
+    },
+    {
+      "epoch": 7.81,
+      "grad_norm": 0.06743381917476654,
+      "learning_rate": 7.248516413873847e-05,
+      "loss": 0.0025,
+      "step": 4107
+    },
+    {
+      "epoch": 7.81,
+      "grad_norm": 0.09489069133996964,
+      "learning_rate": 7.236480039329319e-05,
+      "loss": 0.0036,
+      "step": 4108
+    },
+    {
+      "epoch": 7.81,
+      "grad_norm": 0.07150664180517197,
+      "learning_rate": 7.224452295727661e-05,
+      "loss": 0.0035,
+      "step": 4109
+    },
+    {
+      "epoch": 7.81,
+      "grad_norm": 0.09626176953315735,
+      "learning_rate": 7.212433187629264e-05,
+      "loss": 0.0022,
+      "step": 4110
+    },
+    {
+      "epoch": 7.82,
+      "grad_norm": 0.1656753122806549,
+      "learning_rate": 7.200422719591254e-05,
+      "loss": 0.0044,
+      "step": 4111
+    },
+    {
+      "epoch": 7.82,
+      "grad_norm": 0.09036063402891159,
+      "learning_rate": 7.188420896167502e-05,
+      "loss": 0.0016,
+      "step": 4112
+    },
+    {
+      "epoch": 7.82,
+      "grad_norm": 0.10115846991539001,
+      "learning_rate": 7.176427721908547e-05,
+      "loss": 0.003,
+      "step": 4113
+    },
+    {
+      "epoch": 7.82,
+      "grad_norm": 0.18600879609584808,
+      "learning_rate": 7.164443201361711e-05,
+      "loss": 0.0037,
+      "step": 4114
+    },
+    {
+      "epoch": 7.82,
+      "grad_norm": 0.05557117238640785,
+      "learning_rate": 7.152467339070996e-05,
+      "loss": 0.0013,
+      "step": 4115
+    },
+    {
+      "epoch": 7.83,
+      "grad_norm": 0.12741954624652863,
+      "learning_rate": 7.140500139577133e-05,
+      "loss": 0.004,
+      "step": 4116
+    },
+    {
+      "epoch": 7.83,
+      "grad_norm": 0.07878808677196503,
+      "learning_rate": 7.128541607417562e-05,
+      "loss": 0.0014,
+      "step": 4117
+    },
+    {
+      "epoch": 7.83,
+      "grad_norm": 0.05502631142735481,
+      "learning_rate": 7.116591747126445e-05,
+      "loss": 0.0011,
+      "step": 4118
+    },
+    {
+      "epoch": 7.83,
+      "grad_norm": 0.06397872418165207,
+      "learning_rate": 7.104650563234652e-05,
+      "loss": 0.0016,
+      "step": 4119
+    },
+    {
+      "epoch": 7.83,
+      "grad_norm": 0.1640637367963791,
+      "learning_rate": 7.092718060269763e-05,
+      "loss": 0.0032,
+      "step": 4120
+    },
+    {
+      "epoch": 7.83,
+      "grad_norm": 0.1759433001279831,
+      "learning_rate": 7.080794242756069e-05,
+      "loss": 0.0026,
+      "step": 4121
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 0.10722186416387558,
+      "learning_rate": 7.068879115214558e-05,
+      "loss": 0.0035,
+      "step": 4122
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 0.19047614932060242,
+      "learning_rate": 7.056972682162953e-05,
+      "loss": 0.0054,
+      "step": 4123
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 0.11871615797281265,
+      "learning_rate": 7.04507494811563e-05,
+      "loss": 0.0027,
+      "step": 4124
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 0.1056840792298317,
+      "learning_rate": 7.03318591758372e-05,
+      "loss": 0.0032,
+      "step": 4125
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 0.21510253846645355,
+      "learning_rate": 7.021305595075014e-05,
+      "loss": 0.0053,
+      "step": 4126
+    },
+    {
+      "epoch": 7.85,
+      "grad_norm": 0.12256897985935211,
+      "learning_rate": 7.009433985094025e-05,
+      "loss": 0.0031,
+      "step": 4127
+    },
+    {
+      "epoch": 7.85,
+      "grad_norm": 0.10061811655759811,
+      "learning_rate": 6.997571092141955e-05,
+      "loss": 0.004,
+      "step": 4128
+    },
+    {
+      "epoch": 7.85,
+      "grad_norm": 0.08170633018016815,
+      "learning_rate": 6.985716920716692e-05,
+      "loss": 0.0016,
+      "step": 4129
+    },
+    {
+      "epoch": 7.85,
+      "grad_norm": 0.02924034558236599,
+      "learning_rate": 6.973871475312835e-05,
+      "loss": 0.0008,
+      "step": 4130
+    },
+    {
+      "epoch": 7.85,
+      "grad_norm": 0.1070796400308609,
+      "learning_rate": 6.962034760421646e-05,
+      "loss": 0.0023,
+      "step": 4131
+    },
+    {
+      "epoch": 7.86,
+      "grad_norm": 0.10657835751771927,
+      "learning_rate": 6.95020678053113e-05,
+      "loss": 0.0049,
+      "step": 4132
+    },
+    {
+      "epoch": 7.86,
+      "grad_norm": 0.21173113584518433,
+      "learning_rate": 6.938387540125911e-05,
+      "loss": 0.0016,
+      "step": 4133
+    },
+    {
+      "epoch": 7.86,
+      "grad_norm": 0.07949985563755035,
+      "learning_rate": 6.926577043687355e-05,
+      "loss": 0.0037,
+      "step": 4134
+    },
+    {
+      "epoch": 7.86,
+      "grad_norm": 0.04517035186290741,
+      "learning_rate": 6.914775295693486e-05,
+      "loss": 0.0013,
+      "step": 4135
+    },
+    {
+      "epoch": 7.86,
+      "grad_norm": 0.21489663422107697,
+      "learning_rate": 6.902982300619014e-05,
+      "loss": 0.0143,
+      "step": 4136
+    },
+    {
+      "epoch": 7.87,
+      "grad_norm": 0.15263940393924713,
+      "learning_rate": 6.891198062935339e-05,
+      "loss": 0.0032,
+      "step": 4137
+    },
+    {
+      "epoch": 7.87,
+      "grad_norm": 0.09760142862796783,
+      "learning_rate": 6.879422587110529e-05,
+      "loss": 0.0026,
+      "step": 4138
+    },
+    {
+      "epoch": 7.87,
+      "grad_norm": 0.13082119822502136,
+      "learning_rate": 6.867655877609337e-05,
+      "loss": 0.0034,
+      "step": 4139
+    },
+    {
+      "epoch": 7.87,
+      "grad_norm": 0.07971687614917755,
+      "learning_rate": 6.855897938893184e-05,
+      "loss": 0.0022,
+      "step": 4140
+    },
+    {
+      "epoch": 7.87,
+      "grad_norm": 0.22364968061447144,
+      "learning_rate": 6.844148775420195e-05,
+      "loss": 0.012,
+      "step": 4141
+    },
+    {
+      "epoch": 7.87,
+      "grad_norm": 0.08361897617578506,
+      "learning_rate": 6.832408391645112e-05,
+      "loss": 0.002,
+      "step": 4142
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 0.060622554272413254,
+      "learning_rate": 6.820676792019404e-05,
+      "loss": 0.0015,
+      "step": 4143
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 0.07741209119558334,
+      "learning_rate": 6.808953980991185e-05,
+      "loss": 0.0018,
+      "step": 4144
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 0.06131153181195259,
+      "learning_rate": 6.797239963005234e-05,
+      "loss": 0.001,
+      "step": 4145
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 0.10538023710250854,
+      "learning_rate": 6.785534742503e-05,
+      "loss": 0.0072,
+      "step": 4146
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 0.1410081833600998,
+      "learning_rate": 6.773838323922594e-05,
+      "loss": 0.0026,
+      "step": 4147
+    },
+    {
+      "epoch": 7.89,
+      "grad_norm": 0.07217789441347122,
+      "learning_rate": 6.762150711698788e-05,
+      "loss": 0.0013,
+      "step": 4148
+    },
+    {
+      "epoch": 7.89,
+      "grad_norm": 0.2765141725540161,
+      "learning_rate": 6.750471910263045e-05,
+      "loss": 0.0219,
+      "step": 4149
+    },
+    {
+      "epoch": 7.89,
+      "grad_norm": 0.11924959719181061,
+      "learning_rate": 6.738801924043424e-05,
+      "loss": 0.0025,
+      "step": 4150
+    },
+    {
+      "epoch": 7.89,
+      "grad_norm": 0.04084214195609093,
+      "learning_rate": 6.727140757464709e-05,
+      "loss": 0.0008,
+      "step": 4151
+    },
+    {
+      "epoch": 7.89,
+      "grad_norm": 0.24126484990119934,
+      "learning_rate": 6.715488414948301e-05,
+      "loss": 0.0097,
+      "step": 4152
+    },
+    {
+      "epoch": 7.9,
+      "grad_norm": 0.0894797220826149,
+      "learning_rate": 6.703844900912261e-05,
+      "loss": 0.0011,
+      "step": 4153
+    },
+    {
+      "epoch": 7.9,
+      "grad_norm": 0.07531911134719849,
+      "learning_rate": 6.692210219771312e-05,
+      "loss": 0.0019,
+      "step": 4154
+    },
+    {
+      "epoch": 7.9,
+      "grad_norm": 0.06870628148317337,
+      "learning_rate": 6.680584375936817e-05,
+      "loss": 0.0024,
+      "step": 4155
+    },
+    {
+      "epoch": 7.9,
+      "grad_norm": 0.3415929675102234,
+      "learning_rate": 6.668967373816796e-05,
+      "loss": 0.0107,
+      "step": 4156
+    },
+    {
+      "epoch": 7.9,
+      "grad_norm": 0.14048947393894196,
+      "learning_rate": 6.657359217815905e-05,
+      "loss": 0.0036,
+      "step": 4157
+    },
+    {
+      "epoch": 7.9,
+      "grad_norm": 0.026627441868185997,
+      "learning_rate": 6.645759912335481e-05,
+      "loss": 0.0007,
+      "step": 4158
+    },
+    {
+      "epoch": 7.91,
+      "grad_norm": 0.10162100195884705,
+      "learning_rate": 6.634169461773446e-05,
+      "loss": 0.0042,
+      "step": 4159
+    },
+    {
+      "epoch": 7.91,
+      "grad_norm": 0.1429271697998047,
+      "learning_rate": 6.622587870524422e-05,
+      "loss": 0.0031,
+      "step": 4160
+    },
+    {
+      "epoch": 7.91,
+      "grad_norm": 0.10257145762443542,
+      "learning_rate": 6.611015142979641e-05,
+      "loss": 0.0032,
+      "step": 4161
+    },
+    {
+      "epoch": 7.91,
+      "grad_norm": 0.11412620544433594,
+      "learning_rate": 6.599451283526977e-05,
+      "loss": 0.0028,
+      "step": 4162
+    },
+    {
+      "epoch": 7.91,
+      "grad_norm": 0.08232184499502182,
+      "learning_rate": 6.587896296550949e-05,
+      "loss": 0.0036,
+      "step": 4163
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 0.14282815158367157,
+      "learning_rate": 6.57635018643271e-05,
+      "loss": 0.0036,
+      "step": 4164
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 0.08698391169309616,
+      "learning_rate": 6.564812957550044e-05,
+      "loss": 0.0018,
+      "step": 4165
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 0.1678047776222229,
+      "learning_rate": 6.55328461427736e-05,
+      "loss": 0.0041,
+      "step": 4166
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 0.21196235716342926,
+      "learning_rate": 6.541765160985734e-05,
+      "loss": 0.0039,
+      "step": 4167
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 0.0818399116396904,
+      "learning_rate": 6.530254602042811e-05,
+      "loss": 0.0026,
+      "step": 4168
+    },
+    {
+      "epoch": 7.93,
+      "grad_norm": 0.07230453193187714,
+      "learning_rate": 6.518752941812929e-05,
+      "loss": 0.0011,
+      "step": 4169
+    },
+    {
+      "epoch": 7.93,
+      "grad_norm": 0.0816914439201355,
+      "learning_rate": 6.507260184656991e-05,
+      "loss": 0.0019,
+      "step": 4170
+    },
+    {
+      "epoch": 7.93,
+      "grad_norm": 0.142021045088768,
+      "learning_rate": 6.495776334932576e-05,
+      "loss": 0.0029,
+      "step": 4171
+    },
+    {
+      "epoch": 7.93,
+      "grad_norm": 0.11716854572296143,
+      "learning_rate": 6.484301396993857e-05,
+      "loss": 0.0027,
+      "step": 4172
+    },
+    {
+      "epoch": 7.93,
+      "grad_norm": 0.1062198206782341,
+      "learning_rate": 6.472835375191634e-05,
+      "loss": 0.0016,
+      "step": 4173
+    },
+    {
+      "epoch": 7.94,
+      "grad_norm": 0.1554822027683258,
+      "learning_rate": 6.461378273873326e-05,
+      "loss": 0.0033,
+      "step": 4174
+    },
+    {
+      "epoch": 7.94,
+      "grad_norm": 0.08375168591737747,
+      "learning_rate": 6.44993009738297e-05,
+      "loss": 0.0019,
+      "step": 4175
+    },
+    {
+      "epoch": 7.94,
+      "grad_norm": 0.0982712134718895,
+      "learning_rate": 6.43849085006122e-05,
+      "loss": 0.0036,
+      "step": 4176
+    },
+    {
+      "epoch": 7.94,
+      "grad_norm": 0.11767299473285675,
+      "learning_rate": 6.427060536245338e-05,
+      "loss": 0.0034,
+      "step": 4177
+    },
+    {
+      "epoch": 7.94,
+      "grad_norm": 0.1260092556476593,
+      "learning_rate": 6.415639160269226e-05,
+      "loss": 0.004,
+      "step": 4178
+    },
+    {
+      "epoch": 7.94,
+      "grad_norm": 0.1670764982700348,
+      "learning_rate": 6.404226726463347e-05,
+      "loss": 0.0048,
+      "step": 4179
+    },
+    {
+      "epoch": 7.95,
+      "grad_norm": 0.06806144118309021,
+      "learning_rate": 6.392823239154824e-05,
+      "loss": 0.0015,
+      "step": 4180
+    },
+    {
+      "epoch": 7.95,
+      "grad_norm": 0.09423622488975525,
+      "learning_rate": 6.381428702667361e-05,
+      "loss": 0.0041,
+      "step": 4181
+    },
+    {
+      "epoch": 7.95,
+      "grad_norm": 0.12842021882534027,
+      "learning_rate": 6.370043121321274e-05,
+      "loss": 0.0053,
+      "step": 4182
+    },
+    {
+      "epoch": 7.95,
+      "grad_norm": 0.046266820281744,
+      "learning_rate": 6.35866649943348e-05,
+      "loss": 0.0011,
+      "step": 4183
+    },
+    {
+      "epoch": 7.95,
+      "grad_norm": 0.09481174498796463,
+      "learning_rate": 6.347298841317506e-05,
+      "loss": 0.0028,
+      "step": 4184
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 0.1452246606349945,
+      "learning_rate": 6.335940151283471e-05,
+      "loss": 0.003,
+      "step": 4185
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 0.1533493846654892,
+      "learning_rate": 6.3245904336381e-05,
+      "loss": 0.0033,
+      "step": 4186
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 0.15482865273952484,
+      "learning_rate": 6.313249692684732e-05,
+      "loss": 0.0073,
+      "step": 4187
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 0.20679812133312225,
+      "learning_rate": 6.301917932723258e-05,
+      "loss": 0.0036,
+      "step": 4188
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 0.16006958484649658,
+      "learning_rate": 6.290595158050212e-05,
+      "loss": 0.0073,
+      "step": 4189
+    },
+    {
+      "epoch": 7.97,
+      "grad_norm": 0.09196721762418747,
+      "learning_rate": 6.279281372958693e-05,
+      "loss": 0.0029,
+      "step": 4190
+    },
+    {
+      "epoch": 7.97,
+      "grad_norm": 0.028726940974593163,
+      "learning_rate": 6.267976581738405e-05,
+      "loss": 0.0007,
+      "step": 4191
+    },
+    {
+      "epoch": 7.97,
+      "grad_norm": 0.1468905806541443,
+      "learning_rate": 6.256680788675623e-05,
+      "loss": 0.0097,
+      "step": 4192
+    },
+    {
+      "epoch": 7.97,
+      "grad_norm": 0.09179184585809708,
+      "learning_rate": 6.245393998053248e-05,
+      "loss": 0.0024,
+      "step": 4193
+    },
+    {
+      "epoch": 7.97,
+      "grad_norm": 0.06782804429531097,
+      "learning_rate": 6.234116214150712e-05,
+      "loss": 0.0016,
+      "step": 4194
+    },
+    {
+      "epoch": 7.98,
+      "grad_norm": 0.17963318526744843,
+      "learning_rate": 6.222847441244092e-05,
+      "loss": 0.0025,
+      "step": 4195
+    },
+    {
+      "epoch": 7.98,
+      "grad_norm": 0.3854430317878723,
+      "learning_rate": 6.211587683605995e-05,
+      "loss": 0.0373,
+      "step": 4196
+    },
+    {
+      "epoch": 7.98,
+      "grad_norm": 0.08931051194667816,
+      "learning_rate": 6.200336945505652e-05,
+      "loss": 0.0019,
+      "step": 4197
+    },
+    {
+      "epoch": 7.98,
+      "grad_norm": 0.22708584368228912,
+      "learning_rate": 6.18909523120885e-05,
+      "loss": 0.0105,
+      "step": 4198
+    },
+    {
+      "epoch": 7.98,
+      "grad_norm": 0.1673729568719864,
+      "learning_rate": 6.177862544977965e-05,
+      "loss": 0.003,
+      "step": 4199
+    },
+    {
+      "epoch": 7.98,
+      "grad_norm": 0.07132242619991302,
+      "learning_rate": 6.16663889107194e-05,
+      "loss": 0.0016,
+      "step": 4200
+    },
+    {
+      "epoch": 7.98,
+      "eval_blimp_filtered_avg": 0.5444776119402985,
+      "eval_blimp_filtered_std": 0.005853411675913941,
+      "step": 4200
+    },
+    {
+      "epoch": 7.98,
+      "eval_blimp_supplement_avg": 0.540948275862069,
+      "eval_blimp_supplement_std": 0.023184529763758722,
+      "step": 4200
+    },
+    {
+      "epoch": 7.98,
+      "eval_vqa_filtered_avg": 0.11,
+      "eval_vqa_filtered_std": 0.031446603773522035,
+      "step": 4200
+    },
+    {
+      "epoch": 7.98,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 4200
+    },
+    {
+      "epoch": 7.99,
+      "grad_norm": 0.17357394099235535,
+      "learning_rate": 6.155424273746304e-05,
+      "loss": 0.0038,
+      "step": 4201
+    },
+    {
+      "epoch": 7.99,
+      "grad_norm": 0.10920485854148865,
+      "learning_rate": 6.144218697253149e-05,
+      "loss": 0.0015,
+      "step": 4202
+    },
+    {
+      "epoch": 7.99,
+      "grad_norm": 0.029102910310029984,
+      "learning_rate": 6.133022165841146e-05,
+      "loss": 0.0006,
+      "step": 4203
+    },
+    {
+      "epoch": 7.99,
+      "grad_norm": 0.14735791087150574,
+      "learning_rate": 6.121834683755558e-05,
+      "loss": 0.0071,
+      "step": 4204
+    },
+    {
+      "epoch": 7.99,
+      "grad_norm": 0.03072136640548706,
+      "learning_rate": 6.110656255238154e-05,
+      "loss": 0.0007,
+      "step": 4205
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 0.058113355189561844,
+      "learning_rate": 6.099486884527346e-05,
+      "loss": 0.0014,
+      "step": 4206
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 0.2486698478460312,
+      "learning_rate": 6.0883265758580615e-05,
+      "loss": 0.0094,
+      "step": 4207
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 0.08145719021558762,
+      "learning_rate": 6.077175333461809e-05,
+      "loss": 0.0017,
+      "step": 4208
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 0.01730223000049591,
+      "learning_rate": 6.06603316156666e-05,
+      "loss": 0.0004,
+      "step": 4209
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 0.08303141593933105,
+      "learning_rate": 6.054900064397242e-05,
+      "loss": 0.0033,
+      "step": 4210
+    },
+    {
+      "epoch": 8.01,
+      "grad_norm": 0.0773901715874672,
+      "learning_rate": 6.043776046174742e-05,
+      "loss": 0.0017,
+      "step": 4211
+    },
+    {
+      "epoch": 8.01,
+      "grad_norm": 0.20950832962989807,
+      "learning_rate": 6.032661111116905e-05,
+      "loss": 0.0063,
+      "step": 4212
+    },
+    {
+      "epoch": 8.01,
+      "grad_norm": 0.041525948792696,
+      "learning_rate": 6.021555263438052e-05,
+      "loss": 0.0011,
+      "step": 4213
+    },
+    {
+      "epoch": 8.01,
+      "grad_norm": 0.06181616336107254,
+      "learning_rate": 6.0104585073490116e-05,
+      "loss": 0.0017,
+      "step": 4214
+    },
+    {
+      "epoch": 8.01,
+      "grad_norm": 0.09867484867572784,
+      "learning_rate": 5.999370847057221e-05,
+      "loss": 0.0046,
+      "step": 4215
+    },
+    {
+      "epoch": 8.02,
+      "grad_norm": 0.04409953951835632,
+      "learning_rate": 5.988292286766626e-05,
+      "loss": 0.0008,
+      "step": 4216
+    },
+    {
+      "epoch": 8.02,
+      "grad_norm": 0.04870391637086868,
+      "learning_rate": 5.977222830677744e-05,
+      "loss": 0.0004,
+      "step": 4217
+    },
+    {
+      "epoch": 8.02,
+      "grad_norm": 0.05668798089027405,
+      "learning_rate": 5.966162482987633e-05,
+      "loss": 0.001,
+      "step": 4218
+    },
+    {
+      "epoch": 8.02,
+      "grad_norm": 0.0252094604074955,
+      "learning_rate": 5.955111247889896e-05,
+      "loss": 0.0006,
+      "step": 4219
+    },
+    {
+      "epoch": 8.02,
