diff --git "a/checkpoint-1984307/trainer_state.json" "b/checkpoint-1984307/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1984307/trainer_state.json"
@@ -0,0 +1,23483 @@
+{
+  "best_metric": 3.99186372756958,
+  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/rel-cl/lstm/1/checkpoints/checkpoint-1984307",
+  "epoch": 0.025000606015738065,
+  "eval_steps": 10,
+  "global_step": 1984307,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998362119627e-05,
+      "loss": 10.8173,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999161405248948e-05,
+      "loss": 7.5621,
+      "step": 512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998322810497896e-05,
+      "loss": 7.0552,
+      "step": 1024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997484215746844e-05,
+      "loss": 6.9778,
+      "step": 1536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996645620995792e-05,
+      "loss": 6.9512,
+      "step": 2048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99580702624474e-05,
+      "loss": 6.8598,
+      "step": 2560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994968431493688e-05,
+      "loss": 6.7142,
+      "step": 3072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994129836742636e-05,
+      "loss": 6.6064,
+      "step": 3584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993291241991584e-05,
+      "loss": 6.5223,
+      "step": 4096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992452647240532e-05,
+      "loss": 6.4391,
+      "step": 4608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99161405248948e-05,
+      "loss": 6.405,
+      "step": 5120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.990775457738428e-05,
+      "loss": 6.347,
+      "step": 5632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989936862987376e-05,
+      "loss": 6.2854,
+      "step": 6144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989098268236324e-05,
+      "loss": 6.2171,
+      "step": 6656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.988259673485272e-05,
+      "loss": 6.1624,
+      "step": 7168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.98742107873422e-05,
+      "loss": 6.1169,
+      "step": 7680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.986582483983168e-05,
+      "loss": 6.0706,
+      "step": 8192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.985743889232116e-05,
+      "loss": 6.0137,
+      "step": 8704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984905294481064e-05,
+      "loss": 5.9735,
+      "step": 9216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984066699730012e-05,
+      "loss": 5.9362,
+      "step": 9728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.983229742859333e-05,
+      "loss": 5.8977,
+      "step": 10240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.982391148108281e-05,
+      "loss": 5.8699,
+      "step": 10752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.981552553357229e-05,
+      "loss": 5.8394,
+      "step": 11264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9807155964865506e-05,
+      "loss": 5.807,
+      "step": 11776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9798770017354986e-05,
+      "loss": 5.7783,
+      "step": 12288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9790384069844466e-05,
+      "loss": 5.7604,
+      "step": 12800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9781998122333946e-05,
+      "loss": 5.7276,
+      "step": 13312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9773612174823426e-05,
+      "loss": 5.7055,
+      "step": 13824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9765242606116635e-05,
+      "loss": 5.6716,
+      "step": 14336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9756856658606115e-05,
+      "loss": 5.667,
+      "step": 14848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9748470711095595e-05,
+      "loss": 5.6342,
+      "step": 15360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.974008476358507e-05,
+      "loss": 5.6179,
+      "step": 15872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.973169881607455e-05,
+      "loss": 5.6075,
+      "step": 16384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.972331286856403e-05,
+      "loss": 5.589,
+      "step": 16896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9714926921053515e-05,
+      "loss": 5.5594,
+      "step": 17408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9706540973542995e-05,
+      "loss": 5.5456,
+      "step": 17920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9698171404836204e-05,
+      "loss": 5.5373,
+      "step": 18432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9689785457325684e-05,
+      "loss": 5.5153,
+      "step": 18944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9681399509815164e-05,
+      "loss": 5.5032,
+      "step": 19456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9673013562304644e-05,
+      "loss": 5.4829,
+      "step": 19968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.966464399359785e-05,
+      "loss": 5.4577,
+      "step": 20480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965625804608733e-05,
+      "loss": 5.4464,
+      "step": 20992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.964787209857681e-05,
+      "loss": 5.4322,
+      "step": 21504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.963948615106629e-05,
+      "loss": 5.4251,
+      "step": 22016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96311165823595e-05,
+      "loss": 5.4132,
+      "step": 22528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.962273063484898e-05,
+      "loss": 5.4073,
+      "step": 23040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.961434468733847e-05,