+      "grad_norm": 0.19413913786411285,
+      "learning_rate": 5.9440691295746856e-05,
+      "loss": 0.0032,
+      "step": 4220
+    },
+    {
+      "epoch": 8.02,
+      "grad_norm": 0.06941913813352585,
+      "learning_rate": 5.933036132228698e-05,
+      "loss": 0.0009,
+      "step": 4221
+    },
+    {
+      "epoch": 8.03,
+      "grad_norm": 0.05203106626868248,
+      "learning_rate": 5.922012260035167e-05,
+      "loss": 0.0013,
+      "step": 4222
+    },
+    {
+      "epoch": 8.03,
+      "grad_norm": 0.05390901491045952,
+      "learning_rate": 5.91099751717386e-05,
+      "loss": 0.0009,
+      "step": 4223
+    },
+    {
+      "epoch": 8.03,
+      "grad_norm": 0.023860977962613106,
+      "learning_rate": 5.8999919078211156e-05,
+      "loss": 0.0007,
+      "step": 4224
+    },
+    {
+      "epoch": 8.03,
+      "grad_norm": 0.03017040714621544,
+      "learning_rate": 5.888995436149753e-05,
+      "loss": 0.0007,
+      "step": 4225
+    },
+    {
+      "epoch": 8.03,
+      "grad_norm": 0.020714610815048218,
+      "learning_rate": 5.878008106329183e-05,
+      "loss": 0.0006,
+      "step": 4226
+    },
+    {
+      "epoch": 8.04,
+      "grad_norm": 0.009745297953486443,
+      "learning_rate": 5.86702992252532e-05,
+      "loss": 0.0003,
+      "step": 4227
+    },
+    {
+      "epoch": 8.04,
+      "grad_norm": 0.07059819996356964,
+      "learning_rate": 5.8560608889006125e-05,
+      "loss": 0.0015,
+      "step": 4228
+    },
+    {
+      "epoch": 8.04,
+      "grad_norm": 0.03880246728658676,
+      "learning_rate": 5.845101009614048e-05,
+      "loss": 0.0008,
+      "step": 4229
+    },
+    {
+      "epoch": 8.04,
+      "grad_norm": 0.12333130836486816,
+      "learning_rate": 5.834150288821141e-05,
+      "loss": 0.0018,
+      "step": 4230
+    },
+    {
+      "epoch": 8.04,
+      "grad_norm": 0.06224041432142258,
+      "learning_rate": 5.8232087306739225e-05,
+      "loss": 0.0012,
+      "step": 4231
+    },
+    {
+      "epoch": 8.05,
+      "grad_norm": 0.08667173981666565,
+      "learning_rate": 5.812276339320961e-05,
+      "loss": 0.0027,
+      "step": 4232
+    },
+    {
+      "epoch": 8.05,
+      "grad_norm": 0.02627207525074482,
+      "learning_rate": 5.8013531189073645e-05,
+      "loss": 0.0008,
+      "step": 4233
+    },
+    {
+      "epoch": 8.05,
+      "grad_norm": 0.014420422725379467,
+      "learning_rate": 5.790439073574713e-05,
+      "loss": 0.0003,
+      "step": 4234
+    },
+    {
+      "epoch": 8.05,
+      "grad_norm": 0.05990182235836983,
+      "learning_rate": 5.77953420746117e-05,
+      "loss": 0.0016,
+      "step": 4235
+    },
+    {
+      "epoch": 8.05,
+      "grad_norm": 0.13517257571220398,
+      "learning_rate": 5.7686385247013763e-05,
+      "loss": 0.0028,
+      "step": 4236
+    },
+    {
+      "epoch": 8.06,
+      "grad_norm": 0.029454976320266724,
+      "learning_rate": 5.75775202942651e-05,
+      "loss": 0.0006,
+      "step": 4237
+    },
+    {
+      "epoch": 8.06,
+      "grad_norm": 0.03725656867027283,
+      "learning_rate": 5.746874725764244e-05,
+      "loss": 0.0012,
+      "step": 4238
+    },
+    {
+      "epoch": 8.06,
+      "grad_norm": 0.36969083547592163,
+      "learning_rate": 5.736006617838811e-05,
+      "loss": 0.0031,
+      "step": 4239
+    },
+    {
+      "epoch": 8.06,
+      "grad_norm": 0.08323214203119278,
+      "learning_rate": 5.725147709770897e-05,
+      "loss": 0.0031,
+      "step": 4240
+    },
+    {
+      "epoch": 8.06,
+      "grad_norm": 0.015371928922832012,
+      "learning_rate": 5.7142980056777476e-05,
+      "loss": 0.0004,
+      "step": 4241
+    },
+    {
+      "epoch": 8.06,
+      "grad_norm": 0.05135523900389671,
+      "learning_rate": 5.7034575096730976e-05,
+      "loss": 0.0008,
+      "step": 4242
+    },
+    {
+      "epoch": 8.07,
+      "grad_norm": 0.017483821138739586,
+      "learning_rate": 5.692626225867197e-05,
+      "loss": 0.0003,
+      "step": 4243
+    },
+    {
+      "epoch": 8.07,
+      "grad_norm": 0.026310743764042854,
+      "learning_rate": 5.6818041583667964e-05,
+      "loss": 0.0005,
+      "step": 4244
+    },
+    {
+      "epoch": 8.07,
+      "grad_norm": 0.01471678726375103,
+      "learning_rate": 5.670991311275162e-05,
+      "loss": 0.0003,
+      "step": 4245
+    },
+    {
+      "epoch": 8.07,
+      "grad_norm": 0.012983615510165691,
+      "learning_rate": 5.660187688692047e-05,
+      "loss": 0.0003,
+      "step": 4246
+    },
+    {
+      "epoch": 8.07,
+      "grad_norm": 0.11409618705511093,
+      "learning_rate": 5.64939329471373e-05,
+      "loss": 0.0021,
+      "step": 4247
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 0.01801317371428013,
+      "learning_rate": 5.6386081334329725e-05,
+      "loss": 0.0004,
+      "step": 4248
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 0.04905932769179344,
+      "learning_rate": 5.627832208939037e-05,
+      "loss": 0.0018,
+      "step": 4249
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 0.11396396160125732,
+      "learning_rate": 5.6170655253177055e-05,
+      "loss": 0.0028,
+      "step": 4250
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 0.07774129509925842,
+      "learning_rate": 5.606308086651216e-05,
+      "loss": 0.0016,
+      "step": 4251
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 0.04101789742708206,
+      "learning_rate": 5.595559897018341e-05,
+      "loss": 0.0011,
+      "step": 4252
+    },
+    {
+      "epoch": 8.09,
+      "grad_norm": 0.020992033183574677,
+      "learning_rate": 5.584820960494323e-05,
+      "loss": 0.0003,
+      "step": 4253
+    },
+    {
+      "epoch": 8.09,
+      "grad_norm": 0.04150540754199028,
+      "learning_rate": 5.574091281150906e-05,
+      "loss": 0.0013,
+      "step": 4254
+    },
+    {
+      "epoch": 8.09,
+      "grad_norm": 0.09226648509502411,
+      "learning_rate": 5.563370863056318e-05,
+      "loss": 0.0015,
+      "step": 4255
+    },
+    {
+      "epoch": 8.09,
+      "grad_norm": 0.026781700551509857,
+      "learning_rate": 5.552659710275279e-05,
+      "loss": 0.0009,
+      "step": 4256
+    },
+    {
+      "epoch": 8.09,
+      "grad_norm": 0.010981214232742786,
+      "learning_rate": 5.5419578268689944e-05,
+      "loss": 0.0002,
+      "step": 4257
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 0.1364150047302246,
+      "learning_rate": 5.531265216895149e-05,
+      "loss": 0.004,
+      "step": 4258
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 0.07817291468381882,
+      "learning_rate": 5.520581884407942e-05,
+      "loss": 0.0009,
+      "step": 4259
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 0.12471527606248856,
+      "learning_rate": 5.509907833457997e-05,
+      "loss": 0.0036,
+      "step": 4260
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 0.01729252189397812,
+      "learning_rate": 5.499243068092477e-05,
+      "loss": 0.0004,
+      "step": 4261
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 0.04824591428041458,
+      "learning_rate": 5.4885875923549974e-05,
+      "loss": 0.0012,
+      "step": 4262
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 0.11739405244588852,
+      "learning_rate": 5.477941410285649e-05,
+      "loss": 0.0017,
+      "step": 4263
+    },
+    {
+      "epoch": 8.11,
+      "grad_norm": 0.19470249116420746,
+      "learning_rate": 5.4673045259210066e-05,
+      "loss": 0.0075,
+      "step": 4264
+    },
+    {
+      "epoch": 8.11,
+      "grad_norm": 0.05893266573548317,
+      "learning_rate": 5.4566769432941114e-05,
+      "loss": 0.0012,
+      "step": 4265
+    },
+    {
+      "epoch": 8.11,
+      "grad_norm": 0.043742720037698746,
+      "learning_rate": 5.446058666434493e-05,
+      "loss": 0.001,
+      "step": 4266
+    },
+    {
+      "epoch": 8.11,
+      "grad_norm": 0.00886616576462984,
+      "learning_rate": 5.435449699368134e-05,
+      "loss": 0.0003,
+      "step": 4267
+    },
+    {
+      "epoch": 8.11,
+      "grad_norm": 0.03803609311580658,
+      "learning_rate": 5.4248500461175004e-05,
+      "loss": 0.0009,
+      "step": 4268
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 0.058297544717788696,
+      "learning_rate": 5.414259710701514e-05,
+      "loss": 0.0005,
+      "step": 4269
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 0.05710463225841522,
+      "learning_rate": 5.403678697135596e-05,
+      "loss": 0.001,
+      "step": 4270
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 0.04109049215912819,
+      "learning_rate": 5.393107009431581e-05,
+      "loss": 0.0009,
+      "step": 4271
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 0.07118772715330124,
+      "learning_rate": 5.382544651597812e-05,
+      "loss": 0.0012,
+      "step": 4272
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 0.012347332201898098,
+      "learning_rate": 5.3719916276390765e-05,
+      "loss": 0.0003,
+      "step": 4273
+    },
+    {
+      "epoch": 8.13,
+      "grad_norm": 0.11072738468647003,
+      "learning_rate": 5.361447941556626e-05,
+      "loss": 0.0021,
+      "step": 4274
+    },
+    {
+      "epoch": 8.13,
+      "grad_norm": 0.07087287306785583,
+      "learning_rate": 5.3509135973481666e-05,
+      "loss": 0.0018,
+      "step": 4275
+    },
+    {
+      "epoch": 8.13,
+      "grad_norm": 0.024044062942266464,
+      "learning_rate": 5.340388599007866e-05,
+      "loss": 0.0003,
+      "step": 4276
+    },
+    {
+      "epoch": 8.13,
+      "grad_norm": 0.07172052562236786,
+      "learning_rate": 5.329872950526354e-05,
+      "loss": 0.0031,
+      "step": 4277
+    },
+    {
+      "epoch": 8.13,
+      "grad_norm": 0.04435458779335022,
+      "learning_rate": 5.319366655890696e-05,
+      "loss": 0.0005,
+      "step": 4278
+    },
+    {
+      "epoch": 8.13,
+      "grad_norm": 0.08529604971408844,
+      "learning_rate": 5.308869719084452e-05,
+      "loss": 0.001,
+      "step": 4279
+    },
+    {
+      "epoch": 8.14,
+      "grad_norm": 0.0468132346868515,
+      "learning_rate": 5.298382144087576e-05,
+      "loss": 0.0009,
+      "step": 4280
+    },
+    {
+      "epoch": 8.14,
+      "grad_norm": 0.03188987448811531,
+      "learning_rate": 5.287903934876524e-05,
+      "loss": 0.0005,
+      "step": 4281
+    },
+    {
+      "epoch": 8.14,
+      "grad_norm": 0.022515878081321716,
+      "learning_rate": 5.277435095424171e-05,
+      "loss": 0.0007,
+      "step": 4282
+    },
+    {
+      "epoch": 8.14,
+      "grad_norm": 0.04725012183189392,
+      "learning_rate": 5.2669756296998545e-05,
+      "loss": 0.0005,
+      "step": 4283
+    },
+    {
+      "epoch": 8.14,
+      "grad_norm": 0.02353043295443058,
+      "learning_rate": 5.25652554166934e-05,
+      "loss": 0.0003,
+      "step": 4284
+    },
+    {
+      "epoch": 8.15,
+      "grad_norm": 0.050125449895858765,
+      "learning_rate": 5.2460848352948704e-05,
+      "loss": 0.0015,
+      "step": 4285
+    },
+    {
+      "epoch": 8.15,
+      "grad_norm": 0.07164198160171509,
+      "learning_rate": 5.235653514535083e-05,
+      "loss": 0.0042,
+      "step": 4286
+    },
+    {
+      "epoch": 8.15,
+      "grad_norm": 0.0601380355656147,
+      "learning_rate": 5.225231583345104e-05,
+      "loss": 0.0024,
+      "step": 4287
+    },
+    {
+      "epoch": 8.15,
+      "grad_norm": 0.02986823208630085,
+      "learning_rate": 5.214819045676476e-05,
+      "loss": 0.0007,
+      "step": 4288
+    },
+    {
+      "epoch": 8.15,
+      "grad_norm": 0.07346397638320923,
+      "learning_rate": 5.204415905477177e-05,
+      "loss": 0.0015,
+      "step": 4289
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 0.03785846009850502,
+      "learning_rate": 5.194022166691636e-05,
+      "loss": 0.0007,
+      "step": 4290
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 0.04165830463171005,
+      "learning_rate": 5.183637833260703e-05,
+      "loss": 0.0022,
+      "step": 4291
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 0.021388184279203415,
+      "learning_rate": 5.173262909121669e-05,
+      "loss": 0.0005,
+      "step": 4292
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 0.08778490126132965,
+      "learning_rate": 5.1628973982082575e-05,
+      "loss": 0.0016,
+      "step": 4293
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 0.05691933631896973,
+      "learning_rate": 5.152541304450625e-05,
+      "loss": 0.001,
+      "step": 4294
+    },
+    {
+      "epoch": 8.17,
+      "grad_norm": 0.06112895905971527,
+      "learning_rate": 5.142194631775346e-05,
+      "loss": 0.001,
+      "step": 4295
+    },
+    {
+      "epoch": 8.17,
+      "grad_norm": 0.09804000705480576,
+      "learning_rate": 5.13185738410545e-05,
+      "loss": 0.0024,
+      "step": 4296
+    },
+    {
+      "epoch": 8.17,
+      "grad_norm": 0.017314493656158447,
+      "learning_rate": 5.121529565360352e-05,
+      "loss": 0.0006,
+      "step": 4297
+    },
+    {
+      "epoch": 8.17,
+      "grad_norm": 0.021534794941544533,
+      "learning_rate": 5.111211179455929e-05,
+      "loss": 0.0004,
+      "step": 4298
+    },
+    {
+      "epoch": 8.17,
+      "grad_norm": 0.06889753043651581,
+      "learning_rate": 5.1009022303044644e-05,
+      "loss": 0.0014,
+      "step": 4299
+    },
+    {
+      "epoch": 8.17,
+      "grad_norm": 0.02045854553580284,
+      "learning_rate": 5.090602721814663e-05,
+      "loss": 0.0006,
+      "step": 4300
+    },
+    {
+      "epoch": 8.17,
+      "eval_blimp_filtered_avg": 0.5446268656716418,
+      "eval_blimp_filtered_std": 0.005854199684154884,
+      "step": 4300
+    },
+    {
+      "epoch": 8.17,
+      "eval_blimp_supplement_avg": 0.5538793103448276,
+      "eval_blimp_supplement_std": 0.023107506957053266,
+      "step": 4300
+    },
+    {
+      "epoch": 8.17,
+      "eval_vqa_filtered_avg": 0.15,
+      "eval_vqa_filtered_std": 0.03588702812826371,
+      "step": 4300
+    },
+    {
+      "epoch": 8.17,
+      "eval_winoground_filtered_avg": 0.51,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 4300
+    },
+    {
+      "epoch": 8.18,
+      "grad_norm": 0.055534545332193375,
+      "learning_rate": 5.0803126578916566e-05,
+      "loss": 0.0008,
+      "step": 4301
+    },
+    {
+      "epoch": 8.18,
+      "grad_norm": 0.06010944023728371,
+      "learning_rate": 5.070032042436991e-05,
+      "loss": 0.002,
+      "step": 4302
+    },
+    {
+      "epoch": 8.18,
+      "grad_norm": 0.010478295385837555,
+      "learning_rate": 5.059760879348629e-05,
+      "loss": 0.0003,
+      "step": 4303
+    },
+    {
+      "epoch": 8.18,
+      "grad_norm": 0.044492509216070175,
+      "learning_rate": 5.049499172520951e-05,
+      "loss": 0.001,
+      "step": 4304
+    },
+    {
+      "epoch": 8.18,
+      "grad_norm": 0.07323823869228363,
+      "learning_rate": 5.039246925844765e-05,
+      "loss": 0.0013,
+      "step": 4305
+    },
+    {
+      "epoch": 8.19,
+      "grad_norm": 0.01724736951291561,
+      "learning_rate": 5.029004143207259e-05,
+      "loss": 0.0005,
+      "step": 4306
+    },
+    {
+      "epoch": 8.19,
+      "grad_norm": 0.26461127400398254,
+      "learning_rate": 5.0187708284920724e-05,
+      "loss": 0.0091,
+      "step": 4307
+    },
+    {
+      "epoch": 8.19,
+      "grad_norm": 0.09975370764732361,
+      "learning_rate": 5.008546985579228e-05,
+      "loss": 0.0013,
+      "step": 4308
+    },
+    {
+      "epoch": 8.19,
+      "grad_norm": 0.10302262008190155,
+      "learning_rate": 4.9983326183451654e-05,
+      "loss": 0.0039,
+      "step": 4309
+    },
+    {
+      "epoch": 8.19,
+      "grad_norm": 0.08520229160785675,
+      "learning_rate": 4.988127730662733e-05,
+      "loss": 0.0064,
+      "step": 4310
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 0.10618811100721359,
+      "learning_rate": 4.97793232640118e-05,
+      "loss": 0.0036,
+      "step": 4311
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 0.05473819747567177,
+      "learning_rate": 4.9677464094261656e-05,
+      "loss": 0.0009,
+      "step": 4312
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 0.019265465438365936,
+      "learning_rate": 4.9575699835997386e-05,
+      "loss": 0.0004,
+      "step": 4313
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 0.05932788923382759,
+      "learning_rate": 4.947403052780385e-05,
+      "loss": 0.0006,
+      "step": 4314
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 0.01034677866846323,
+      "learning_rate": 4.937245620822933e-05,
+      "loss": 0.0003,
+      "step": 4315
+    },
+    {
+      "epoch": 8.21,
+      "grad_norm": 0.08104673773050308,
+      "learning_rate": 4.9270976915786765e-05,
+      "loss": 0.0036,
+      "step": 4316
+    },
+    {
+      "epoch": 8.21,
+      "grad_norm": 0.05957690253853798,
+      "learning_rate": 4.9169592688952383e-05,
+      "loss": 0.0009,
+      "step": 4317
+    },
+    {
+      "epoch": 8.21,
+      "grad_norm": 0.028382649645209312,
+      "learning_rate": 4.906830356616691e-05,
+      "loss": 0.0006,
+      "step": 4318
+    },
+    {
+      "epoch": 8.21,
+      "grad_norm": 0.029100971296429634,
+      "learning_rate": 4.896710958583475e-05,
+      "loss": 0.0003,
+      "step": 4319
+    },
+    {
+      "epoch": 8.21,
+      "grad_norm": 0.07133439183235168,
+      "learning_rate": 4.886601078632432e-05,
+      "loss": 0.0014,
+      "step": 4320
+    },
+    {
+      "epoch": 8.21,
+      "grad_norm": 0.009207689203321934,
+      "learning_rate": 4.876500720596784e-05,
+      "loss": 0.0002,
+      "step": 4321
+    },
+    {
+      "epoch": 8.22,
+      "grad_norm": 0.011953119188547134,
+      "learning_rate": 4.866409888306156e-05,
+      "loss": 0.0003,
+      "step": 4322
+    },
+    {
+      "epoch": 8.22,
+      "grad_norm": 0.013479550369083881,
+      "learning_rate": 4.8563285855865545e-05,
+      "loss": 0.0003,
+      "step": 4323
+    },
+    {
+      "epoch": 8.22,
+      "grad_norm": 0.013635807670652866,
+      "learning_rate": 4.846256816260363e-05,
+      "loss": 0.0004,
+      "step": 4324
+    },
+    {
+      "epoch": 8.22,
+      "grad_norm": 0.020905883982777596,
+      "learning_rate": 4.836194584146389e-05,
+      "loss": 0.0005,
+      "step": 4325
+    },
+    {
+      "epoch": 8.22,
+      "grad_norm": 0.06611500680446625,
+      "learning_rate": 4.8261418930597594e-05,
+      "loss": 0.0008,
+      "step": 4326
+    },
+    {
+      "epoch": 8.23,
+      "grad_norm": 0.03861399367451668,
+      "learning_rate": 4.8160987468120496e-05,
+      "loss": 0.0009,
+      "step": 4327
+    },
+    {
+      "epoch": 8.23,
+      "grad_norm": 0.030351996421813965,
+      "learning_rate": 4.806065149211172e-05,
+      "loss": 0.0006,
+      "step": 4328
+    },
+    {
+      "epoch": 8.23,
+      "grad_norm": 0.04442434385418892,
+      "learning_rate": 4.796041104061439e-05,
+      "loss": 0.0004,
+      "step": 4329
+    },
+    {
+      "epoch": 8.23,
+      "grad_norm": 0.09961090236902237,
+      "learning_rate": 4.786026615163528e-05,
+      "loss": 0.0061,
+      "step": 4330
+    },
+    {
+      "epoch": 8.23,
+      "grad_norm": 0.032169148325920105,
+      "learning_rate": 4.7760216863145175e-05,
+      "loss": 0.0006,
+      "step": 4331
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 0.02211141027510166,
+      "learning_rate": 4.766026321307818e-05,
+      "loss": 0.0004,
+      "step": 4332
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 0.08686751127243042,
+      "learning_rate": 4.756040523933261e-05,
+      "loss": 0.0018,
+      "step": 4333
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 0.103911392390728,
+      "learning_rate": 4.746064297977019e-05,
+      "loss": 0.0014,
+      "step": 4334
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 0.015344129875302315,
+      "learning_rate": 4.736097647221653e-05,
+      "loss": 0.0005,
+      "step": 4335
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 0.13901478052139282,
+      "learning_rate": 4.7261405754460826e-05,
+      "loss": 0.0031,
+      "step": 4336
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 0.16673126816749573,
+      "learning_rate": 4.716193086425595e-05,
+      "loss": 0.0066,
+      "step": 4337
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 0.09676972776651382,
+      "learning_rate": 4.706255183931853e-05,
+      "loss": 0.0023,
+      "step": 4338
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 0.046336449682712555,
+      "learning_rate": 4.696326871732869e-05,
+      "loss": 0.0013,
+      "step": 4339
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 0.03479794040322304,
+      "learning_rate": 4.6864081535930565e-05,
+      "loss": 0.0008,
+      "step": 4340
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 0.012033471837639809,
+      "learning_rate": 4.676499033273132e-05,
+      "loss": 0.0002,
+      "step": 4341
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 0.0974738597869873,
+      "learning_rate": 4.6665995145302346e-05,
+      "loss": 0.0028,
+      "step": 4342
+    },
+    {
+      "epoch": 8.26,
+      "grad_norm": 0.010699262842535973,
+      "learning_rate": 4.656709601117808e-05,
+      "loss": 0.0002,
+      "step": 4343
+    },
+    {
+      "epoch": 8.26,
+      "grad_norm": 0.03722322732210159,
+      "learning_rate": 4.646829296785697e-05,
+      "loss": 0.0006,
+      "step": 4344
+    },
+    {
+      "epoch": 8.26,
+      "grad_norm": 0.026214975863695145,
+      "learning_rate": 4.63695860528008e-05,
+      "loss": 0.0005,
+      "step": 4345
+    },
+    {
+      "epoch": 8.26,
+      "grad_norm": 0.007844100706279278,
+      "learning_rate": 4.627097530343501e-05,
+      "loss": 0.0002,
+      "step": 4346
+    },
+    {
+      "epoch": 8.26,
+      "grad_norm": 0.02118430659174919,
+      "learning_rate": 4.617246075714849e-05,
+      "loss": 0.0004,
+      "step": 4347
+    },
+    {
+      "epoch": 8.27,
+      "grad_norm": 0.09241119027137756,
+      "learning_rate": 4.6074042451293694e-05,
+      "loss": 0.0029,
+      "step": 4348
+    },
+    {
+      "epoch": 8.27,
+      "grad_norm": 0.012289599515497684,
+      "learning_rate": 4.59757204231866e-05,
+      "loss": 0.0002,
+      "step": 4349
+    },
+    {
+      "epoch": 8.27,
+      "grad_norm": 0.0178693775087595,
+      "learning_rate": 4.5877494710106635e-05,
+      "loss": 0.0004,
+      "step": 4350
+    },
+    {
+      "epoch": 8.27,
+      "grad_norm": 0.06423716247081757,
+      "learning_rate": 4.5779365349296894e-05,
+      "loss": 0.0007,
+      "step": 4351
+    },
+    {
+      "epoch": 8.27,
+      "grad_norm": 0.011614951305091381,
+      "learning_rate": 4.568133237796355e-05,
+      "loss": 0.0003,
+      "step": 4352
+    },
+    {
+      "epoch": 8.28,
+      "grad_norm": 0.052367158234119415,
+      "learning_rate": 4.5583395833276645e-05,
+      "loss": 0.0009,
+      "step": 4353
+    },
+    {
+      "epoch": 8.28,
+      "grad_norm": 0.08815959095954895,
+      "learning_rate": 4.5485555752369474e-05,
+      "loss": 0.0019,
+      "step": 4354
+    },
+    {
+      "epoch": 8.28,
+      "grad_norm": 0.01068867463618517,
+      "learning_rate": 4.5387812172338725e-05,
+      "loss": 0.0003,
+      "step": 4355
+    },
+    {
+      "epoch": 8.28,
+      "grad_norm": 0.02823864482343197,
+      "learning_rate": 4.5290165130244516e-05,
+      "loss": 0.0005,
+      "step": 4356
+    },
+    {
+      "epoch": 8.28,
+      "grad_norm": 0.01555906143039465,
+      "learning_rate": 4.519261466311044e-05,
+      "loss": 0.0003,
+      "step": 4357
+    },
+    {
+      "epoch": 8.29,
+      "grad_norm": 0.17226672172546387,
+      "learning_rate": 4.509516080792337e-05,
+      "loss": 0.0111,
+      "step": 4358
+    },
+    {
+      "epoch": 8.29,
+      "grad_norm": 0.09022840857505798,
+      "learning_rate": 4.499780360163352e-05,
+      "loss": 0.0027,
+      "step": 4359
+    },
+    {
+      "epoch": 8.29,
+      "grad_norm": 0.012838329188525677,
+      "learning_rate": 4.490054308115483e-05,
+      "loss": 0.0002,
+      "step": 4360
+    },
+    {
+      "epoch": 8.29,
+      "grad_norm": 0.05284389108419418,
+      "learning_rate": 4.4803379283363896e-05,
+      "loss": 0.0007,
+      "step": 4361
+    },
+    {
+      "epoch": 8.29,
+      "grad_norm": 0.033422037959098816,
+      "learning_rate": 4.470631224510132e-05,
+      "loss": 0.0004,
+      "step": 4362
+    },
+    {
+      "epoch": 8.29,
+      "grad_norm": 0.008204043842852116,
+      "learning_rate": 4.460934200317059e-05,
+      "loss": 0.0002,
+      "step": 4363
+    },
+    {
+      "epoch": 8.3,
+      "grad_norm": 0.0778932049870491,
+      "learning_rate": 4.45124685943387e-05,
+      "loss": 0.002,
+      "step": 4364
+    },
+    {
+      "epoch": 8.3,
+      "grad_norm": 0.03624528646469116,
+      "learning_rate": 4.44156920553358e-05,
+      "loss": 0.0006,
+      "step": 4365
+    },
+    {
+      "epoch": 8.3,
+      "grad_norm": 0.046308085322380066,
+      "learning_rate": 4.4319012422855415e-05,
+      "loss": 0.001,
+      "step": 4366
+    },
+    {
+      "epoch": 8.3,
+      "grad_norm": 0.011219562031328678,
+      "learning_rate": 4.4222429733554274e-05,
+      "loss": 0.0003,
+      "step": 4367
+    },
+    {
+      "epoch": 8.3,
+      "grad_norm": 0.03580864518880844,
+      "learning_rate": 4.412594402405233e-05,
+      "loss": 0.0009,
+      "step": 4368
+    },
+    {
+      "epoch": 8.31,
+      "grad_norm": 0.058891791850328445,
+      "learning_rate": 4.4029555330932845e-05,
+      "loss": 0.001,
+      "step": 4369
+    },
+    {
+      "epoch": 8.31,
+      "grad_norm": 0.03365139663219452,
+      "learning_rate": 4.3933263690742115e-05,
+      "loss": 0.0008,
+      "step": 4370
+    },
+    {
+      "epoch": 8.31,
+      "grad_norm": 0.03657792881131172,
+      "learning_rate": 4.383706913999001e-05,
+      "loss": 0.0005,
+      "step": 4371
+    },
+    {
+      "epoch": 8.31,
+      "grad_norm": 0.03834382817149162,
+      "learning_rate": 4.374097171514905e-05,
+      "loss": 0.0006,
+      "step": 4372
+    },
+    {
+      "epoch": 8.31,
+      "grad_norm": 0.00687333894893527,
+      "learning_rate": 4.3644971452655465e-05,
+      "loss": 0.0002,
+      "step": 4373
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 0.033968765288591385,
+      "learning_rate": 4.3549068388908326e-05,
+      "loss": 0.0006,
+      "step": 4374
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 0.045775312930345535,
+      "learning_rate": 4.3453262560269884e-05,
+      "loss": 0.0012,
+      "step": 4375
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 0.03564378619194031,
+      "learning_rate": 4.335755400306559e-05,
+      "loss": 0.0009,
+      "step": 4376
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 0.02691163308918476,
+      "learning_rate": 4.32619427535841e-05,
+      "loss": 0.0005,
+      "step": 4377
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 0.07282459735870361,
+      "learning_rate": 4.316642884807686e-05,
+      "loss": 0.0009,
+      "step": 4378
+    },
+    {
+      "epoch": 8.33,
+      "grad_norm": 0.0641508623957634,
+      "learning_rate": 4.307101232275877e-05,
+      "loss": 0.0022,
+      "step": 4379
+    },
+    {
+      "epoch": 8.33,
+      "grad_norm": 0.13768525421619415,
+      "learning_rate": 4.297569321380765e-05,
+      "loss": 0.0031,
+      "step": 4380
+    },
+    {
+      "epoch": 8.33,
+      "grad_norm": 0.07876940071582794,
+      "learning_rate": 4.288047155736429e-05,
+      "loss": 0.0028,
+      "step": 4381
+    },
+    {
+      "epoch": 8.33,
+      "grad_norm": 0.17705893516540527,
+      "learning_rate": 4.278534738953275e-05,
+      "loss": 0.0056,
+      "step": 4382
+    },
+    {
+      "epoch": 8.33,
+      "grad_norm": 0.0933384969830513,
+      "learning_rate": 4.269032074637988e-05,
+      "loss": 0.002,
+      "step": 4383
+    },
+    {
+      "epoch": 8.33,
+      "grad_norm": 0.17234458029270172,
+      "learning_rate": 4.259539166393578e-05,
+      "loss": 0.0052,
+      "step": 4384
+    },
+    {
+      "epoch": 8.34,
+      "grad_norm": 0.09836405515670776,
+      "learning_rate": 4.250056017819331e-05,
+      "loss": 0.0029,
+      "step": 4385
+    },
+    {
+      "epoch": 8.34,
+      "grad_norm": 0.019654318690299988,
+      "learning_rate": 4.240582632510869e-05,
+      "loss": 0.0004,
+      "step": 4386
+    },
+    {
+      "epoch": 8.34,
+      "grad_norm": 0.00906306877732277,
+      "learning_rate": 4.231119014060067e-05,
+      "loss": 0.0002,
+      "step": 4387
+    },
+    {
+      "epoch": 8.34,
+      "grad_norm": 0.06786053627729416,
+      "learning_rate": 4.221665166055137e-05,
+      "loss": 0.0024,
+      "step": 4388
+    },
+    {
+      "epoch": 8.34,
+      "grad_norm": 0.09945710748434067,
+      "learning_rate": 4.2122210920805657e-05,
+      "loss": 0.001,
+      "step": 4389
+    },
+    {
+      "epoch": 8.35,
+      "grad_norm": 0.08035233616828918,
+      "learning_rate": 4.202786795717135e-05,
+      "loss": 0.0018,
+      "step": 4390
+    },
+    {
+      "epoch": 8.35,
+      "grad_norm": 0.10572786629199982,
+      "learning_rate": 4.1933622805419196e-05,
+      "loss": 0.004,
+      "step": 4391
+    },
+    {
+      "epoch": 8.35,
+      "grad_norm": 0.022922731935977936,
+      "learning_rate": 4.183947550128296e-05,
+      "loss": 0.0005,
+      "step": 4392
+    },
+    {
+      "epoch": 8.35,
+      "grad_norm": 0.039655156433582306,
+      "learning_rate": 4.1745426080459144e-05,
+      "loss": 0.0008,
+      "step": 4393
+    },
+    {
+      "epoch": 8.35,
+      "grad_norm": 0.03428768739104271,
+      "learning_rate": 4.165147457860729e-05,
+      "loss": 0.001,
+      "step": 4394
+    },
+    {
+      "epoch": 8.36,
+      "grad_norm": 0.013347703963518143,
+      "learning_rate": 4.155762103134974e-05,
+      "loss": 0.0002,
+      "step": 4395
+    },
+    {
+      "epoch": 8.36,
+      "grad_norm": 0.040046025067567825,
+      "learning_rate": 4.1463865474271595e-05,
+      "loss": 0.0005,
+      "step": 4396
+    },
+    {
+      "epoch": 8.36,
+      "grad_norm": 0.08258210867643356,
+      "learning_rate": 4.1370207942921154e-05,
+      "loss": 0.0035,
+      "step": 4397
+    },
+    {
+      "epoch": 8.36,
+      "grad_norm": 0.060599200427532196,
+      "learning_rate": 4.127664847280897e-05,
+      "loss": 0.0013,
+      "step": 4398
+    },
+    {
+      "epoch": 8.36,
+      "grad_norm": 0.10431927442550659,
+      "learning_rate": 4.118318709940904e-05,
+      "loss": 0.001,
+      "step": 4399
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 0.012674281373620033,
+      "learning_rate": 4.108982385815775e-05,
+      "loss": 0.0003,
+      "step": 4400
+    },
+    {
+      "epoch": 8.37,
+      "eval_blimp_filtered_avg": 0.5434328358208955,
+      "eval_blimp_filtered_std": 0.005856274905660744,
+      "step": 4400
+    },
+    {
+      "epoch": 8.37,
+      "eval_blimp_supplement_avg": 0.5517241379310345,
+      "eval_blimp_supplement_std": 0.023136888272642143,
+      "step": 4400
+    },
+    {
+      "epoch": 8.37,
+      "eval_vqa_filtered_avg": 0.14,
+      "eval_vqa_filtered_std": 0.03487350880197771,
+      "step": 4400
+    },
+    {
+      "epoch": 8.37,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 4400
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 0.029184820130467415,
+      "learning_rate": 4.0996558784454456e-05,
+      "loss": 0.0005,
+      "step": 4401
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 0.13495193421840668,
+      "learning_rate": 4.09033919136612e-05,
+      "loss": 0.0065,
+      "step": 4402
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 0.011668450199067593,
+      "learning_rate": 4.081032328110282e-05,
+      "loss": 0.0002,
+      "step": 4403
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 0.044473156332969666,
+      "learning_rate": 4.071735292206696e-05,
+      "loss": 0.0008,
+      "step": 4404
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 0.022545911371707916,
+      "learning_rate": 4.062448087180385e-05,
+      "loss": 0.0004,
+      "step": 4405
+    },
+    {
+      "epoch": 8.38,
+      "grad_norm": 0.06875132024288177,
+      "learning_rate": 4.053170716552681e-05,
+      "loss": 0.0014,
+      "step": 4406
+    },
+    {
+      "epoch": 8.38,
+      "grad_norm": 0.02562439627945423,
+      "learning_rate": 4.043903183841134e-05,
+      "loss": 0.0005,
+      "step": 4407
+    },
+    {
+      "epoch": 8.38,
+      "grad_norm": 0.030132809653878212,
+      "learning_rate": 4.0346454925596085e-05,
+      "loss": 0.0005,
+      "step": 4408
+    },
+    {
+      "epoch": 8.38,
+      "grad_norm": 0.015905573964118958,
+      "learning_rate": 4.025397646218216e-05,
+      "loss": 0.0003,
+      "step": 4409
+    },
+    {
+      "epoch": 8.38,
+      "grad_norm": 0.07482285052537918,
+      "learning_rate": 4.0161596483233415e-05,
+      "loss": 0.0005,
+      "step": 4410
+    },
+    {
+      "epoch": 8.39,
+      "grad_norm": 0.04825481399893761,
+      "learning_rate": 4.006931502377627e-05,
+      "loss": 0.0009,
+      "step": 4411
+    },
+    {
+      "epoch": 8.39,
+      "grad_norm": 0.08112642914056778,
+      "learning_rate": 3.997713211880007e-05,
+      "loss": 0.0008,
+      "step": 4412
+    },
+    {
+      "epoch": 8.39,
+      "grad_norm": 0.07325179874897003,
+      "learning_rate": 3.988504780325635e-05,
+      "loss": 0.001,
+      "step": 4413
+    },
+    {
+      "epoch": 8.39,
+      "grad_norm": 0.04721332713961601,
+      "learning_rate": 3.9793062112059626e-05,
+      "loss": 0.001,
+      "step": 4414
+    },
+    {
+      "epoch": 8.39,
+      "grad_norm": 0.09540034830570221,
+      "learning_rate": 3.9701175080086854e-05,
+      "loss": 0.0046,
+      "step": 4415
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 0.016756029799580574,
+      "learning_rate": 3.9609386742177576e-05,
+      "loss": 0.0003,
+      "step": 4416
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 0.02511107549071312,
+      "learning_rate": 3.951769713313417e-05,
+      "loss": 0.0005,
+      "step": 4417
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 0.03500399366021156,
+      "learning_rate": 3.9426106287721115e-05,
+      "loss": 0.0004,
+      "step": 4418
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 0.09632732719182968,
+      "learning_rate": 3.933461424066585e-05,
+      "loss": 0.0008,
+      "step": 4419
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 0.047380443662405014,
+      "learning_rate": 3.92432210266582e-05,
+      "loss": 0.0008,
+      "step": 4420
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 0.04805818200111389,
+      "learning_rate": 3.9151926680350475e-05,
+      "loss": 0.0013,
+      "step": 4421
+    },
+    {
+      "epoch": 8.41,
+      "grad_norm": 0.13765691220760345,
+      "learning_rate": 3.906073123635749e-05,
+      "loss": 0.0025,
+      "step": 4422
+    },
+    {
+      "epoch": 8.41,
+      "grad_norm": 0.10342316329479218,
+      "learning_rate": 3.896963472925685e-05,
+      "loss": 0.0046,
+      "step": 4423
+    },
+    {
+      "epoch": 8.41,
+      "grad_norm": 0.1158265769481659,
+      "learning_rate": 3.887863719358809e-05,
+      "loss": 0.0051,
+      "step": 4424
+    },
+    {
+      "epoch": 8.41,
+      "grad_norm": 0.010652510449290276,
+      "learning_rate": 3.8787738663853784e-05,
+      "loss": 0.0002,
+      "step": 4425
+    },
+    {
+      "epoch": 8.41,
+      "grad_norm": 0.13669829070568085,
+      "learning_rate": 3.869693917451863e-05,
+      "loss": 0.0048,
+      "step": 4426
+    },
+    {
+      "epoch": 8.42,
+      "grad_norm": 0.024357859045267105,
+      "learning_rate": 3.8606238760009914e-05,
+      "loss": 0.0005,
+      "step": 4427
+    },
+    {
+      "epoch": 8.42,
+      "grad_norm": 0.10841548442840576,
+      "learning_rate": 3.851563745471726e-05,
+      "loss": 0.0018,
+      "step": 4428
+    },
+    {
+      "epoch": 8.42,
+      "grad_norm": 0.05363420397043228,
+      "learning_rate": 3.842513529299277e-05,
+      "loss": 0.0008,
+      "step": 4429
+    },
+    {
+      "epoch": 8.42,
+      "grad_norm": 0.014884863048791885,
+      "learning_rate": 3.833473230915102e-05,
+      "loss": 0.0003,
+      "step": 4430
+    },
+    {
+      "epoch": 8.42,
+      "grad_norm": 0.029448531568050385,
+      "learning_rate": 3.8244428537468784e-05,
+      "loss": 0.0007,
+      "step": 4431
+    },
+    {
+      "epoch": 8.43,
+      "grad_norm": 0.020677851513028145,
+      "learning_rate": 3.815422401218555e-05,
+      "loss": 0.0005,
+      "step": 4432
+    },
+    {
+      "epoch": 8.43,
+      "grad_norm": 0.053480371832847595,
+      "learning_rate": 3.806411876750273e-05,
+      "loss": 0.001,
+      "step": 4433
+    },
+    {
+      "epoch": 8.43,
+      "grad_norm": 0.03507504239678383,
+      "learning_rate": 3.7974112837584544e-05,
+      "loss": 0.0004,
+      "step": 4434
+    },
+    {
+      "epoch": 8.43,
+      "grad_norm": 0.022233450785279274,
+      "learning_rate": 3.788420625655725e-05,
+      "loss": 0.0003,
+      "step": 4435
+    },
+    {
+      "epoch": 8.43,
+      "grad_norm": 0.030552612617611885,
+      "learning_rate": 3.779439905850956e-05,
+      "loss": 0.0004,
+      "step": 4436
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 0.026204507797956467,
+      "learning_rate": 3.7704691277492514e-05,
+      "loss": 0.0005,
+      "step": 4437
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 0.10012747347354889,
+      "learning_rate": 3.761508294751937e-05,
+      "loss": 0.0028,
+      "step": 4438
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 0.09680014103651047,
+      "learning_rate": 3.752557410256575e-05,
+      "loss": 0.0023,
+      "step": 4439
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 0.07710690051317215,
+      "learning_rate": 3.743616477656953e-05,
+      "loss": 0.0007,
+      "step": 4440
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 0.02869727648794651,
+      "learning_rate": 3.734685500343091e-05,
+      "loss": 0.0006,
+      "step": 4441
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 0.2002260386943817,
+      "learning_rate": 3.725764481701219e-05,
+      "loss": 0.004,
+      "step": 4442
+    },
+    {
+      "epoch": 8.45,
+      "grad_norm": 0.051229327917099,
+      "learning_rate": 3.7168534251138216e-05,
+      "loss": 0.0013,
+      "step": 4443
+    },
+    {
+      "epoch": 8.45,
+      "grad_norm": 0.05989079922437668,
+      "learning_rate": 3.707952333959559e-05,
+      "loss": 0.0012,
+      "step": 4444
+    },
+    {
+      "epoch": 8.45,