+      "loss": 5.3832,
+      "step": 23552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.960595873982795e-05,
+      "loss": 5.3828,
+      "step": 24064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.959757279231743e-05,
+      "loss": 5.3878,
+      "step": 24576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958920322361064e-05,
+      "loss": 5.3593,
+      "step": 25088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958081727610012e-05,
+      "loss": 5.3489,
+      "step": 25600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.95724313285896e-05,
+      "loss": 5.3364,
+      "step": 26112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.956404538107908e-05,
+      "loss": 5.3456,
+      "step": 26624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.955565943356856e-05,
+      "loss": 5.3071,
+      "step": 27136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.954732262246923e-05,
+      "loss": 5.3256,
+      "step": 27648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.953893667495871e-05,
+      "loss": 5.2948,
+      "step": 28160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.953055072744819e-05,
+      "loss": 5.2928,
+      "step": 28672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.952216477993767e-05,
+      "loss": 5.2861,
+      "step": 29184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.951377883242715e-05,
+      "loss": 5.2759,
+      "step": 29696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.950539288491663e-05,
+      "loss": 5.2628,
+      "step": 30208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.949700693740611e-05,
+      "loss": 5.2518,
+      "step": 30720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948862098989559e-05,
+      "loss": 5.2376,
+      "step": 31232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948023504238507e-05,
+      "loss": 5.2442,
+      "step": 31744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.947186547367828e-05,
+      "loss": 5.2314,
+      "step": 32256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.946347952616776e-05,
+      "loss": 5.2213,
+      "step": 32768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.945509357865724e-05,
+      "loss": 5.2138,
+      "step": 33280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.944670763114672e-05,
+      "loss": 5.2024,
+      "step": 33792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.94383216836362e-05,
+      "loss": 5.1858,
+      "step": 34304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.942993573612568e-05,
+      "loss": 5.203,
+      "step": 34816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.942154978861516e-05,
+      "loss": 5.1795,
+      "step": 35328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.941316384110464e-05,
+      "loss": 5.1736,
+      "step": 35840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.940477789359412e-05,
+      "loss": 5.1824,
+      "step": 36352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.93963919460836e-05,
+      "loss": 5.1575,
+      "step": 36864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9388022377376816e-05,
+      "loss": 5.1701,
+      "step": 37376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9379636429866296e-05,
+      "loss": 5.1621,
+      "step": 37888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9371250482355776e-05,
+      "loss": 5.1505,
+      "step": 38400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9362864534845256e-05,
+      "loss": 5.1492,
+      "step": 38912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.935447858733473e-05,
+      "loss": 5.1231,
+      "step": 39424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.934609263982421e-05,
+      "loss": 5.1214,
+      "step": 39936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.933770669231369e-05,
+      "loss": 5.1192,
+      "step": 40448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9329337123606905e-05,
+      "loss": 5.1184,
+      "step": 40960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.932095117609638e-05,
+      "loss": 5.1077,
+      "step": 41472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.931256522858586e-05,
+      "loss": 5.0946,
+      "step": 41984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.930417928107534e-05,
+      "loss": 5.0808,
+      "step": 42496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9295793333564825e-05,
+      "loss": 5.0878,
+      "step": 43008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9287407386054305e-05,
+      "loss": 5.0868,
+      "step": 43520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9279037817347514e-05,
+      "loss": 5.0809,
+      "step": 44032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9270651869836994e-05,
+      "loss": 5.0794,
+      "step": 44544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9262265922326474e-05,
+      "loss": 5.0745,
+      "step": 45056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9253879974815954e-05,
+      "loss": 5.0617,
+      "step": 45568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9245494027305433e-05,
+      "loss": 5.0432,
+      "step": 46080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.923712445859864e-05,
+      "loss": 5.0402,
+      "step": 46592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922873851108812e-05,
+      "loss": 5.0459,
+      "step": 47104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.92203525635776e-05,
+      "loss": 5.0425,
+      "step": 47616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.921196661606708e-05,
+      "loss": 5.0189,
+      "step": 48128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.920359704736029e-05,