+      "grad_norm": 0.030590804293751717,
+      "learning_rate": 3.6990612116133635e-05,
+      "loss": 0.0006,
+      "step": 4445
+    },
+    {
+      "epoch": 8.45,
+      "grad_norm": 0.016818048432469368,
+      "learning_rate": 3.690180061446353e-05,
+      "loss": 0.0003,
+      "step": 4446
+    },
+    {
+      "epoch": 8.45,
+      "grad_norm": 0.04373672977089882,
+      "learning_rate": 3.681308886825881e-05,
+      "loss": 0.001,
+      "step": 4447
+    },
+    {
+      "epoch": 8.46,
+      "grad_norm": 0.026299884542822838,
+      "learning_rate": 3.672447691115511e-05,
+      "loss": 0.0006,
+      "step": 4448
+    },
+    {
+      "epoch": 8.46,
+      "grad_norm": 0.012565869837999344,
+      "learning_rate": 3.663596477675025e-05,
+      "loss": 0.0003,
+      "step": 4449
+    },
+    {
+      "epoch": 8.46,
+      "grad_norm": 0.02149553969502449,
+      "learning_rate": 3.654755249860418e-05,
+      "loss": 0.0005,
+      "step": 4450
+    },
+    {
+      "epoch": 8.46,
+      "grad_norm": 0.06040337309241295,
+      "learning_rate": 3.645924011023903e-05,
+      "loss": 0.0032,
+      "step": 4451
+    },
+    {
+      "epoch": 8.46,
+      "grad_norm": 0.07061257213354111,
+      "learning_rate": 3.6371027645139136e-05,
+      "loss": 0.0012,
+      "step": 4452
+    },
+    {
+      "epoch": 8.47,
+      "grad_norm": 0.03867549076676369,
+      "learning_rate": 3.628291513675069e-05,
+      "loss": 0.0008,
+      "step": 4453
+    },
+    {
+      "epoch": 8.47,
+      "grad_norm": 0.03233369439840317,
+      "learning_rate": 3.6194902618482255e-05,
+      "loss": 0.0006,
+      "step": 4454
+    },
+    {
+      "epoch": 8.47,
+      "grad_norm": 0.041974134743213654,
+      "learning_rate": 3.610699012370435e-05,
+      "loss": 0.0006,
+      "step": 4455
+    },
+    {
+      "epoch": 8.47,
+      "grad_norm": 0.0544402040541172,
+      "learning_rate": 3.60191776857496e-05,
+      "loss": 0.0011,
+      "step": 4456
+    },
+    {
+      "epoch": 8.47,
+      "grad_norm": 0.05878666788339615,
+      "learning_rate": 3.593146533791261e-05,
+      "loss": 0.0011,
+      "step": 4457
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 0.014738723635673523,
+      "learning_rate": 3.584385311345028e-05,
+      "loss": 0.0002,
+      "step": 4458
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 0.09206192195415497,
+      "learning_rate": 3.5756341045581236e-05,
+      "loss": 0.002,
+      "step": 4459
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 0.04009959474205971,
+      "learning_rate": 3.566892916748627e-05,
+      "loss": 0.0007,
+      "step": 4460
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 0.010608205571770668,
+      "learning_rate": 3.55816175123083e-05,
+      "loss": 0.0003,
+      "step": 4461
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 0.15474827587604523,
+      "learning_rate": 3.549440611315199e-05,
+      "loss": 0.0011,
+      "step": 4462
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 0.05248108506202698,
+      "learning_rate": 3.54072950030843e-05,
+      "loss": 0.001,
+      "step": 4463
+    },
+    {
+      "epoch": 8.49,
+      "grad_norm": 0.15078531205654144,
+      "learning_rate": 3.532028421513376e-05,
+      "loss": 0.0019,
+      "step": 4464
+    },
+    {
+      "epoch": 8.49,
+      "grad_norm": 0.07364580035209656,
+      "learning_rate": 3.523337378229132e-05,
+      "loss": 0.003,
+      "step": 4465
+    },
+    {
+      "epoch": 8.49,
+      "grad_norm": 0.1627316027879715,
+      "learning_rate": 3.5146563737509567e-05,
+      "loss": 0.0046,
+      "step": 4466
+    },
+    {
+      "epoch": 8.49,
+      "grad_norm": 0.020456992089748383,
+      "learning_rate": 3.5059854113703155e-05,
+      "loss": 0.0004,
+      "step": 4467
+    },
+    {
+      "epoch": 8.49,
+      "grad_norm": 0.1129571795463562,
+      "learning_rate": 3.4973244943748544e-05,
+      "loss": 0.0036,
+      "step": 4468
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 0.025371085852384567,
+      "learning_rate": 3.488673626048434e-05,
+      "loss": 0.0003,
+      "step": 4469
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 0.04040466994047165,
+      "learning_rate": 3.4800328096710694e-05,
+      "loss": 0.0007,
+      "step": 4470
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 0.10386474430561066,
+      "learning_rate": 3.471402048519004e-05,
+      "loss": 0.0021,
+      "step": 4471
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 0.13852186501026154,
+      "learning_rate": 3.462781345864637e-05,
+      "loss": 0.0008,
+      "step": 4472
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 0.011225958354771137,
+      "learning_rate": 3.4541707049765734e-05,
+      "loss": 0.0002,
+      "step": 4473
+    },
+    {
+      "epoch": 8.51,
+      "grad_norm": 0.05949229374527931,
+      "learning_rate": 3.445570129119588e-05,
+      "loss": 0.0012,
+      "step": 4474
+    },
+    {
+      "epoch": 8.51,
+      "grad_norm": 0.024381890892982483,
+      "learning_rate": 3.436979621554656e-05,
+      "loss": 0.0005,
+      "step": 4475
+    },
+    {
+      "epoch": 8.51,
+      "grad_norm": 0.04067368432879448,
+      "learning_rate": 3.4283991855389227e-05,
+      "loss": 0.0008,
+      "step": 4476
+    },
+    {
+      "epoch": 8.51,
+      "grad_norm": 0.0177047997713089,
+      "learning_rate": 3.419828824325711e-05,
+      "loss": 0.0002,
+      "step": 4477
+    },
+    {
+      "epoch": 8.51,
+      "grad_norm": 0.012734758667647839,
+      "learning_rate": 3.4112685411645526e-05,
+      "loss": 0.0002,
+      "step": 4478
+    },
+    {
+      "epoch": 8.52,
+      "grad_norm": 0.020619388669729233,
+      "learning_rate": 3.40271833930111e-05,
+      "loss": 0.0004,
+      "step": 4479
+    },
+    {
+      "epoch": 8.52,
+      "grad_norm": 0.055596668273210526,
+      "learning_rate": 3.394178221977272e-05,
+      "loss": 0.0011,
+      "step": 4480
+    },
+    {
+      "epoch": 8.52,
+      "grad_norm": 0.03161495551466942,
+      "learning_rate": 3.385648192431073e-05,
+      "loss": 0.0004,
+      "step": 4481
+    },
+    {
+      "epoch": 8.52,
+      "grad_norm": 0.24485482275485992,
+      "learning_rate": 3.3771282538967305e-05,
+      "loss": 0.0031,
+      "step": 4482
+    },
+    {
+      "epoch": 8.52,
+      "grad_norm": 0.06473598629236221,
+      "learning_rate": 3.36861840960464e-05,
+      "loss": 0.0014,
+      "step": 4483
+    },
+    {
+      "epoch": 8.52,
+      "grad_norm": 0.03849637508392334,
+      "learning_rate": 3.360118662781364e-05,
+      "loss": 0.0009,
+      "step": 4484
+    },
+    {
+      "epoch": 8.53,
+      "grad_norm": 0.025242341682314873,
+      "learning_rate": 3.3516290166496376e-05,
+      "loss": 0.0005,
+      "step": 4485
+    },
+    {
+      "epoch": 8.53,
+      "grad_norm": 0.13845162093639374,
+      "learning_rate": 3.343149474428366e-05,
+      "loss": 0.0057,
+      "step": 4486
+    },
+    {
+      "epoch": 8.53,
+      "grad_norm": 0.10756086558103561,
+      "learning_rate": 3.334680039332645e-05,
+      "loss": 0.0034,
+      "step": 4487
+    },
+    {
+      "epoch": 8.53,
+      "grad_norm": 0.1167188435792923,
+      "learning_rate": 3.326220714573688e-05,
+      "loss": 0.0012,
+      "step": 4488
+    },
+    {
+      "epoch": 8.53,
+      "grad_norm": 0.024480702355504036,
+      "learning_rate": 3.317771503358929e-05,
+      "loss": 0.0004,
+      "step": 4489
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 0.0609021857380867,
+      "learning_rate": 3.309332408891927e-05,
+      "loss": 0.0008,
+      "step": 4490
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 0.0712532103061676,
+      "learning_rate": 3.300903434372432e-05,
+      "loss": 0.0015,
+      "step": 4491
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 0.026839759200811386,
+      "learning_rate": 3.292484582996346e-05,
+      "loss": 0.0007,
+      "step": 4492
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 0.1404748111963272,
+      "learning_rate": 3.284075857955735e-05,
+      "loss": 0.0055,
+      "step": 4493
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 0.07714039832353592,
+      "learning_rate": 3.275677262438816e-05,
+      "loss": 0.0014,
+      "step": 4494
+    },
+    {
+      "epoch": 8.55,
+      "grad_norm": 0.12577293813228607,
+      "learning_rate": 3.26728879962998e-05,
+      "loss": 0.0015,
+      "step": 4495
+    },
+    {
+      "epoch": 8.55,
+      "grad_norm": 0.03133292868733406,
+      "learning_rate": 3.258910472709769e-05,
+      "loss": 0.0007,
+      "step": 4496
+    },
+    {
+      "epoch": 8.55,
+      "grad_norm": 0.036218348890542984,
+      "learning_rate": 3.250542284854876e-05,
+      "loss": 0.0009,
+      "step": 4497
+    },
+    {
+      "epoch": 8.55,
+      "grad_norm": 0.04037129133939743,
+      "learning_rate": 3.2421842392381714e-05,
+      "loss": 0.0009,
+      "step": 4498
+    },
+    {
+      "epoch": 8.55,
+      "grad_norm": 0.05914370343089104,
+      "learning_rate": 3.233836339028645e-05,
+      "loss": 0.0012,
+      "step": 4499
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 0.13102252781391144,
+      "learning_rate": 3.225498587391474e-05,
+      "loss": 0.0039,
+      "step": 4500
+    },
+    {
+      "epoch": 8.56,
+      "eval_blimp_filtered_avg": 0.5408955223880597,
+      "eval_blimp_filtered_std": 0.0058696884590671555,
+      "step": 4500
+    },
+    {
+      "epoch": 8.56,
+      "eval_blimp_supplement_avg": 0.5474137931034483,
+      "eval_blimp_supplement_std": 0.023171362129813227,
+      "step": 4500
+    },
+    {
+      "epoch": 8.56,
+      "eval_vqa_filtered_avg": 0.15,
+      "eval_vqa_filtered_std": 0.03588702812826371,
+      "step": 4500
+    },
+    {
+      "epoch": 8.56,
+      "eval_winoground_filtered_avg": 0.51,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 4500
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 0.10031779110431671,
+      "learning_rate": 3.2171709874879705e-05,
+      "loss": 0.0011,
+      "step": 4501
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 0.017441604286432266,
+      "learning_rate": 3.208853542475598e-05,
+      "loss": 0.0003,
+      "step": 4502
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 0.013597894459962845,
+      "learning_rate": 3.200546255507968e-05,
+      "loss": 0.0003,
+      "step": 4503
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 0.02666313946247101,
+      "learning_rate": 3.19224912973486e-05,
+      "loss": 0.0005,
+      "step": 4504
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 0.13654325902462006,
+      "learning_rate": 3.183962168302171e-05,
+      "loss": 0.0058,
+      "step": 4505
+    },
+    {
+      "epoch": 8.57,
+      "grad_norm": 0.06931483745574951,
+      "learning_rate": 3.1756853743519584e-05,
+      "loss": 0.0016,
+      "step": 4506
+    },
+    {
+      "epoch": 8.57,
+      "grad_norm": 0.06919556111097336,
+      "learning_rate": 3.167418751022438e-05,
+      "loss": 0.001,
+      "step": 4507
+    },
+    {
+      "epoch": 8.57,
+      "grad_norm": 0.10675889253616333,
+      "learning_rate": 3.1591623014479365e-05,
+      "loss": 0.0009,
+      "step": 4508
+    },
+    {
+      "epoch": 8.57,
+      "grad_norm": 0.011571124196052551,
+      "learning_rate": 3.1509160287589583e-05,
+      "loss": 0.0004,
+      "step": 4509
+    },
+    {
+      "epoch": 8.57,
+      "grad_norm": 0.006723396014422178,
+      "learning_rate": 3.142679936082131e-05,
+      "loss": 0.0002,
+      "step": 4510
+    },
+    {
+      "epoch": 8.58,
+      "grad_norm": 0.04153607040643692,
+      "learning_rate": 3.1344540265402216e-05,
+      "loss": 0.0007,
+      "step": 4511
+    },
+    {
+      "epoch": 8.58,
+      "grad_norm": 0.04115785285830498,
+      "learning_rate": 3.126238303252143e-05,
+      "loss": 0.0008,
+      "step": 4512
+    },
+    {
+      "epoch": 8.58,
+      "grad_norm": 0.04473372548818588,
+      "learning_rate": 3.1180327693329364e-05,
+      "loss": 0.0008,
+      "step": 4513
+    },
+    {
+      "epoch": 8.58,
+      "grad_norm": 0.010500053875148296,
+      "learning_rate": 3.1098374278937866e-05,
+      "loss": 0.0002,
+      "step": 4514
+    },
+    {
+      "epoch": 8.58,
+      "grad_norm": 0.030172478407621384,
+      "learning_rate": 3.101652282042024e-05,
+      "loss": 0.0008,
+      "step": 4515
+    },
+    {
+      "epoch": 8.59,
+      "grad_norm": 0.19653673470020294,
+      "learning_rate": 3.093477334881084e-05,
+      "loss": 0.0078,
+      "step": 4516
+    },
+    {
+      "epoch": 8.59,
+      "grad_norm": 0.08757874369621277,
+      "learning_rate": 3.085312589510566e-05,
+      "loss": 0.0045,
+      "step": 4517
+    },
+    {
+      "epoch": 8.59,
+      "grad_norm": 0.04365585371851921,
+      "learning_rate": 3.077158049026187e-05,
+      "loss": 0.0007,
+      "step": 4518
+    },
+    {
+      "epoch": 8.59,
+      "grad_norm": 0.008203418925404549,
+      "learning_rate": 3.069013716519795e-05,
+      "loss": 0.0002,
+      "step": 4519
+    },
+    {
+      "epoch": 8.59,
+      "grad_norm": 0.15351255238056183,
+      "learning_rate": 3.060879595079369e-05,
+      "loss": 0.0044,
+      "step": 4520
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 0.04348806291818619,
+      "learning_rate": 3.052755687789015e-05,
+      "loss": 0.0004,
+      "step": 4521
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 0.16879421472549438,
+      "learning_rate": 3.044641997728966e-05,
+      "loss": 0.0024,
+      "step": 4522
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 0.043063245713710785,
+      "learning_rate": 3.0365385279755817e-05,
+      "loss": 0.0011,
+      "step": 4523
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 0.0307457335293293,
+      "learning_rate": 3.0284452816013604e-05,
+      "loss": 0.0005,
+      "step": 4524
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 0.03550812974572182,
+      "learning_rate": 3.0203622616748914e-05,
+      "loss": 0.0008,
+      "step": 4525
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 0.04273543134331703,
+      "learning_rate": 3.0122894712609225e-05,
+      "loss": 0.0007,
+      "step": 4526
+    },
+    {
+      "epoch": 8.61,
+      "grad_norm": 0.033510174602270126,
+      "learning_rate": 3.004226913420299e-05,
+      "loss": 0.0005,
+      "step": 4527
+    },
+    {
+      "epoch": 8.61,
+      "grad_norm": 0.11794678866863251,
+      "learning_rate": 2.9961745912099987e-05,
+      "loss": 0.0015,
+      "step": 4528
+    },
+    {
+      "epoch": 8.61,
+      "grad_norm": 0.01926802657544613,
+      "learning_rate": 2.98813250768311e-05,
+      "loss": 0.0004,
+      "step": 4529
+    },
+    {
+      "epoch": 8.61,
+      "grad_norm": 0.07529684156179428,
+      "learning_rate": 2.980100665888846e-05,
+      "loss": 0.0013,
+      "step": 4530
+    },
+    {
+      "epoch": 8.61,
+      "grad_norm": 0.03784902021288872,
+      "learning_rate": 2.972079068872535e-05,
+      "loss": 0.0008,
+      "step": 4531
+    },
+    {
+      "epoch": 8.62,
+      "grad_norm": 0.02026723325252533,
+      "learning_rate": 2.96406771967561e-05,
+      "loss": 0.0005,
+      "step": 4532
+    },
+    {
+      "epoch": 8.62,
+      "grad_norm": 0.10392842441797256,
+      "learning_rate": 2.9560666213356486e-05,
+      "loss": 0.0032,
+      "step": 4533
+    },
+    {
+      "epoch": 8.62,
+      "grad_norm": 0.02050996758043766,
+      "learning_rate": 2.9480757768862973e-05,
+      "loss": 0.0004,
+      "step": 4534
+    },
+    {
+      "epoch": 8.62,
+      "grad_norm": 0.04556230083107948,
+      "learning_rate": 2.9400951893573587e-05,
+      "loss": 0.0011,
+      "step": 4535
+    },
+    {
+      "epoch": 8.62,
+      "grad_norm": 0.06512311100959778,
+      "learning_rate": 2.932124861774716e-05,
+      "loss": 0.0016,
+      "step": 4536
+    },
+    {
+      "epoch": 8.63,
+      "grad_norm": 0.036821089684963226,
+      "learning_rate": 2.9241647971603755e-05,
+      "loss": 0.0008,
+      "step": 4537
+    },
+    {
+      "epoch": 8.63,
+      "grad_norm": 0.0744747743010521,
+      "learning_rate": 2.9162149985324524e-05,
+      "loss": 0.0036,
+      "step": 4538
+    },
+    {
+      "epoch": 8.63,
+      "grad_norm": 0.051339294761419296,
+      "learning_rate": 2.9082754689051622e-05,
+      "loss": 0.0015,
+      "step": 4539
+    },
+    {
+      "epoch": 8.63,
+      "grad_norm": 0.11158092319965363,
+      "learning_rate": 2.9003462112888344e-05,
+      "loss": 0.0042,
+      "step": 4540
+    },
+    {
+      "epoch": 8.63,
+      "grad_norm": 0.03232022747397423,
+      "learning_rate": 2.8924272286898997e-05,
+      "loss": 0.0004,
+      "step": 4541
+    },
+    {
+      "epoch": 8.63,
+      "grad_norm": 0.01533407624810934,
+      "learning_rate": 2.8845185241108937e-05,
+      "loss": 0.0003,
+      "step": 4542
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 0.017064673826098442,
+      "learning_rate": 2.8766201005504506e-05,
+      "loss": 0.0003,
+      "step": 4543
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 0.06315390020608902,
+      "learning_rate": 2.868731961003329e-05,
+      "loss": 0.0008,
+      "step": 4544
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 0.02539408765733242,
+      "learning_rate": 2.860854108460342e-05,
+      "loss": 0.0005,
+      "step": 4545
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 0.10095426440238953,
+      "learning_rate": 2.8529865459084577e-05,
+      "loss": 0.0016,
+      "step": 4546
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 0.044226743280887604,
+      "learning_rate": 2.8451292763306998e-05,
+      "loss": 0.0006,
+      "step": 4547
+    },
+    {
+      "epoch": 8.65,
+      "grad_norm": 0.025293508544564247,
+      "learning_rate": 2.837282302706212e-05,
+      "loss": 0.0003,
+      "step": 4548
+    },
+    {
+      "epoch": 8.65,
+      "grad_norm": 0.0302597489207983,
+      "learning_rate": 2.8294456280102206e-05,
+      "loss": 0.0006,
+      "step": 4549
+    },
+    {
+      "epoch": 8.65,
+      "grad_norm": 0.14611980319023132,
+      "learning_rate": 2.8216192552140694e-05,
+      "loss": 0.0054,
+      "step": 4550
+    },
+    {
+      "epoch": 8.65,
+      "grad_norm": 0.02318945713341236,
+      "learning_rate": 2.8138031872851672e-05,
+      "loss": 0.0004,
+      "step": 4551
+    },
+    {
+      "epoch": 8.65,
+      "grad_norm": 0.06577591598033905,
+      "learning_rate": 2.8059974271870245e-05,
+      "loss": 0.0023,
+      "step": 4552
+    },
+    {
+      "epoch": 8.66,
+      "grad_norm": 0.03316377475857735,
+      "learning_rate": 2.7982019778792662e-05,
+      "loss": 0.0006,
+      "step": 4553
+    },
+    {
+      "epoch": 8.66,
+      "grad_norm": 0.015374165028333664,
+      "learning_rate": 2.7904168423175687e-05,
+      "loss": 0.0002,
+      "step": 4554
+    },
+    {
+      "epoch": 8.66,
+      "grad_norm": 0.024228712543845177,
+      "learning_rate": 2.7826420234537338e-05,
+      "loss": 0.0004,
+      "step": 4555
+    },
+    {
+      "epoch": 8.66,
+      "grad_norm": 0.052051618695259094,
+      "learning_rate": 2.774877524235637e-05,
+      "loss": 0.0013,
+      "step": 4556
+    },
+    {
+      "epoch": 8.66,
+      "grad_norm": 0.035197481513023376,
+      "learning_rate": 2.767123347607233e-05,
+      "loss": 0.0004,
+      "step": 4557
+    },
+    {
+      "epoch": 8.67,
+      "grad_norm": 0.08831775188446045,
+      "learning_rate": 2.759379496508567e-05,
+      "loss": 0.0024,
+      "step": 4558
+    },
+    {
+      "epoch": 8.67,
+      "grad_norm": 0.11312561482191086,
+      "learning_rate": 2.7516459738757933e-05,
+      "loss": 0.0009,
+      "step": 4559
+    },
+    {
+      "epoch": 8.67,
+      "grad_norm": 0.07478821277618408,
+      "learning_rate": 2.7439227826411036e-05,
+      "loss": 0.0022,
+      "step": 4560
+    },
+    {
+      "epoch": 8.67,
+      "grad_norm": 0.010302568785846233,
+      "learning_rate": 2.7362099257328217e-05,
+      "loss": 0.0002,
+      "step": 4561
+    },
+    {
+      "epoch": 8.67,
+      "grad_norm": 0.25420042872428894,
+      "learning_rate": 2.7285074060753055e-05,
+      "loss": 0.0144,
+      "step": 4562
+    },
+    {
+      "epoch": 8.67,
+      "grad_norm": 0.08430655300617218,
+      "learning_rate": 2.7208152265890347e-05,
+      "loss": 0.0018,
+      "step": 4563
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 0.024280227720737457,
+      "learning_rate": 2.7131333901905506e-05,
+      "loss": 0.0003,
+      "step": 4564
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 0.01762748323380947,
+      "learning_rate": 2.705461899792466e-05,
+      "loss": 0.0005,
+      "step": 4565
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 0.06539709866046906,
+      "learning_rate": 2.6978007583034845e-05,
+      "loss": 0.0017,
+      "step": 4566
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 0.006906154565513134,
+      "learning_rate": 2.6901499686283757e-05,
+      "loss": 0.0002,
+      "step": 4567
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 0.11644479632377625,
+      "learning_rate": 2.682509533667989e-05,
+      "loss": 0.0012,
+      "step": 4568
+    },
+    {
+      "epoch": 8.69,
+      "grad_norm": 0.11060871183872223,
+      "learning_rate": 2.6748794563192445e-05,
+      "loss": 0.0011,
+      "step": 4569
+    },
+    {
+      "epoch": 8.69,
+      "grad_norm": 0.10060733556747437,
+      "learning_rate": 2.6672597394751504e-05,
+      "loss": 0.0028,
+      "step": 4570
+    },
+    {
+      "epoch": 8.69,
+      "grad_norm": 0.10362262278795242,
+      "learning_rate": 2.6596503860247566e-05,
+      "loss": 0.0012,
+      "step": 4571
+    },
+    {
+      "epoch": 8.69,
+      "grad_norm": 0.12248428165912628,
+      "learning_rate": 2.6520513988532112e-05,
+      "loss": 0.001,
+      "step": 4572
+    },
+    {
+      "epoch": 8.69,
+      "grad_norm": 0.014800452627241611,
+      "learning_rate": 2.6444627808417228e-05,
+      "loss": 0.0003,
+      "step": 4573
+    },
+    {
+      "epoch": 8.7,
+      "grad_norm": 0.021233754232525826,
+      "learning_rate": 2.6368845348675638e-05,
+      "loss": 0.0003,
+      "step": 4574
+    },
+    {
+      "epoch": 8.7,
+      "grad_norm": 0.13268721103668213,
+      "learning_rate": 2.6293166638040786e-05,
+      "loss": 0.0019,
+      "step": 4575
+    },
+    {
+      "epoch": 8.7,
+      "grad_norm": 0.08089374750852585,
+      "learning_rate": 2.6217591705206753e-05,
+      "loss": 0.0022,
+      "step": 4576
+    },
+    {
+      "epoch": 8.7,
+      "grad_norm": 0.02264082431793213,
+      "learning_rate": 2.6142120578828285e-05,
+      "loss": 0.0004,
+      "step": 4577
+    },
+    {
+      "epoch": 8.7,
+      "grad_norm": 0.034675825387239456,
+      "learning_rate": 2.6066753287520727e-05,
+      "loss": 0.0012,
+      "step": 4578
+    },
+    {
+      "epoch": 8.71,
+      "grad_norm": 0.04020014777779579,
+      "learning_rate": 2.5991489859860282e-05,
+      "loss": 0.0006,
+      "step": 4579
+    },
+    {
+      "epoch": 8.71,
+      "grad_norm": 0.05711415782570839,
+      "learning_rate": 2.591633032438336e-05,
+      "loss": 0.0011,
+      "step": 4580
+    },
+    {
+      "epoch": 8.71,
+      "grad_norm": 0.056122370064258575,
+      "learning_rate": 2.5841274709587335e-05,
+      "loss": 0.0008,
+      "step": 4581
+    },
+    {
+      "epoch": 8.71,
+      "grad_norm": 0.025714989751577377,
+      "learning_rate": 2.5766323043930036e-05,
+      "loss": 0.0005,
+      "step": 4582
+    },
+    {
+      "epoch": 8.71,
+      "grad_norm": 0.06139739602804184,
+      "learning_rate": 2.5691475355829906e-05,
+      "loss": 0.002,
+      "step": 4583
+    },
+    {
+      "epoch": 8.71,
+      "grad_norm": 0.02577328309416771,
+      "learning_rate": 2.5616731673665902e-05,
+      "loss": 0.0005,
+      "step": 4584
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 0.0053535462357103825,
+      "learning_rate": 2.554209202577766e-05,
+      "loss": 0.0001,
+      "step": 4585
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 0.013915237039327621,
+      "learning_rate": 2.5467556440465243e-05,
+      "loss": 0.0002,
+      "step": 4586
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 0.029134774580597878,
+      "learning_rate": 2.5393124945989375e-05,
+      "loss": 0.0005,
+      "step": 4587
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 0.03808876499533653,
+      "learning_rate": 2.5318797570571247e-05,
+      "loss": 0.0006,
+      "step": 4588
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 0.07604074478149414,
+      "learning_rate": 2.524457434239252e-05,
+      "loss": 0.0005,
+      "step": 4589
+    },
+    {
+      "epoch": 8.73,
+      "grad_norm": 0.030406009405851364,
+      "learning_rate": 2.517045528959558e-05,
+      "loss": 0.0009,
+      "step": 4590
+    },
+    {
+      "epoch": 8.73,
+      "grad_norm": 0.08354295790195465,
+      "learning_rate": 2.5096440440283015e-05,
+      "loss": 0.0012,
+      "step": 4591
+    },
+    {
+      "epoch": 8.73,
+      "grad_norm": 0.17017613351345062,
+      "learning_rate": 2.502252982251818e-05,
+      "loss": 0.0056,
+      "step": 4592
+    },
+    {
+      "epoch": 8.73,
+      "grad_norm": 0.03694745898246765,
+      "learning_rate": 2.4948723464324715e-05,
+      "loss": 0.0011,
+      "step": 4593
+    },
+    {
+      "epoch": 8.73,
+      "grad_norm": 0.026754502207040787,
+      "learning_rate": 2.4875021393686845e-05,
+      "loss": 0.0004,
+      "step": 4594
+    },
+    {
+      "epoch": 8.74,
+      "grad_norm": 0.022085588425397873,
+      "learning_rate": 2.4801423638549112e-05,
+      "loss": 0.0005,
+      "step": 4595
+    },
+    {
+      "epoch": 8.74,
+      "grad_norm": 0.045793142169713974,
+      "learning_rate": 2.472793022681676e-05,
+      "loss": 0.0013,
+      "step": 4596
+    },
+    {
+      "epoch": 8.74,
+      "grad_norm": 0.08854474872350693,
+      "learning_rate": 2.4654541186355202e-05,
+      "loss": 0.0023,
+      "step": 4597
+    },
+    {
+      "epoch": 8.74,
+      "grad_norm": 0.09237746894359589,
+      "learning_rate": 2.458125654499038e-05,
+      "loss": 0.0042,
+      "step": 4598
+    },
+    {
+      "epoch": 8.74,
+      "grad_norm": 0.10978224873542786,
+      "learning_rate": 2.4508076330508798e-05,
+      "loss": 0.001,
+      "step": 4599
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 0.061948955059051514,
+      "learning_rate": 2.4435000570656992e-05,
+      "loss": 0.0015,
+      "step": 4600
+    },
+    {
+      "epoch": 8.75,
+      "eval_blimp_filtered_avg": 0.5425373134328358,
+      "eval_blimp_filtered_std": 0.005868749166288542,
+      "step": 4600
+    },
+    {
+      "epoch": 8.75,
+      "eval_blimp_supplement_avg": 0.5517241379310345,
+      "eval_blimp_supplement_std": 0.023157173227488538,
+      "step": 4600
+    },
+    {
+      "epoch": 8.75,
+      "eval_vqa_filtered_avg": 0.13,
+      "eval_vqa_filtered_std": 0.0337997668989631,
+      "step": 4600
+    },
+    {
+      "epoch": 8.75,
+      "eval_winoground_filtered_avg": 0.51,
+      "eval_winoground_filtered_std": 0.05024183937956912,
+      "step": 4600
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 0.04136449843645096,
+      "learning_rate": 2.436202929314237e-05,
+      "loss": 0.0012,
+      "step": 4601
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 0.028166690841317177,
+      "learning_rate": 2.4289162525632367e-05,
+      "loss": 0.0004,
+      "step": 4602
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 0.1685723215341568,
+      "learning_rate": 2.4216400295754944e-05,
+      "loss": 0.0051,
+      "step": 4603
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 0.031605374068021774,
+      "learning_rate": 2.4143742631098305e-05,
+      "loss": 0.0003,
+      "step": 4604
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 0.057000722736120224,
+      "learning_rate": 2.4071189559211312e-05,
+      "loss": 0.001,
+      "step": 4605
+    },
+    {
+      "epoch": 8.76,
+      "grad_norm": 0.019879955798387527,
+      "learning_rate": 2.399874110760269e-05,
+      "loss": 0.0004,
+      "step": 4606
+    },
+    {
+      "epoch": 8.76,
+      "grad_norm": 0.01609973795711994,
+      "learning_rate": 2.392639730374194e-05,
+      "loss": 0.0003,
+      "step": 4607
+    },
+    {
+      "epoch": 8.76,
+      "grad_norm": 0.01955130323767662,
+      "learning_rate": 2.385415817505868e-05,
+      "loss": 0.0004,
+      "step": 4608
+    },
+    {
+      "epoch": 8.76,
+      "grad_norm": 0.11622583121061325,
+      "learning_rate": 2.37820237489428e-05,
+      "loss": 0.0032,
+      "step": 4609
+    },
+    {
+      "epoch": 8.76,
+      "grad_norm": 0.007163602858781815,
+      "learning_rate": 2.3709994052744598e-05,
+      "loss": 0.0001,
+      "step": 4610
+    },
+    {
+      "epoch": 8.77,
+      "grad_norm": 0.05935308709740639,
+      "learning_rate": 2.363806911377465e-05,
+      "loss": 0.0007,
+      "step": 4611
+    },
+    {
+      "epoch": 8.77,
+      "grad_norm": 0.05218588560819626,
+      "learning_rate": 2.356624895930371e-05,
+      "loss": 0.001,
+      "step": 4612
+    },
+    {
+      "epoch": 8.77,
+      "grad_norm": 0.060240715742111206,
+      "learning_rate": 2.349453361656294e-05,
+      "loss": 0.0009,
+      "step": 4613
+    },
+    {
+      "epoch": 8.77,
+      "grad_norm": 0.0210797730833292,
+      "learning_rate": 2.3422923112743675e-05,
+      "loss": 0.0005,
+      "step": 4614
+    },
+    {
+      "epoch": 8.77,
+      "grad_norm": 0.038335222750902176,
+      "learning_rate": 2.335141747499747e-05,
+      "loss": 0.0006,
+      "step": 4615
+    },
+    {
+      "epoch": 8.78,
+      "grad_norm": 0.015231002122163773,
+      "learning_rate": 2.3280016730436336e-05,
+      "loss": 0.0003,
+      "step": 4616
+    },
+    {
+      "epoch": 8.78,
+      "grad_norm": 0.009975885041058064,
+      "learning_rate": 2.3208720906132118e-05,
+      "loss": 0.0002,
+      "step": 4617
+    },
+    {
+      "epoch": 8.78,
+      "grad_norm": 0.04658019542694092,
+      "learning_rate": 2.313753002911729e-05,
+      "loss": 0.0005,
+      "step": 4618
+    },
+    {
+      "epoch": 8.78,
+      "grad_norm": 0.03578156977891922,
+      "learning_rate": 2.30664441263843e-05,
+      "loss": 0.0007,
+      "step": 4619
+    },
+    {
+      "epoch": 8.78,
+      "grad_norm": 0.04919728636741638,
+      "learning_rate": 2.2995463224885847e-05,
+      "loss": 0.0013,
+      "step": 4620
+    },
+    {
+      "epoch": 8.79,
+      "grad_norm": 0.02024107240140438,
+      "learning_rate": 2.2924587351534806e-05,
+      "loss": 0.0003,
+      "step": 4621
+    },
+    {
+      "epoch": 8.79,
+      "grad_norm": 0.005165771581232548,
+      "learning_rate": 2.2853816533204284e-05,
+      "loss": 0.0001,
+      "step": 4622
+    },
+    {
+      "epoch": 8.79,
+      "grad_norm": 0.06311006098985672,
+      "learning_rate": 2.278315079672748e-05,
+      "loss": 0.0009,
+      "step": 4623
+    },
+    {
+      "epoch": 8.79,
+      "grad_norm": 0.06116212159395218,
+      "learning_rate": 2.271259016889776e-05,
+      "loss": 0.0018,
+      "step": 4624
+    },
+    {
+      "epoch": 8.79,
+      "grad_norm": 0.03700055181980133,
+      "learning_rate": 2.2642134676468793e-05,
+      "loss": 0.0006,
+      "step": 4625
+    },
+    {
+      "epoch": 8.79,
+      "grad_norm": 0.17214542627334595,
+      "learning_rate": 2.2571784346154043e-05,
+      "loss": 0.0044,
+      "step": 4626
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 0.12935200333595276,
+      "learning_rate": 2.250153920462754e-05,
+      "loss": 0.0021,
+      "step": 4627
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 0.03808349370956421,
+      "learning_rate": 2.2431399278523075e-05,
+      "loss": 0.0011,
+      "step": 4628
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 0.043291859328746796,
+      "learning_rate": 2.2361364594434682e-05,
+      "loss": 0.0005,
+      "step": 4629
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 0.015344783663749695,
+      "learning_rate": 2.229143517891655e-05,
+      "loss": 0.0003,
+      "step": 4630
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 0.05191527307033539,
+      "learning_rate": 2.2221611058482835e-05,
+      "loss": 0.0013,
+      "step": 4631
+    },
+    {
+      "epoch": 8.81,
+      "grad_norm": 0.02831227332353592,
+      "learning_rate": 2.2151892259607828e-05,
+      "loss": 0.0007,
+      "step": 4632
+    },
+    {
+      "epoch": 8.81,
+      "grad_norm": 0.05065396800637245,
+      "learning_rate": 2.2082278808725852e-05,
+      "loss": 0.001,
+      "step": 4633
+    },
+    {
+      "epoch": 8.81,
+      "grad_norm": 0.06894666701555252,
+      "learning_rate": 2.2012770732231487e-05,
+      "loss": 0.0011,
+      "step": 4634
+    },
+    {
+      "epoch": 8.81,
+      "grad_norm": 0.0317835733294487,
+      "learning_rate": 2.194336805647896e-05,
+      "loss": 0.0012,
+      "step": 4635
+    },
+    {
+      "epoch": 8.81,
+      "grad_norm": 0.09472806006669998,
+      "learning_rate": 2.187407080778295e-05,
+      "loss": 0.0024,
+      "step": 4636
+    },
+    {
+      "epoch": 8.82,
+      "grad_norm": 0.10331946611404419,
+      "learning_rate": 2.180487901241781e-05,
+      "loss": 0.0029,
+      "step": 4637
+    },
+    {
+      "epoch": 8.82,
+      "grad_norm": 0.16278430819511414,
+      "learning_rate": 2.1735792696618216e-05,
+      "loss": 0.0077,
+      "step": 4638
+    },
+    {
+      "epoch": 8.82,
+      "grad_norm": 0.027624640613794327,
+      "learning_rate": 2.1666811886578684e-05,
+      "loss": 0.0006,
+      "step": 4639
+    },
+    {
+      "epoch": 8.82,
+      "grad_norm": 0.05396478250622749,
+      "learning_rate": 2.1597936608453724e-05,
+      "loss": 0.0008,
+      "step": 4640
+    },
+    {
+      "epoch": 8.82,
+      "grad_norm": 0.02827777899801731,
+      "learning_rate": 2.1529166888357853e-05,
+      "loss": 0.0003,
+      "step": 4641
+    },
+    {
+      "epoch": 8.83,
+      "grad_norm": 0.0488787405192852,
+      "learning_rate": 2.1460502752365673e-05,
+      "loss": 0.0003,
+      "step": 4642
+    },
+    {
+      "epoch": 8.83,
+      "grad_norm": 0.036069147288799286,
+      "learning_rate": 2.1391944226511526e-05,
+      "loss": 0.0007,
+      "step": 4643
+    },
+    {
+      "epoch": 8.83,
+      "grad_norm": 0.040185559540987015,
+      "learning_rate": 2.1323491336789887e-05,
+      "loss": 0.0008,
+      "step": 4644
+    },
+    {
+      "epoch": 8.83,
+      "grad_norm": 0.030585620552301407,
+      "learning_rate": 2.125514410915523e-05,
+      "loss": 0.0006,
+      "step": 4645
+    },
+    {
+      "epoch": 8.83,
+      "grad_norm": 0.01663673296570778,
+      "learning_rate": 2.118690256952169e-05,
+      "loss": 0.0003,
+      "step": 4646
+    },
+    {
+      "epoch": 8.83,
+      "grad_norm": 0.12323743104934692,
+      "learning_rate": 2.111876674376367e-05,
+      "loss": 0.0016,
+      "step": 4647
+    },
+    {
+      "epoch": 8.84,
+      "grad_norm": 0.011169657111167908,
+      "learning_rate": 2.105073665771524e-05,
+      "loss": 0.0002,
+      "step": 4648
+    },
+    {
+      "epoch": 8.84,
+      "grad_norm": 0.011027532629668713,
+      "learning_rate": 2.0982812337170495e-05,
+      "loss": 0.0002,
+      "step": 4649
+    },
+    {
+      "epoch": 8.84,
+      "grad_norm": 0.06997191905975342,
+      "learning_rate": 2.0914993807883372e-05,
+      "loss": 0.0021,
+      "step": 4650
+    },
+    {
+      "epoch": 8.84,
+      "grad_norm": 0.015608487650752068,
+      "learning_rate": 2.0847281095567835e-05,
+      "loss": 0.0004,
+      "step": 4651
+    },
+    {
+      "epoch": 8.84,
+      "grad_norm": 0.045463934540748596,
+      "learning_rate": 2.0779674225897445e-05,
+      "loss": 0.0006,
+      "step": 4652
+    },
+    {
+      "epoch": 8.85,
+      "grad_norm": 0.22060415148735046,
+      "learning_rate": 2.0712173224505968e-05,
+      "loss": 0.0017,
+      "step": 4653
+    },
+    {
+      "epoch": 8.85,
+      "grad_norm": 0.054026030004024506,
+      "learning_rate": 2.0644778116986792e-05,
+      "loss": 0.001,
+      "step": 4654
+    },
+    {
+      "epoch": 8.85,
+      "grad_norm": 0.08282148092985153,
+      "learning_rate": 2.0577488928893214e-05,
+      "loss": 0.0025,
+      "step": 4655
+    },
+    {
+      "epoch": 8.85,
+      "grad_norm": 0.04451335594058037,
+      "learning_rate": 2.0510305685738392e-05,
+      "loss": 0.0007,
+      "step": 4656
+    },
+    {
+      "epoch": 8.85,
+      "grad_norm": 0.028186440467834473,
+      "learning_rate": 2.044322841299535e-05,
+      "loss": 0.0007,
+      "step": 4657
+    },
+    {
+      "epoch": 8.86,
+      "grad_norm": 0.0054627833887934685,
+      "learning_rate": 2.037625713609684e-05,
+      "loss": 0.0001,
+      "step": 4658
+    },
+    {
+      "epoch": 8.86,
+      "grad_norm": 0.09249133616685867,
+      "learning_rate": 2.0309391880435455e-05,
+      "loss": 0.0023,
+      "step": 4659
+    },
+    {
+      "epoch": 8.86,
+      "grad_norm": 0.06305146217346191,
+      "learning_rate": 2.0242632671363744e-05,
+      "loss": 0.004,
+      "step": 4660
+    },
+    {
+      "epoch": 8.86,
+      "grad_norm": 0.020345529541373253,
+      "learning_rate": 2.0175979534193763e-05,
+      "loss": 0.0003,
+      "step": 4661
+    },
+    {
+      "epoch": 8.86,
+      "grad_norm": 0.007300288416445255,
+      "learning_rate": 2.0109432494197597e-05,
+      "loss": 0.0002,
+      "step": 4662
+    },
+    {
+      "epoch": 8.87,
+      "grad_norm": 0.014178336597979069,
+      "learning_rate": 2.0042991576606894e-05,
+      "loss": 0.0003,
+      "step": 4663
+    },
+    {
+      "epoch": 8.87,
+      "grad_norm": 0.02150767669081688,
+      "learning_rate": 1.9976656806613332e-05,
+      "loss": 0.0004,
+      "step": 4664
+    },
+    {
+      "epoch": 8.87,
+      "grad_norm": 0.07122684270143509,
+      "learning_rate": 1.9910428209368067e-05,
+      "loss": 0.0029,
+      "step": 4665
+    },
+    {
+      "epoch": 8.87,
+      "grad_norm": 0.06315658241510391,
+      "learning_rate": 1.9844305809982174e-05,
+      "loss": 0.0008,
+      "step": 4666
+    },
+    {
+      "epoch": 8.87,
+      "grad_norm": 0.07094325125217438,
+      "learning_rate": 1.97782896335264e-05,
+      "loss": 0.0011,
+      "step": 4667
+    },
+    {
+      "epoch": 8.87,
+      "grad_norm": 0.015643475577235222,
+      "learning_rate": 1.971237970503122e-05,
+      "loss": 0.0003,
+      "step": 4668
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 0.06544893980026245,
+      "learning_rate": 1.964657604948685e-05,
+      "loss": 0.001,
+      "step": 4669
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 0.04757843539118767,
+      "learning_rate": 1.9580878691843125e-05,
+      "loss": 0.0005,
+      "step": 4670
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 0.09861589223146439,
+      "learning_rate": 1.951528765700976e-05,
+      "loss": 0.0066,
+      "step": 4671
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 0.030316265299916267,
+      "learning_rate": 1.944980296985592e-05,
+      "loss": 0.0007,
+      "step": 4672
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 0.021473750472068787,
+      "learning_rate": 1.938442465521065e-05,
+      "loss": 0.0005,
+      "step": 4673
+    },
+    {
+      "epoch": 8.89,
+      "grad_norm": 0.026415321975946426,
+      "learning_rate": 1.9319152737862552e-05,
+      "loss": 0.0003,
+      "step": 4674
+    },
+    {
+      "epoch": 8.89,
+      "grad_norm": 0.0325302891433239,
+      "learning_rate": 1.92539872425599e-05,
+      "loss": 0.0006,
+      "step": 4675
+    },
+    {
+      "epoch": 8.89,
+      "grad_norm": 0.07460860162973404,
+      "learning_rate": 1.918892819401069e-05,
+      "loss": 0.001,
+      "step": 4676
+    },
+    {
+      "epoch": 8.89,
+      "grad_norm": 0.02745869755744934,
+      "learning_rate": 1.9123975616882427e-05,
+      "loss": 0.0006,
+      "step": 4677
+    },
+    {
+      "epoch": 8.89,
+      "grad_norm": 0.037749528884887695,
+      "learning_rate": 1.9059129535802374e-05,
+      "loss": 0.0006,
+      "step": 4678
+    },
+    {
+      "epoch": 8.9,
+      "grad_norm": 0.15057669579982758,
+      "learning_rate": 1.8994389975357272e-05,
+      "loss": 0.0019,
+      "step": 4679
+    },
+    {
+      "epoch": 8.9,
+      "grad_norm": 0.02101430669426918,
+      "learning_rate": 1.8929756960093746e-05,
+      "loss": 0.0004,
+      "step": 4680
+    },
+    {
+      "epoch": 8.9,
+      "grad_norm": 0.04377562180161476,
+      "learning_rate": 1.886523051451766e-05,
+      "loss": 0.0009,
+      "step": 4681
+    },
+    {
+      "epoch": 8.9,
+      "grad_norm": 0.097110316157341,
+      "learning_rate": 1.8800810663094744e-05,
+      "loss": 0.0034,
+      "step": 4682
+    },
+    {
+      "epoch": 8.9,
+      "grad_norm": 0.0217836182564497,
+      "learning_rate": 1.8736497430250197e-05,
+      "loss": 0.0002,
+      "step": 4683
+    },
+    {
+      "epoch": 8.9,
+      "grad_norm": 0.07222891598939896,
+      "learning_rate": 1.867229084036884e-05,
+      "loss": 0.0024,
+      "step": 4684
+    },
+    {
+      "epoch": 8.91,
+      "grad_norm": 0.01122148334980011,
+      "learning_rate": 1.8608190917795006e-05,
+      "loss": 0.0003,
+      "step": 4685
+    },
+    {
+      "epoch": 8.91,
+      "grad_norm": 0.02092917263507843,
+      "learning_rate": 1.854419768683265e-05,
+      "loss": 0.0003,
+      "step": 4686
+    },
+    {
+      "epoch": 8.91,
+      "grad_norm": 0.015466089360415936,
+      "learning_rate": 1.84803111717452e-05,
+      "loss": 0.0003,
+      "step": 4687
+    },
+    {
+      "epoch": 8.91,
+      "grad_norm": 0.010018778964877129,
+      "learning_rate": 1.8416531396755674e-05,
+      "loss": 0.0002,
+      "step": 4688
+    },
+    {
+      "epoch": 8.91,
+      "grad_norm": 0.02366429567337036,
+      "learning_rate": 1.83528583860466e-05,
+      "loss": 0.0002,
+      "step": 4689
+    },
+    {
+      "epoch": 8.92,
+      "grad_norm": 0.03789089620113373,
+      "learning_rate": 1.8289292163760027e-05,
+      "loss": 0.0004,
+      "step": 4690
+    },
+    {
+      "epoch": 8.92,
+      "grad_norm": 0.011624742299318314,
+      "learning_rate": 1.822583275399758e-05,
+      "loss": 0.0003,
+      "step": 4691
+    },
+    {
+      "epoch": 8.92,
+      "grad_norm": 0.008515357039868832,
+      "learning_rate": 1.8162480180820216e-05,
+      "loss": 0.0002,
+      "step": 4692
+    },
+    {
+      "epoch": 8.92,
+      "grad_norm": 0.022082965821027756,
+      "learning_rate": 1.8099234468248613e-05,
+      "loss": 0.0006,
+      "step": 4693
+    },
+    {
+      "epoch": 8.92,
+      "grad_norm": 0.23262417316436768,
+      "learning_rate": 1.8036095640262726e-05,
+      "loss": 0.0053,
+      "step": 4694
+    },
+    {
+      "epoch": 8.93,
+      "grad_norm": 0.05679648369550705,
+      "learning_rate": 1.7973063720802105e-05,
+      "loss": 0.0015,
+      "step": 4695
+    },
+    {
+      "epoch": 8.93,
+      "grad_norm": 0.030516983941197395,
+      "learning_rate": 1.7910138733765666e-05,
+      "loss": 0.0004,
+      "step": 4696
+    },
+    {
+      "epoch": 8.93,
+      "grad_norm": 0.009400640614330769,
+      "learning_rate": 1.7847320703012026e-05,
+      "loss": 0.0002,
+      "step": 4697
+    },
+    {
+      "epoch": 8.93,
+      "grad_norm": 0.042748529464006424,
+      "learning_rate": 1.7784609652358827e-05,
+      "loss": 0.0008,
+      "step": 4698
+    },
+    {
+      "epoch": 8.93,
+      "grad_norm": 0.013971800915896893,
+      "learning_rate": 1.7722005605583523e-05,
+      "loss": 0.0003,
+      "step": 4699
+    },
+    {
+      "epoch": 8.94,
+      "grad_norm": 0.017422590404748917,
+      "learning_rate": 1.7659508586422855e-05,
+      "loss": 0.0003,
+      "step": 4700
+    },
+    {
+      "epoch": 8.94,
+      "eval_blimp_filtered_avg": 0.5407462686567164,
+      "eval_blimp_filtered_std": 0.005863110249573203,
+      "step": 4700
+    },
+    {
+      "epoch": 8.94,
+      "eval_blimp_supplement_avg": 0.5495689655172413,
+      "eval_blimp_supplement_std": 0.023148047197586195,
+      "step": 4700
+    },
+    {
+      "epoch": 8.94,
+      "eval_vqa_filtered_avg": 0.12,
+      "eval_vqa_filtered_std": 0.03265986323710906,
+      "step": 4700
+    },
+    {
+      "epoch": 8.94,
+      "eval_winoground_filtered_avg": 0.54,
+      "eval_winoground_filtered_std": 0.05009082659620333,
+      "step": 4700
+    },
+    {
+      "epoch": 8.94,
+      "grad_norm": 0.06172678992152214,
+      "learning_rate": 1.7597118618573003e-05,
+      "loss": 0.0011,
+      "step": 4701
+    },
+    {
+      "epoch": 8.94,
+      "grad_norm": 0.0625278577208519,
+      "learning_rate": 1.753483572568951e-05,
+      "loss": 0.0016,
+      "step": 4702
+    },
+    {
+      "epoch": 8.94,
+      "grad_norm": 0.03934096172451973,
+      "learning_rate": 1.7472659931387357e-05,
+      "loss": 0.0019,
+      "step": 4703
+    },
+    {
+      "epoch": 8.94,
+      "grad_norm": 0.005482279229909182,
+      "learning_rate": 1.7410591259240913e-05,
+      "loss": 0.0001,
+      "step": 4704
+    },
+    {
+      "epoch": 8.94,
+      "grad_norm": 0.01977839693427086,
+      "learning_rate": 1.7348629732783925e-05,
+      "loss": 0.0003,
+      "step": 4705
+    },
+    {
+      "epoch": 8.95,
+      "grad_norm": 0.007910947315394878,
+      "learning_rate": 1.7286775375509665e-05,
+      "loss": 0.0002,
+      "step": 4706
+    },
+    {
+      "epoch": 8.95,
+      "grad_norm": 0.018948592245578766,
+      "learning_rate": 1.7225028210870406e-05,
+      "loss": 0.0002,
+      "step": 4707
+    },
+    {
+      "epoch": 8.95,
+      "grad_norm": 0.02717820554971695,
+      "learning_rate": 1.7163388262278154e-05,
+      "loss": 0.0005,
+      "step": 4708
+    },
+    {
+      "epoch": 8.95,
+      "grad_norm": 0.011470044031739235,
+      "learning_rate": 1.710185555310408e-05,
+      "loss": 0.0002,
+      "step": 4709
+    },
+    {
+      "epoch": 8.95,
+      "grad_norm": 0.06851696223020554,
+      "learning_rate": 1.7040430106678727e-05,
+      "loss": 0.0011,
+      "step": 4710
+    },
+    {
+      "epoch": 8.96,
+      "grad_norm": 0.07709036022424698,
+      "learning_rate": 1.6979111946291956e-05,
+      "loss": 0.0016,
+      "step": 4711
+    },
+    {
+      "epoch": 8.96,
+      "grad_norm": 0.013667052611708641,
+      "learning_rate": 1.691790109519301e-05,
+      "loss": 0.0003,
+      "step": 4712
+    },
+    {
+      "epoch": 8.96,
+      "grad_norm": 0.05704985931515694,
+      "learning_rate": 1.685679757659032e-05,
+      "loss": 0.0015,
+      "step": 4713
+    },
+    {
+      "epoch": 8.96,
+      "grad_norm": 0.06836773455142975,
+      "learning_rate": 1.6795801413651754e-05,
+      "loss": 0.0018,
+      "step": 4714
+    },
+    {
+      "epoch": 8.96,
+      "grad_norm": 0.05811435356736183,
+      "learning_rate": 1.6734912629504415e-05,
+      "loss": 0.0009,
+      "step": 4715
+    },
+    {
+      "epoch": 8.97,
+      "grad_norm": 0.011817816644906998,
+      "learning_rate": 1.6674131247234635e-05,
+      "loss": 0.0002,
+      "step": 4716
+    },
+    {
+      "epoch": 8.97,
+      "grad_norm": 0.05658109113574028,
+      "learning_rate": 1.6613457289888243e-05,
+      "loss": 0.0015,
+      "step": 4717
+    },
+    {
+      "epoch": 8.97,
+      "grad_norm": 0.010744825005531311,
+      "learning_rate": 1.6552890780469972e-05,
+      "loss": 0.0002,
+      "step": 4718
+    },
+    {
+      "epoch": 8.97,
+      "grad_norm": 0.05544554814696312,
+      "learning_rate": 1.6492431741944188e-05,
+      "loss": 0.0017,
+      "step": 4719
+    },
+    {
+      "epoch": 8.97,
+      "grad_norm": 0.009559288620948792,
+      "learning_rate": 1.6432080197234288e-05,
+      "loss": 0.0001,
+      "step": 4720
+    },
+    {
+      "epoch": 8.98,
+      "grad_norm": 0.047412097454071045,
+      "learning_rate": 1.6371836169222997e-05,
+      "loss": 0.0008,
+      "step": 4721
+    },
+    {
+      "epoch": 8.98,
+      "grad_norm": 0.034955523908138275,
+      "learning_rate": 1.6311699680752176e-05,
+      "loss": 0.0005,
+      "step": 4722
+    },
+    {
+      "epoch": 8.98,
+      "grad_norm": 0.08693701773881912,
+      "learning_rate": 1.625167075462306e-05,
+      "loss": 0.0014,
+      "step": 4723
+    },
+    {
+      "epoch": 8.98,
+      "grad_norm": 0.00765823433175683,
+      "learning_rate": 1.6191749413595966e-05,
+      "loss": 0.0001,
+      "step": 4724
+    },
+    {
+      "epoch": 8.98,
+      "grad_norm": 0.07216187566518784,
+      "learning_rate": 1.6131935680390497e-05,
+      "loss": 0.0021,
+      "step": 4725
+    },
+    {
+      "epoch": 8.98,
+      "grad_norm": 0.03461804613471031,
+      "learning_rate": 1.607222957768547e-05,
+      "loss": 0.0007,
+      "step": 4726
+    },
+    {
+      "epoch": 8.99,
+      "grad_norm": 0.011866027489304543,
+      "learning_rate": 1.601263112811878e-05,
+      "loss": 0.0002,
+      "step": 4727
+    },
+    {
+      "epoch": 8.99,
+      "grad_norm": 0.011933130212128162,
+      "learning_rate": 1.5953140354287652e-05,
+      "loss": 0.0002,
+      "step": 4728
+    },
+    {
+      "epoch": 8.99,
+      "grad_norm": 0.04855145886540413,
+      "learning_rate": 1.5893757278748408e-05,
+      "loss": 0.0005,
+      "step": 4729
+    },
+    {
+      "epoch": 8.99,
+      "grad_norm": 0.10093075782060623,
+      "learning_rate": 1.583448192401654e-05,
+      "loss": 0.0017,
+      "step": 4730
+    },
+    {
+      "epoch": 8.99,
+      "grad_norm": 0.03847431018948555,
+      "learning_rate": 1.5775314312566633e-05,
+      "loss": 0.001,
+      "step": 4731
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 0.0071837035939097404,
+      "learning_rate": 1.5716254466832612e-05,
+      "loss": 0.0002,
+      "step": 4732
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 0.01854768767952919,
+      "learning_rate": 1.5657302409207328e-05,
+      "loss": 0.0005,
+      "step": 4733
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 0.02752833068370819,
+      "learning_rate": 1.559845816204287e-05,
+      "loss": 0.0004,
+      "step": 4734
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 0.039274703711271286,
+      "learning_rate": 1.5539721747650457e-05,
+      "loss": 0.0011,
+      "step": 4735
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 0.005300817545503378,
+      "learning_rate": 1.548109318830034e-05,
+      "loss": 0.0002,
+      "step": 4736
+    },
+    {
+      "epoch": 9.01,
+      "grad_norm": 0.01491699181497097,
+      "learning_rate": 1.542257250622204e-05,
+      "loss": 0.0002,
+      "step": 4737
+    },
+    {
+      "epoch": 9.01,
+      "grad_norm": 0.06962891668081284,
+      "learning_rate": 1.5364159723603942e-05,
+      "loss": 0.001,
+      "step": 4738
+    },
+    {
+      "epoch": 9.01,
+      "grad_norm": 0.014101231470704079,
+      "learning_rate": 1.5305854862593725e-05,
+      "loss": 0.0003,
+      "step": 4739
+    },
+    {
+      "epoch": 9.01,
+      "grad_norm": 0.03290229290723801,
+      "learning_rate": 1.5247657945298075e-05,
+      "loss": 0.0005,
+      "step": 4740
+    },
+    {
+      "epoch": 9.01,
+      "grad_norm": 0.007186095230281353,
+      "learning_rate": 1.5189568993782742e-05,
+      "loss": 0.0002,
+      "step": 4741
+    },
+    {
+      "epoch": 9.02,
+      "grad_norm": 0.006306251976639032,
+      "learning_rate": 1.5131588030072473e-05,
+      "loss": 0.0002,
+      "step": 4742
+    },
+    {
+      "epoch": 9.02,
+      "grad_norm": 0.03247540444135666,
+      "learning_rate": 1.5073715076151315e-05,
+      "loss": 0.0012,
+      "step": 4743
+    },
+    {
+      "epoch": 9.02,
+      "grad_norm": 0.05208312347531319,
+      "learning_rate": 1.5015950153961987e-05,
+      "loss": 0.001,
+      "step": 4744
+    },
+    {
+      "epoch": 9.02,
+      "grad_norm": 0.01133365835994482,
+      "learning_rate": 1.4958293285406564e-05,
+      "loss": 0.0003,
+      "step": 4745
+    },
+    {
+      "epoch": 9.02,
+      "grad_norm": 0.04553275927901268,
+      "learning_rate": 1.4900744492346029e-05,
+      "loss": 0.0003,
+      "step": 4746
+    },
+    {
+      "epoch": 9.02,
+      "grad_norm": 0.00782103929668665,
+      "learning_rate": 1.4843303796600425e-05,
+      "loss": 0.0002,
+      "step": 4747
+    },
+    {
+      "epoch": 9.03,
+      "grad_norm": 0.021717559546232224,
+      "learning_rate": 1.4785971219948701e-05,
+      "loss": 0.0003,
+      "step": 4748
+    },
+    {
+      "epoch": 9.03,
+      "grad_norm": 0.04631790891289711,
+      "learning_rate": 1.4728746784128965e-05,
+      "loss": 0.0014,
+      "step": 4749
+    },
+    {
+      "epoch": 9.03,
+      "grad_norm": 0.0211868304759264,
+      "learning_rate": 1.4671630510838195e-05,
+      "loss": 0.0006,
+      "step": 4750
+    },
+    {
+      "epoch": 9.03,
+      "grad_norm": 0.03237876296043396,
+      "learning_rate": 1.4614622421732403e-05,
+      "loss": 0.0003,
+      "step": 4751
+    },
+    {
+      "epoch": 9.03,
+      "grad_norm": 0.005018625874072313,
+      "learning_rate": 1.4557722538426698e-05,
+      "loss": 0.0002,
+      "step": 4752
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 0.007093494758009911,
+      "learning_rate": 1.4500930882494921e-05,
+      "loss": 0.0002,
+      "step": 4753
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 0.009090973995625973,
+      "learning_rate": 1.4444247475470117e-05,
+      "loss": 0.0002,
+      "step": 4754
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 0.005744739435613155,
+      "learning_rate": 1.4387672338844158e-05,
+      "loss": 0.0002,
+      "step": 4755
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 0.04501582682132721,
+      "learning_rate": 1.4331205494067855e-05,
+      "loss": 0.0013,
+      "step": 4756
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 0.032413747161626816,
+      "learning_rate": 1.4274846962551013e-05,
+      "loss": 0.0005,
+      "step": 4757
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 0.04222416132688522,
+      "learning_rate": 1.4218596765662372e-05,
+      "loss": 0.0015,
+      "step": 4758
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 0.02050110325217247,
+      "learning_rate": 1.4162454924729571e-05,
+      "loss": 0.0006,
+      "step": 4759
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 0.01810363121330738,
+      "learning_rate": 1.4106421461039118e-05,
+      "loss": 0.0004,
+      "step": 4760
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 0.01858523115515709,
+      "learning_rate": 1.4050496395836553e-05,
+      "loss": 0.0003,
+      "step": 4761
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 0.0450543537735939,
+      "learning_rate": 1.3994679750326143e-05,
+      "loss": 0.001,
+      "step": 4762
+    },
+    {
+      "epoch": 9.06,
+      "grad_norm": 0.011183064430952072,
+      "learning_rate": 1.3938971545671295e-05,
+      "loss": 0.0003,
+      "step": 4763
+    },
+    {
+      "epoch": 9.06,
+      "grad_norm": 0.01352730207145214,
+      "learning_rate": 1.3883371802994014e-05,
+      "loss": 0.0003,
+      "step": 4764
+    },
+    {
+      "epoch": 9.06,
+      "grad_norm": 0.013757997192442417,
+      "learning_rate": 1.3827880543375402e-05,
+      "loss": 0.0003,
+      "step": 4765
+    },
+    {
+      "epoch": 9.06,
+      "grad_norm": 0.039414580911397934,
+      "learning_rate": 1.377249778785533e-05,
+      "loss": 0.001,
+      "step": 4766
+    },
+    {
+      "epoch": 9.06,
+      "grad_norm": 0.025043528527021408,
+      "learning_rate": 1.3717223557432533e-05,
+      "loss": 0.0006,
+      "step": 4767
+    },
+    {
+      "epoch": 9.06,
+      "grad_norm": 0.009793458506464958,
+      "learning_rate": 1.3662057873064614e-05,
+      "loss": 0.0002,
+      "step": 4768
+    },
+    {
+      "epoch": 9.07,
+      "grad_norm": 0.0521082803606987,
+      "learning_rate": 1.3607000755668006e-05,
+      "loss": 0.0004,
+      "step": 4769
+    },
+    {
+      "epoch": 9.07,
+      "grad_norm": 0.05542886257171631,
+      "learning_rate": 1.3552052226118016e-05,
+      "loss": 0.0007,
+      "step": 4770
+    },
+    {
+      "epoch": 9.07,
+      "grad_norm": 0.01751817762851715,
+      "learning_rate": 1.3497212305248673e-05,
+      "loss": 0.0003,
+      "step": 4771
+    },
+    {
+      "epoch": 9.07,
+      "grad_norm": 0.021827522665262222,
+      "learning_rate": 1.3442481013853046e-05,
+      "loss": 0.0005,
+      "step": 4772
+    },
+    {
+      "epoch": 9.07,
+      "grad_norm": 0.0338696725666523,
+      "learning_rate": 1.3387858372682736e-05,
+      "loss": 0.0006,
+      "step": 4773
+    },
+    {
+      "epoch": 9.08,
+      "grad_norm": 0.0755721926689148,
+      "learning_rate": 1.3333344402448376e-05,
+      "loss": 0.0018,
+      "step": 4774
+    },
+    {
+      "epoch": 9.08,
+      "grad_norm": 0.05858979374170303,
+      "learning_rate": 1.3278939123819265e-05,
+      "loss": 0.0005,
+      "step": 4775
+    },
+    {
+      "epoch": 9.08,
+      "grad_norm": 0.08012661337852478,
+      "learning_rate": 1.3224642557423536e-05,
+      "loss": 0.0009,
+      "step": 4776
+    },
+    {
+      "epoch": 9.08,
+      "grad_norm": 0.03330380097031593,
+      "learning_rate": 1.317045472384809e-05,
+      "loss": 0.0009,
+      "step": 4777
+    },
+    {
+      "epoch": 9.08,
+      "grad_norm": 0.031149402260780334,
+      "learning_rate": 1.3116375643638688e-05,
+      "loss": 0.0003,
+      "step": 4778
+    },
+    {
+      "epoch": 9.09,
+      "grad_norm": 0.014514913782477379,
+      "learning_rate": 1.3062405337299698e-05,
+      "loss": 0.0002,
+      "step": 4779
+    },
+    {
+      "epoch": 9.09,
+      "grad_norm": 0.013790026307106018,
+      "learning_rate": 1.3008543825294316e-05,
+      "loss": 0.0003,
+      "step": 4780
+    },
+    {
+      "epoch": 9.09,
+      "grad_norm": 0.009036700241267681,
+      "learning_rate": 1.2954791128044606e-05,
+      "loss": 0.0002,
+      "step": 4781
+    },
+    {
+      "epoch": 9.09,
+      "grad_norm": 0.06039751321077347,
+      "learning_rate": 1.2901147265931099e-05,
+      "loss": 0.0012,
+      "step": 4782
+    },
+    {
+      "epoch": 9.09,
+      "grad_norm": 0.01323123648762703,
+      "learning_rate": 1.2847612259293428e-05,
+      "loss": 0.0003,
+      "step": 4783
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 0.013130691833794117,
+      "learning_rate": 1.2794186128429584e-05,
+      "loss": 0.0002,
+      "step": 4784
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 0.010957039892673492,
+      "learning_rate": 1.2740868893596535e-05,
+      "loss": 0.0002,
+      "step": 4785
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 0.020512251183390617,
+      "learning_rate": 1.2687660575009839e-05,
+      "loss": 0.0003,
+      "step": 4786
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 0.0051098656840622425,
+      "learning_rate": 1.2634561192843828e-05,
+      "loss": 0.0001,
+      "step": 4787
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 0.02049040049314499,
+      "learning_rate": 1.2581570767231396e-05,
+      "loss": 0.0003,
+      "step": 4788
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 0.014327622950077057,
+      "learning_rate": 1.2528689318264406e-05,
+      "loss": 0.0003,
+      "step": 4789
+    },
+    {
+      "epoch": 9.11,
+      "grad_norm": 0.01204295177012682,
+      "learning_rate": 1.2475916865993018e-05,
+      "loss": 0.0002,
+      "step": 4790
+    },
+    {
+      "epoch": 9.11,
+      "grad_norm": 0.028350476175546646,
+      "learning_rate": 1.2423253430426394e-05,
+      "loss": 0.0006,
+      "step": 4791
+    },
+    {
+      "epoch": 9.11,
+      "grad_norm": 0.012895573861896992,
+      "learning_rate": 1.237069903153216e-05,
+      "loss": 0.0002,
+      "step": 4792
+    },
+    {
+      "epoch": 9.11,
+      "grad_norm": 0.03143145516514778,
+      "learning_rate": 1.2318253689236745e-05,
+      "loss": 0.0012,
+      "step": 4793
+    },
+    {
+      "epoch": 9.11,
+      "grad_norm": 0.019256848841905594,
+      "learning_rate": 1.2265917423425076e-05,
+      "loss": 0.0004,
+      "step": 4794
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 0.005416986532509327,
+      "learning_rate": 1.2213690253940844e-05,
+      "loss": 0.0002,
+      "step": 4795
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 0.03305651620030403,
+      "learning_rate": 1.2161572200586346e-05,
+      "loss": 0.0005,
+      "step": 4796
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 0.00651888269931078,
+      "learning_rate": 1.2109563283122437e-05,
+      "loss": 0.0001,
+      "step": 4797
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 0.04809172824025154,
+      "learning_rate": 1.2057663521268779e-05,
+      "loss": 0.0007,
+      "step": 4798
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 0.011113783344626427,
+      "learning_rate": 1.2005872934703364e-05,
+      "loss": 0.0002,
+      "step": 4799
+    },
+    {
+      "epoch": 9.13,
+      "grad_norm": 0.009929557330906391,
+      "learning_rate": 1.195419154306302e-05,
+      "loss": 0.0003,
+      "step": 4800
+    },
+    {
+      "epoch": 9.13,
+      "eval_blimp_filtered_avg": 0.5395522388059701,
+      "eval_blimp_filtered_std": 0.005867061895595643,
+      "step": 4800
+    },
+    {
+      "epoch": 9.13,
+      "eval_blimp_supplement_avg": 0.5474137931034483,
+      "eval_blimp_supplement_std": 0.023134858798920886,
+      "step": 4800
+    },
+    {
+      "epoch": 9.13,
+      "eval_vqa_filtered_avg": 0.11,
+      "eval_vqa_filtered_std": 0.031446603773522035,
+      "step": 4800
+    },
+    {
+      "epoch": 9.13,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 4800
+    },
+    {
+      "epoch": 9.13,
+      "grad_norm": 0.011969330720603466,
+      "learning_rate": 1.1902619365943101e-05,
+      "loss": 0.0002,
+      "step": 4801
+    },
+    {
+      "epoch": 9.13,
+      "grad_norm": 0.01297009363770485,
+      "learning_rate": 1.1851156422897568e-05,
+      "loss": 0.0004,
+      "step": 4802
+    },
+    {
+      "epoch": 9.13,
+      "grad_norm": 0.00929530244320631,
+      "learning_rate": 1.1799802733438879e-05,
+      "loss": 0.0002,
+      "step": 4803
+    },
+    {
+      "epoch": 9.13,
+      "grad_norm": 0.03857535868883133,
+      "learning_rate": 1.1748558317038159e-05,
+      "loss": 0.0009,
+      "step": 4804
+    },
+    {
+      "epoch": 9.13,
+      "grad_norm": 0.005206716246902943,
+      "learning_rate": 1.16974231931251e-05,
+      "loss": 0.0001,
+      "step": 4805
+    },
+    {
+      "epoch": 9.14,
+      "grad_norm": 0.0031854859553277493,
+      "learning_rate": 1.1646397381087891e-05,
+      "loss": 0.0001,
+      "step": 4806
+    },
+    {
+      "epoch": 9.14,
+      "grad_norm": 0.021325618028640747,
+      "learning_rate": 1.159548090027339e-05,
+      "loss": 0.0003,
+      "step": 4807
+    },
+    {
+      "epoch": 9.14,
+      "grad_norm": 0.027285540476441383,
+      "learning_rate": 1.1544673769986823e-05,
+      "loss": 0.0005,
+      "step": 4808
+    },
+    {
+      "epoch": 9.14,
+      "grad_norm": 0.02192055806517601,
+      "learning_rate": 1.1493976009492178e-05,
+      "loss": 0.0004,
+      "step": 4809
+    },
+    {
+      "epoch": 9.14,
+      "grad_norm": 0.0089506134390831,
+      "learning_rate": 1.1443387638011681e-05,
+      "loss": 0.0002,
+      "step": 4810
+    },
+    {
+      "epoch": 9.15,
+      "grad_norm": 0.038725849241018295,
+      "learning_rate": 1.139290867472642e-05,
+      "loss": 0.0003,
+      "step": 4811
+    },
+    {
+      "epoch": 9.15,
+      "grad_norm": 0.007556665223091841,
+      "learning_rate": 1.1342539138775786e-05,
+      "loss": 0.0001,
+      "step": 4812
+    },
+    {
+      "epoch": 9.15,
+      "grad_norm": 0.044155560433864594,
+      "learning_rate": 1.1292279049257701e-05,
+      "loss": 0.0003,
+      "step": 4813
+    },
+    {
+      "epoch": 9.15,
+      "grad_norm": 0.06190323084592819,
+      "learning_rate": 1.1242128425228658e-05,
+      "loss": 0.0009,
+      "step": 4814
+    },
+    {
+      "epoch": 9.15,
+      "grad_norm": 0.019286949187517166,
+      "learning_rate": 1.1192087285703576e-05,
+      "loss": 0.0003,
+      "step": 4815
+    },
+    {
+      "epoch": 9.16,
+      "grad_norm": 0.010933559387922287,
+      "learning_rate": 1.1142155649655882e-05,
+      "loss": 0.0003,
+      "step": 4816
+    },
+    {
+      "epoch": 9.16,
+      "grad_norm": 0.017794422805309296,
+      "learning_rate": 1.1092333536017495e-05,
+      "loss": 0.0004,
+      "step": 4817
+    },
+    {
+      "epoch": 9.16,
+      "grad_norm": 0.05798433721065521,
+      "learning_rate": 1.1042620963678906e-05,
+      "loss": 0.0006,
+      "step": 4818
+    },
+    {
+      "epoch": 9.16,
+      "grad_norm": 0.007508455775678158,
+      "learning_rate": 1.0993017951488802e-05,
+      "loss": 0.0002,
+      "step": 4819
+    },
+    {
+      "epoch": 9.16,
+      "grad_norm": 0.015753276646137238,
+      "learning_rate": 1.0943524518254642e-05,
+      "loss": 0.0002,
+      "step": 4820
+    },
+    {
+      "epoch": 9.17,
+      "grad_norm": 0.02246587909758091,
+      "learning_rate": 1.0894140682742148e-05,
+      "loss": 0.0005,
+      "step": 4821
+    },
+    {
+      "epoch": 9.17,
+      "grad_norm": 0.004621716216206551,
+      "learning_rate": 1.084486646367554e-05,
+      "loss": 0.0001,
+      "step": 4822
+    },
+    {
+      "epoch": 9.17,
+      "grad_norm": 0.017892887815833092,
+      "learning_rate": 1.079570187973744e-05,
+      "loss": 0.0004,
+      "step": 4823
+    },
+    {
+      "epoch": 9.17,
+      "grad_norm": 0.03096826933324337,
+      "learning_rate": 1.0746646949569005e-05,
+      "loss": 0.0005,
+      "step": 4824
+    },
+    {
+      "epoch": 9.17,
+      "grad_norm": 0.0093551529571414,
+      "learning_rate": 1.0697701691769689e-05,
+      "loss": 0.0002,
+      "step": 4825
+    },
+    {
+      "epoch": 9.17,
+      "grad_norm": 0.06657213717699051,
+      "learning_rate": 1.0648866124897382e-05,
+      "loss": 0.0011,
+      "step": 4826
+    },
+    {
+      "epoch": 9.18,
+      "grad_norm": 0.007252390030771494,
+      "learning_rate": 1.0600140267468537e-05,
+      "loss": 0.0002,
+      "step": 4827
+    },
+    {
+      "epoch": 9.18,
+      "grad_norm": 0.004124809056520462,
+      "learning_rate": 1.0551524137957745e-05,
+      "loss": 0.0001,
+      "step": 4828
+    },
+    {
+      "epoch": 9.18,
+      "grad_norm": 0.02499365247786045,
+      "learning_rate": 1.050301775479826e-05,
+      "loss": 0.0005,
+      "step": 4829
+    },
+    {
+      "epoch": 9.18,
+      "grad_norm": 0.0033836958464235067,
+      "learning_rate": 1.0454621136381568e-05,
+      "loss": 0.0001,
+      "step": 4830
+    },
+    {
+      "epoch": 9.18,
+      "grad_norm": 0.017911681905388832,
+      "learning_rate": 1.0406334301057595e-05,
+      "loss": 0.0002,
+      "step": 4831
+    },
+    {
+      "epoch": 9.19,
+      "grad_norm": 0.019943321123719215,
+      "learning_rate": 1.035815726713456e-05,
+      "loss": 0.0003,
+      "step": 4832
+    },
+    {
+      "epoch": 9.19,
+      "grad_norm": 0.03875599801540375,
+      "learning_rate": 1.0310090052879183e-05,
+      "loss": 0.0007,
+      "step": 4833
+    },
+    {
+      "epoch": 9.19,
+      "grad_norm": 0.008005265146493912,
+      "learning_rate": 1.0262132676516388e-05,
+      "loss": 0.0001,
+      "step": 4834
+    },
+    {
+      "epoch": 9.19,
+      "grad_norm": 0.005439273081719875,
+      "learning_rate": 1.0214285156229663e-05,
+      "loss": 0.0001,
+      "step": 4835
+    },
+    {
+      "epoch": 9.19,
+      "grad_norm": 0.01605304516851902,
+      "learning_rate": 1.0166547510160594e-05,
+      "loss": 0.0003,
+      "step": 4836
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 0.037182245403528214,
+      "learning_rate": 1.0118919756409305e-05,
+      "loss": 0.0006,
+      "step": 4837
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 0.01738634705543518,
+      "learning_rate": 1.0071401913034182e-05,
+      "loss": 0.0003,
+      "step": 4838
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 0.003443761495873332,
+      "learning_rate": 1.0023993998051915e-05,
+      "loss": 0.0001,
+      "step": 4839
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 0.047170743346214294,
+      "learning_rate": 9.976696029437559e-06,
+      "loss": 0.0005,
+      "step": 4840
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 0.010561298578977585,
+      "learning_rate": 9.9295080251244e-06,
+      "loss": 0.0002,
+      "step": 4841
+    },
+    {
+      "epoch": 9.21,
+      "grad_norm": 0.004716386087238789,
+      "learning_rate": 9.88243000300416e-06,
+      "loss": 0.0001,
+      "step": 4842
+    },
+    {
+      "epoch": 9.21,
+      "grad_norm": 0.006460041273385286,
+      "learning_rate": 9.83546198092673e-06,
+      "loss": 0.0001,
+      "step": 4843
+    },
+    {
+      "epoch": 9.21,
+      "grad_norm": 0.05199669674038887,
+      "learning_rate": 9.78860397670046e-06,
+      "loss": 0.0005,
+      "step": 4844
+    },
+    {
+      "epoch": 9.21,
+      "grad_norm": 0.004283174406737089,
+      "learning_rate": 9.741856008091742e-06,
+      "loss": 0.0001,
+      "step": 4845
+    },
+    {
+      "epoch": 9.21,
+      "grad_norm": 0.027619214728474617,
+      "learning_rate": 9.695218092825462e-06,
+      "loss": 0.0005,
+      "step": 4846
+    },
+    {
+      "epoch": 9.21,
+      "grad_norm": 0.007183215580880642,
+      "learning_rate": 9.648690248584679e-06,
+      "loss": 0.0002,
+      "step": 4847
+    },
+    {
+      "epoch": 9.22,
+      "grad_norm": 0.027417270466685295,
+      "learning_rate": 9.602272493010776e-06,
+      "loss": 0.0003,
+      "step": 4848
+    },
+    {
+      "epoch": 9.22,
+      "grad_norm": 0.0264898668974638,
+      "learning_rate": 9.555964843703345e-06,
+      "loss": 0.0004,
+      "step": 4849
+    },
+    {
+      "epoch": 9.22,
+      "grad_norm": 0.03782255947589874,
+      "learning_rate": 9.509767318220207e-06,
+      "loss": 0.0009,
+      "step": 4850
+    },
+    {
+      "epoch": 9.22,
+      "grad_norm": 0.025884239003062248,
+      "learning_rate": 9.463679934077517e-06,
+      "loss": 0.0003,
+      "step": 4851
+    },
+    {
+      "epoch": 9.22,
+      "grad_norm": 0.02375505305826664,
+      "learning_rate": 9.41770270874953e-06,
+      "loss": 0.0003,
+      "step": 4852
+    },
+    {
+      "epoch": 9.23,
+      "grad_norm": 0.011514200828969479,
+      "learning_rate": 9.37183565966897e-06,
+      "loss": 0.0002,
+      "step": 4853
+    },
+    {
+      "epoch": 9.23,
+      "grad_norm": 0.05336970463395119,
+      "learning_rate": 9.326078804226489e-06,
+      "loss": 0.0021,
+      "step": 4854
+    },
+    {
+      "epoch": 9.23,
+      "grad_norm": 0.039266880601644516,
+      "learning_rate": 9.280432159771245e-06,
+      "loss": 0.0008,
+      "step": 4855
+    },
+    {
+      "epoch": 9.23,
+      "grad_norm": 0.08077745884656906,
+      "learning_rate": 9.23489574361036e-06,
+      "loss": 0.0012,
+      "step": 4856
+    },
+    {
+      "epoch": 9.23,
+      "grad_norm": 0.015188457444310188,
+      "learning_rate": 9.189469573009357e-06,
+      "loss": 0.0003,
+      "step": 4857
+    },
+    {
+      "epoch": 9.24,
+      "grad_norm": 0.03841094300150871,
+      "learning_rate": 9.14415366519179e-06,
+      "loss": 0.0007,
+      "step": 4858
+    },
+    {
+      "epoch": 9.24,
+      "grad_norm": 0.04238761216402054,
+      "learning_rate": 9.098948037339582e-06,
+      "loss": 0.0012,
+      "step": 4859
+    },
+    {
+      "epoch": 9.24,
+      "grad_norm": 0.05981672182679176,
+      "learning_rate": 9.053852706592723e-06,
+      "loss": 0.0011,
+      "step": 4860
+    },
+    {
+      "epoch": 9.24,
+      "grad_norm": 0.011605408973991871,
+      "learning_rate": 9.00886769004937e-06,
+      "loss": 0.0004,
+      "step": 4861
+    },
+    {
+      "epoch": 9.24,
+      "grad_norm": 0.008636330254375935,
+      "learning_rate": 8.963993004765946e-06,
+      "loss": 0.0002,
+      "step": 4862
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 0.01107928715646267,
+      "learning_rate": 8.919228667756939e-06,
+      "loss": 0.0002,
+      "step": 4863
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 0.020969141274690628,
+      "learning_rate": 8.874574695995174e-06,
+      "loss": 0.0004,
+      "step": 4864
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 0.007346865721046925,
+      "learning_rate": 8.83003110641134e-06,
+      "loss": 0.0002,
+      "step": 4865
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 0.02260327897965908,
+      "learning_rate": 8.78559791589456e-06,
+      "loss": 0.0003,
+      "step": 4866
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 0.012212979607284069,
+      "learning_rate": 8.741275141291927e-06,
+      "loss": 0.0002,
+      "step": 4867
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 0.00719645107164979,
+      "learning_rate": 8.69706279940876e-06,
+      "loss": 0.0001,
+      "step": 4868
+    },
+    {
+      "epoch": 9.26,
+      "grad_norm": 0.008783864788711071,
+      "learning_rate": 8.652960907008388e-06,
+      "loss": 0.0002,
+      "step": 4869
+    },
+    {
+      "epoch": 9.26,
+      "grad_norm": 0.013938911259174347,
+      "learning_rate": 8.608969480812466e-06,
+      "loss": 0.0004,
+      "step": 4870
+    },
+    {
+      "epoch": 9.26,
+      "grad_norm": 0.023678578436374664,
+      "learning_rate": 8.565088537500553e-06,
+      "loss": 0.0004,
+      "step": 4871
+    },
+    {
+      "epoch": 9.26,
+      "grad_norm": 0.007330070249736309,
+      "learning_rate": 8.521318093710372e-06,
+      "loss": 0.0002,
+      "step": 4872
+    },
+    {
+      "epoch": 9.26,
+      "grad_norm": 0.05874321609735489,
+      "learning_rate": 8.477658166037948e-06,
+      "loss": 0.001,
+      "step": 4873
+    },
+    {
+      "epoch": 9.27,
+      "grad_norm": 0.004183645825833082,
+      "learning_rate": 8.43410877103704e-06,
+      "loss": 0.0001,
+      "step": 4874
+    },
+    {
+      "epoch": 9.27,
+      "grad_norm": 0.007597120013087988,
+      "learning_rate": 8.390669925219807e-06,
+      "loss": 0.0002,
+      "step": 4875
+    },
+    {
+      "epoch": 9.27,
+      "grad_norm": 0.095968097448349,
+      "learning_rate": 8.347341645056372e-06,
+      "loss": 0.0018,
+      "step": 4876
+    },
+    {
+      "epoch": 9.27,
+      "grad_norm": 0.012134694494307041,
+      "learning_rate": 8.304123946974927e-06,
+      "loss": 0.0003,
+      "step": 4877
+    },
+    {
+      "epoch": 9.27,
+      "grad_norm": 0.03951236605644226,
+      "learning_rate": 8.261016847361735e-06,
+      "loss": 0.0004,
+      "step": 4878
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 0.04378756880760193,
+      "learning_rate": 8.218020362561218e-06,
+      "loss": 0.0006,
+      "step": 4879
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 0.030490253120660782,
+      "learning_rate": 8.175134508875703e-06,
+      "loss": 0.0005,
+      "step": 4880
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 0.01803063414990902,
+      "learning_rate": 8.132359302565717e-06,
+      "loss": 0.0004,
+      "step": 4881
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 0.01897231675684452,
+      "learning_rate": 8.089694759849718e-06,
+      "loss": 0.0002,
+      "step": 4882
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 0.06074994429945946,
+      "learning_rate": 8.04714089690427e-06,
+      "loss": 0.0025,
+      "step": 4883
+    },
+    {
+      "epoch": 9.29,
+      "grad_norm": 0.07157880067825317,
+      "learning_rate": 8.004697729863996e-06,
+      "loss": 0.0024,
+      "step": 4884
+    },
+    {
+      "epoch": 9.29,
+      "grad_norm": 0.034503739327192307,
+      "learning_rate": 7.962365274821459e-06,
+      "loss": 0.0007,
+      "step": 4885
+    },
+    {
+      "epoch": 9.29,
+      "grad_norm": 0.007035516202449799,
+      "learning_rate": 7.92014354782735e-06,
+      "loss": 0.0001,
+      "step": 4886
+    },
+    {
+      "epoch": 9.29,
+      "grad_norm": 0.01385193970054388,
+      "learning_rate": 7.87803256489027e-06,
+      "loss": 0.0002,
+      "step": 4887
+    },
+    {
+      "epoch": 9.29,
+      "grad_norm": 0.054474975913763046,
+      "learning_rate": 7.836032341976906e-06,
+      "loss": 0.001,
+      "step": 4888
+    },
+    {
+      "epoch": 9.29,
+      "grad_norm": 0.03017285466194153,
+      "learning_rate": 7.79414289501189e-06,
+      "loss": 0.001,
+      "step": 4889
+    },
+    {
+      "epoch": 9.3,
+      "grad_norm": 0.029979275539517403,
+      "learning_rate": 7.752364239878017e-06,
+      "loss": 0.0005,
+      "step": 4890
+    },
+    {
+      "epoch": 9.3,
+      "grad_norm": 0.008087504655122757,
+      "learning_rate": 7.71069639241575e-06,
+      "loss": 0.0002,
+      "step": 4891
+    },
+    {
+      "epoch": 9.3,
+      "grad_norm": 0.03132373467087746,
+      "learning_rate": 7.669139368423915e-06,
+      "loss": 0.0003,
+      "step": 4892
+    },
+    {
+      "epoch": 9.3,
+      "grad_norm": 0.0377512201666832,
+      "learning_rate": 7.627693183659045e-06,
+      "loss": 0.0004,
+      "step": 4893
+    },
+    {
+      "epoch": 9.3,
+      "grad_norm": 0.00730771804228425,
+      "learning_rate": 7.586357853835768e-06,
+      "loss": 0.0002,
+      "step": 4894
+    },
+    {
+      "epoch": 9.31,
+      "grad_norm": 0.014911551959812641,
+      "learning_rate": 7.545133394626646e-06,
+      "loss": 0.0003,
+      "step": 4895
+    },
+    {
+      "epoch": 9.31,
+      "grad_norm": 0.03126543015241623,
+      "learning_rate": 7.504019821662211e-06,
+      "loss": 0.0004,
+      "step": 4896
+    },
+    {
+      "epoch": 9.31,
+      "grad_norm": 0.029922982677817345,
+      "learning_rate": 7.4630171505309585e-06,
+      "loss": 0.0006,
+      "step": 4897
+    },
+    {
+      "epoch": 9.31,
+      "grad_norm": 0.04241274669766426,
+      "learning_rate": 7.422125396779322e-06,
+      "loss": 0.0014,
+      "step": 4898
+    },
+    {
+      "epoch": 9.31,
+      "grad_norm": 0.01718105375766754,
+      "learning_rate": 7.38134457591173e-06,
+      "loss": 0.0002,
+      "step": 4899
+    },
+    {
+      "epoch": 9.32,
+      "grad_norm": 0.006644503679126501,
+      "learning_rate": 7.3406747033904145e-06,
+      "loss": 0.0001,
+      "step": 4900
+    },
+    {
+      "epoch": 9.32,
+      "eval_blimp_filtered_avg": 0.5402985074626866,
+      "eval_blimp_filtered_std": 0.005865009673902528,
+      "step": 4900
+    },
+    {
+      "epoch": 9.32,
+      "eval_blimp_supplement_avg": 0.5538793103448276,
+      "eval_blimp_supplement_std": 0.02311760107885112,
+      "step": 4900
+    },
+    {
+      "epoch": 9.32,
+      "eval_vqa_filtered_avg": 0.12,
+      "eval_vqa_filtered_std": 0.03265986323710906,
+      "step": 4900
+    },
+    {
+      "epoch": 9.32,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 4900
+    },
+    {
+      "epoch": 9.32,
+      "grad_norm": 0.007759585045278072,
+      "learning_rate": 7.300115794635708e-06,
+      "loss": 0.0002,
+      "step": 4901
+    },
+    {
+      "epoch": 9.32,
+      "grad_norm": 0.0020849970169365406,
+      "learning_rate": 7.259667865025776e-06,
+      "loss": 0.0001,
+      "step": 4902
+    },
+    {
+      "epoch": 9.32,
+      "grad_norm": 0.026066191494464874,
+      "learning_rate": 7.219330929896749e-06,
+      "loss": 0.0004,
+      "step": 4903
+    },
+    {
+      "epoch": 9.32,
+      "grad_norm": 0.04330383986234665,
+      "learning_rate": 7.179105004542629e-06,
+      "loss": 0.0008,
+      "step": 4904
+    },
+    {
+      "epoch": 9.33,
+      "grad_norm": 0.014445922337472439,
+      "learning_rate": 7.138990104215314e-06,
+      "loss": 0.0002,
+      "step": 4905
+    },
+    {
+      "epoch": 9.33,
+      "grad_norm": 0.005260587204247713,
+      "learning_rate": 7.098986244124705e-06,
+      "loss": 0.0001,
+      "step": 4906
+    },
+    {
+      "epoch": 9.33,
+      "grad_norm": 0.01680564507842064,
+      "learning_rate": 7.059093439438468e-06,
+      "loss": 0.0003,
+      "step": 4907
+    },
+    {
+      "epoch": 9.33,
+      "grad_norm": 0.02826097048819065,
+      "learning_rate": 7.0193117052823045e-06,
+      "loss": 0.0005,
+      "step": 4908
+    },
+    {
+      "epoch": 9.33,
+      "grad_norm": 0.01626162976026535,
+      "learning_rate": 6.9796410567396486e-06,
+      "loss": 0.0003,
+      "step": 4909
+    },
+    {
+      "epoch": 9.33,
+      "grad_norm": 0.023623215034604073,
+      "learning_rate": 6.940081508851969e-06,
+      "loss": 0.0003,
+      "step": 4910
+    },
+    {
+      "epoch": 9.34,
+      "grad_norm": 0.021747058257460594,
+      "learning_rate": 6.900633076618467e-06,
+      "loss": 0.0003,
+      "step": 4911
+    },
+    {
+      "epoch": 9.34,
+      "grad_norm": 0.0523885153234005,
+      "learning_rate": 6.8612957749963445e-06,
+      "loss": 0.0008,
+      "step": 4912
+    },
+    {
+      "epoch": 9.34,
+      "grad_norm": 0.03980701044201851,
+      "learning_rate": 6.822069618900605e-06,
+      "loss": 0.0004,
+      "step": 4913
+    },
+    {
+      "epoch": 9.34,
+      "grad_norm": 0.03702656179666519,
+      "learning_rate": 6.7829546232040514e-06,
+      "loss": 0.0005,
+      "step": 4914
+    },
+    {
+      "epoch": 9.34,
+      "grad_norm": 0.05022647604346275,
+      "learning_rate": 6.743950802737419e-06,
+      "loss": 0.0006,
+      "step": 4915
+    },
+    {
+      "epoch": 9.35,
+      "grad_norm": 0.007565234322100878,
+      "learning_rate": 6.705058172289346e-06,
+      "loss": 0.0001,
+      "step": 4916
+    },
+    {
+      "epoch": 9.35,
+      "grad_norm": 0.010308542288839817,
+      "learning_rate": 6.666276746606169e-06,
+      "loss": 0.0002,
+      "step": 4917
+    },
+    {
+      "epoch": 9.35,
+      "grad_norm": 0.014681599102914333,
+      "learning_rate": 6.627606540392094e-06,
+      "loss": 0.0002,
+      "step": 4918
+    },
+    {
+      "epoch": 9.35,
+      "grad_norm": 0.02058319002389908,
+      "learning_rate": 6.589047568309291e-06,
+      "loss": 0.0005,
+      "step": 4919
+    },
+    {
+      "epoch": 9.35,
+      "grad_norm": 0.020656712353229523,
+      "learning_rate": 6.550599844977533e-06,
+      "loss": 0.0003,
+      "step": 4920
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 0.009465417824685574,
+      "learning_rate": 6.512263384974658e-06,
+      "loss": 0.0002,
+      "step": 4921
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 0.043316178023815155,
+      "learning_rate": 6.474038202836107e-06,
+      "loss": 0.0006,
+      "step": 4922
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 0.013390801846981049,
+      "learning_rate": 6.435924313055285e-06,
+      "loss": 0.0001,
+      "step": 4923
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 0.030062638223171234,
+      "learning_rate": 6.397921730083233e-06,
+      "loss": 0.0006,
+      "step": 4924
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 0.032758016139268875,
+      "learning_rate": 6.360030468329025e-06,
+      "loss": 0.001,
+      "step": 4925
+    },
+    {
+      "epoch": 9.37,
+      "grad_norm": 0.017612773925065994,
+      "learning_rate": 6.322250542159268e-06,
+      "loss": 0.0004,
+      "step": 4926
+    },
+    {
+      "epoch": 9.37,
+      "grad_norm": 0.008636044338345528,
+      "learning_rate": 6.284581965898539e-06,
+      "loss": 0.0001,
+      "step": 4927
+    },
+    {
+      "epoch": 9.37,
+      "grad_norm": 0.051353469491004944,
+      "learning_rate": 6.2470247538291106e-06,
+      "loss": 0.0006,
+      "step": 4928
+    },
+    {
+      "epoch": 9.37,
+      "grad_norm": 0.005215273704379797,
+      "learning_rate": 6.2095789201910606e-06,
+      "loss": 0.0001,
+      "step": 4929
+    },
+    {
+      "epoch": 9.37,
+      "grad_norm": 0.014918804168701172,
+      "learning_rate": 6.1722444791822316e-06,
+      "loss": 0.0002,
+      "step": 4930
+    },
+    {
+      "epoch": 9.37,
+      "grad_norm": 0.009118861518800259,
+      "learning_rate": 6.135021444958266e-06,
+      "loss": 0.0002,
+      "step": 4931
+    },
+    {
+      "epoch": 9.38,
+      "grad_norm": 0.051265668123960495,
+      "learning_rate": 6.097909831632475e-06,
+      "loss": 0.0007,
+      "step": 4932
+    },
+    {
+      "epoch": 9.38,
+      "grad_norm": 0.02163768745958805,
+      "learning_rate": 6.060909653275969e-06,
+      "loss": 0.0004,
+      "step": 4933
+    },
+    {
+      "epoch": 9.38,
+      "grad_norm": 0.005135472398251295,
+      "learning_rate": 6.024020923917694e-06,
+      "loss": 0.0001,
+      "step": 4934
+    },
+    {
+      "epoch": 9.38,
+      "grad_norm": 0.028298521414399147,
+      "learning_rate": 5.9872436575441605e-06,
+      "loss": 0.0008,
+      "step": 4935
+    },
+    {
+      "epoch": 9.38,
+      "grad_norm": 0.11704280227422714,
+      "learning_rate": 5.950577868099815e-06,
+      "loss": 0.002,
+      "step": 4936
+    },
+    {
+      "epoch": 9.39,
+      "grad_norm": 0.01735849678516388,
+      "learning_rate": 5.914023569486636e-06,
+      "loss": 0.0002,
+      "step": 4937
+    },
+    {
+      "epoch": 9.39,
+      "grad_norm": 0.06409275531768799,
+      "learning_rate": 5.877580775564539e-06,
+      "loss": 0.0021,
+      "step": 4938
+    },
+    {
+      "epoch": 9.39,
+      "grad_norm": 0.015518237836658955,
+      "learning_rate": 5.841249500151002e-06,
+      "loss": 0.0004,
+      "step": 4939
+    },
+    {
+      "epoch": 9.39,
+      "grad_norm": 0.012582876719534397,
+      "learning_rate": 5.805029757021307e-06,
+      "loss": 0.0002,
+      "step": 4940
+    },
+    {
+      "epoch": 9.39,
+      "grad_norm": 0.02927466109395027,
+      "learning_rate": 5.768921559908368e-06,
+      "loss": 0.0004,
+      "step": 4941
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 0.04962334781885147,
+      "learning_rate": 5.732924922502868e-06,
+      "loss": 0.0008,
+      "step": 4942
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 0.007469616364687681,
+      "learning_rate": 5.69703985845319e-06,
+      "loss": 0.0001,
+      "step": 4943
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 0.059592898935079575,
+      "learning_rate": 5.6612663813653815e-06,
+      "loss": 0.0009,
+      "step": 4944
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 0.014781125821173191,
+      "learning_rate": 5.62560450480326e-06,
+      "loss": 0.0002,
+      "step": 4945
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 0.020417410880327225,
+      "learning_rate": 5.590054242288144e-06,
+      "loss": 0.0003,
+      "step": 4946
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 0.024615643545985222,
+      "learning_rate": 5.554615607299284e-06,
+      "loss": 0.0003,
+      "step": 4947
+    },
+    {
+      "epoch": 9.41,
+      "grad_norm": 0.018623776733875275,
+      "learning_rate": 5.5192886132733984e-06,
+      "loss": 0.0002,
+      "step": 4948
+    },
+    {
+      "epoch": 9.41,
+      "grad_norm": 0.005950514692813158,
+      "learning_rate": 5.484073273605006e-06,
+      "loss": 0.0001,
+      "step": 4949
+    },
+    {
+      "epoch": 9.41,
+      "grad_norm": 0.02457873895764351,
+      "learning_rate": 5.448969601646225e-06,
+      "loss": 0.0005,
+      "step": 4950
+    },
+    {
+      "epoch": 9.41,
+      "grad_norm": 0.017416909337043762,
+      "learning_rate": 5.41397761070681e-06,
+      "loss": 0.0003,
+      "step": 4951
+    },
+    {
+      "epoch": 9.41,
+      "grad_norm": 0.01658037304878235,
+      "learning_rate": 5.379097314054248e-06,
+      "loss": 0.0004,
+      "step": 4952
+    },
+    {
+      "epoch": 9.42,
+      "grad_norm": 0.009493102319538593,
+      "learning_rate": 5.344328724913627e-06,
+      "loss": 0.0001,
+      "step": 4953
+    },
+    {
+      "epoch": 9.42,
+      "grad_norm": 0.036918871104717255,
+      "learning_rate": 5.309671856467768e-06,
+      "loss": 0.0009,
+      "step": 4954
+    },
+    {
+      "epoch": 9.42,
+      "grad_norm": 0.017537033185362816,
+      "learning_rate": 5.275126721856926e-06,
+      "loss": 0.0005,
+      "step": 4955
+    },
+    {
+      "epoch": 9.42,
+      "grad_norm": 0.0049637313932180405,
+      "learning_rate": 5.240693334179258e-06,
+      "loss": 0.0001,
+      "step": 4956
+    },
+    {
+      "epoch": 9.42,
+      "grad_norm": 0.03268401697278023,
+      "learning_rate": 5.206371706490254e-06,
+      "loss": 0.0005,
+      "step": 4957
+    },
+    {
+      "epoch": 9.43,
+      "grad_norm": 0.02995353564620018,
+      "learning_rate": 5.172161851803369e-06,
+      "loss": 0.0004,
+      "step": 4958
+    },
+    {
+      "epoch": 9.43,
+      "grad_norm": 0.05734610557556152,
+      "learning_rate": 5.138063783089363e-06,
+      "loss": 0.0009,
+      "step": 4959
+    },
+    {
+      "epoch": 9.43,
+      "grad_norm": 0.030067462474107742,
+      "learning_rate": 5.104077513276827e-06,
+      "loss": 0.0007,
+      "step": 4960
+    },
+    {
+      "epoch": 9.43,
+      "grad_norm": 0.08131217211484909,
+      "learning_rate": 5.070203055251853e-06,
+      "loss": 0.0011,
+      "step": 4961
+    },
+    {
+      "epoch": 9.43,
+      "grad_norm": 0.011139553040266037,
+      "learning_rate": 5.036440421858235e-06,
+      "loss": 0.0002,
+      "step": 4962
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 0.006351503077894449,
+      "learning_rate": 5.0027896258971965e-06,
+      "loss": 0.0001,
+      "step": 4963
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 0.007486308459192514,
+      "learning_rate": 4.969250680127701e-06,
+      "loss": 0.0002,
+      "step": 4964
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 0.014201883226633072,
+      "learning_rate": 4.935823597266342e-06,
+      "loss": 0.0002,
+      "step": 4965
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 0.048011649399995804,
+      "learning_rate": 4.902508389987147e-06,
+      "loss": 0.0011,
+      "step": 4966
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 0.014483626000583172,
+      "learning_rate": 4.869305070921814e-06,
+      "loss": 0.0003,
+      "step": 4967
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 0.015461734496057034,
+      "learning_rate": 4.8362136526596374e-06,
+      "loss": 0.0003,
+      "step": 4968
+    },
+    {
+      "epoch": 9.45,
+      "grad_norm": 0.01282887440174818,
+      "learning_rate": 4.803234147747415e-06,
+      "loss": 0.0003,
+      "step": 4969
+    },
+    {
+      "epoch": 9.45,
+      "grad_norm": 0.016918649896979332,
+      "learning_rate": 4.770366568689543e-06,
+      "loss": 0.0002,
+      "step": 4970
+    },
+    {
+      "epoch": 9.45,
+      "grad_norm": 0.01903160847723484,
+      "learning_rate": 4.737610927948055e-06,
+      "loss": 0.0004,
+      "step": 4971
+    },
+    {
+      "epoch": 9.45,
+      "grad_norm": 0.05678436905145645,
+      "learning_rate": 4.704967237942381e-06,
+      "loss": 0.0014,
+      "step": 4972
+    },
+    {
+      "epoch": 9.45,
+      "grad_norm": 0.02332654595375061,
+      "learning_rate": 4.672435511049688e-06,
+      "loss": 0.0005,
+      "step": 4973
+    },
+    {
+      "epoch": 9.46,
+      "grad_norm": 0.010140757076442242,
+      "learning_rate": 4.640015759604543e-06,
+      "loss": 0.0003,
+      "step": 4974
+    },
+    {
+      "epoch": 9.46,
+      "grad_norm": 0.03325306251645088,
+      "learning_rate": 4.607707995899112e-06,
+      "loss": 0.0006,
+      "step": 4975
+    },
+    {
+      "epoch": 9.46,
+      "grad_norm": 0.012683757580816746,
+      "learning_rate": 4.575512232183165e-06,
+      "loss": 0.0003,
+      "step": 4976
+    },
+    {
+      "epoch": 9.46,
+      "grad_norm": 0.014531539753079414,
+      "learning_rate": 4.543428480663868e-06,
+      "loss": 0.0002,
+      "step": 4977
+    },
+    {
+      "epoch": 9.46,
+      "grad_norm": 0.009568165987730026,
+      "learning_rate": 4.511456753506026e-06,
+      "loss": 0.0002,
+      "step": 4978
+    },
+    {
+      "epoch": 9.47,
+      "grad_norm": 0.005120296496897936,
+      "learning_rate": 4.479597062831941e-06,
+      "loss": 0.0001,
+      "step": 4979
+    },
+    {
+      "epoch": 9.47,
+      "grad_norm": 0.010656212456524372,
+      "learning_rate": 4.447849420721416e-06,
+      "loss": 0.0001,
+      "step": 4980
+    },
+    {
+      "epoch": 9.47,
+      "grad_norm": 0.011787001974880695,
+      "learning_rate": 4.4162138392117905e-06,
+      "loss": 0.0002,
+      "step": 4981
+    },
+    {
+      "epoch": 9.47,
+      "grad_norm": 0.05115697905421257,
+      "learning_rate": 4.384690330297935e-06,
+      "loss": 0.0016,
+      "step": 4982
+    },
+    {
+      "epoch": 9.47,
+      "grad_norm": 0.0037611525040119886,
+      "learning_rate": 4.353278905932123e-06,
+      "loss": 0.0001,
+      "step": 4983
+    },
+    {
+      "epoch": 9.48,
+      "grad_norm": 0.0069047268480062485,
+      "learning_rate": 4.321979578024293e-06,
+      "loss": 0.0002,
+      "step": 4984
+    },
+    {
+      "epoch": 9.48,
+      "grad_norm": 0.019202005118131638,
+      "learning_rate": 4.290792358441786e-06,
+      "loss": 0.0004,
+      "step": 4985
+    },
+    {
+      "epoch": 9.48,
+      "grad_norm": 0.053613726049661636,
+      "learning_rate": 4.25971725900941e-06,
+      "loss": 0.0006,
+      "step": 4986
+    },
+    {
+      "epoch": 9.48,
+      "grad_norm": 0.006505091208964586,
+      "learning_rate": 4.22875429150954e-06,
+      "loss": 0.0001,
+      "step": 4987
+    },
+    {
+      "epoch": 9.48,
+      "grad_norm": 0.10912220925092697,
+      "learning_rate": 4.197903467681984e-06,
+      "loss": 0.0031,
+      "step": 4988
+    },
+    {
+      "epoch": 9.48,
+      "grad_norm": 0.047009631991386414,
+      "learning_rate": 4.167164799223987e-06,
+      "loss": 0.0008,
+      "step": 4989
+    },
+    {
+      "epoch": 9.49,
+      "grad_norm": 0.024890968576073647,
+      "learning_rate": 4.136538297790392e-06,
+      "loss": 0.0004,
+      "step": 4990
+    },
+    {
+      "epoch": 9.49,
+      "grad_norm": 0.047617118805646896,
+      "learning_rate": 4.106023974993444e-06,
+      "loss": 0.0009,
+      "step": 4991
+    },
+    {
+      "epoch": 9.49,
+      "grad_norm": 0.0934872254729271,
+      "learning_rate": 4.075621842402755e-06,
+      "loss": 0.0012,
+      "step": 4992
+    },
+    {
+      "epoch": 9.49,
+      "grad_norm": 0.01009327918291092,
+      "learning_rate": 4.045331911545635e-06,
+      "loss": 0.0001,
+      "step": 4993
+    },
+    {
+      "epoch": 9.49,
+      "grad_norm": 0.007179752457886934,
+      "learning_rate": 4.015154193906667e-06,
+      "loss": 0.0001,
+      "step": 4994
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 0.03128325566649437,
+      "learning_rate": 3.985088700927896e-06,
+      "loss": 0.0006,
+      "step": 4995
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 0.054751988500356674,
+      "learning_rate": 3.9551354440089056e-06,
+      "loss": 0.0006,
+      "step": 4996
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 0.0178584773093462,
+      "learning_rate": 3.925294434506676e-06,
+      "loss": 0.0003,
+      "step": 4997
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 0.04130968824028969,
+      "learning_rate": 3.895565683735591e-06,
+      "loss": 0.0021,
+      "step": 4998
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 0.024489473551511765,
+      "learning_rate": 3.8659492029675e-06,
+      "loss": 0.0005,
+      "step": 4999
+    },
+    {
+      "epoch": 9.51,
+      "grad_norm": 0.06334065645933151,
+      "learning_rate": 3.836445003431787e-06,
+      "loss": 0.0009,
+      "step": 5000
+    },
+    {
+      "epoch": 9.51,
+      "eval_blimp_filtered_avg": 0.538955223880597,
+      "eval_blimp_filtered_std": 0.005865642678662206,
+      "step": 5000
+    },
+    {
+      "epoch": 9.51,
+      "eval_blimp_supplement_avg": 0.5517241379310345,
+      "eval_blimp_supplement_std": 0.023126739123082884,
+      "step": 5000
+    },
+    {
+      "epoch": 9.51,
+      "eval_vqa_filtered_avg": 0.13,
+      "eval_vqa_filtered_std": 0.0337997668989631,
+      "step": 5000
+    },
+    {
+      "epoch": 9.51,
+      "eval_winoground_filtered_avg": 0.5,
+      "eval_winoground_filtered_std": 0.050251890762960605,
+      "step": 5000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5260,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 1000,
+  "total_flos": 1.2583758798352548e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}