+      "loss": 5.0237,
+      "step": 48640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.919521109984978e-05,
+      "loss": 5.0224,
+      "step": 49152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.918682515233926e-05,
+      "loss": 5.0212,
+      "step": 49664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.917843920482874e-05,
+      "loss": 5.0117,
+      "step": 50176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.917005325731822e-05,
+      "loss": 5.0127,
+      "step": 50688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.916168368861143e-05,
+      "loss": 5.0045,
+      "step": 51200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.915329774110091e-05,
+      "loss": 4.9936,
+      "step": 51712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.914491179359039e-05,
+      "loss": 4.9952,
+      "step": 52224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.913652584607987e-05,
+      "loss": 4.9946,
+      "step": 52736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.912813989856935e-05,
+      "loss": 4.9785,
+      "step": 53248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9119770329862556e-05,
+      "loss": 4.9824,
+      "step": 53760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9111384382352036e-05,
+      "loss": 4.9603,
+      "step": 54272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9102998434841516e-05,
+      "loss": 4.9551,
+      "step": 54784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9094612487330996e-05,
+      "loss": 4.9542,
+      "step": 55296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9086226539820476e-05,
+      "loss": 4.9587,
+      "step": 55808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.907784059230996e-05,
+      "loss": 4.9513,
+      "step": 56320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906945464479944e-05,
+      "loss": 4.9289,
+      "step": 56832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906106869728892e-05,
+      "loss": 4.9339,
+      "step": 57344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.905269912858213e-05,
+      "loss": 4.9377,
+      "step": 57856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.904431318107161e-05,
+      "loss": 4.9436,
+      "step": 58368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.903592723356109e-05,
+      "loss": 4.9289,
+      "step": 58880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90275576648543e-05,
+      "loss": 4.9288,
+      "step": 59392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.901918809614751e-05,
+      "loss": 4.9254,
+      "step": 59904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.901080214863699e-05,
+      "loss": 4.9262,
+      "step": 60416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.900241620112647e-05,
+      "loss": 4.9102,
+      "step": 60928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.899403025361595e-05,
+      "loss": 4.9102,
+      "step": 61440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.898564430610543e-05,
+      "loss": 4.9012,
+      "step": 61952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.897725835859491e-05,
+      "loss": 4.905,
+      "step": 62464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8968872411084396e-05,
+      "loss": 4.8878,
+      "step": 62976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.896048646357387e-05,
+      "loss": 4.8999,
+      "step": 63488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.895210051606335e-05,
+      "loss": 4.8903,
+      "step": 64000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8943730947356565e-05,
+      "loss": 4.8972,
+      "step": 64512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.893534499984604e-05,
+      "loss": 4.8688,
+      "step": 65024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8926975431139254e-05,
+      "loss": 4.8801,
+      "step": 65536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8918589483628734e-05,
+      "loss": 4.8895,
+      "step": 66048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8910203536118214e-05,
+      "loss": 4.8614,
+      "step": 66560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.890181758860769e-05,
+      "loss": 4.882,
+      "step": 67072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.889343164109717e-05,
+      "loss": 4.8646,
+      "step": 67584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.888504569358665e-05,
+      "loss": 4.868,
+      "step": 68096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8876659746076134e-05,
+      "loss": 4.8582,
+      "step": 68608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8868273798565614e-05,
+      "loss": 4.8538,
+      "step": 69120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8859887851055094e-05,
+      "loss": 4.8555,
+      "step": 69632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8851501903544574e-05,
+      "loss": 4.8526,
+      "step": 70144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.884313233483778e-05,
+      "loss": 4.8526,
+      "step": 70656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.883474638732726e-05,
+      "loss": 4.8449,
+      "step": 71168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.882636043981674e-05,
+      "loss": 4.8474,
+      "step": 71680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.881797449230622e-05,
+      "loss": 4.8465,
+      "step": 72192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88095885447957e-05,
+      "loss": 4.8267,
+      "step": 72704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.880120259728518e-05,
+      "loss": 4.8271,
+      "step": 73216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.879283302857839e-05,
+      "loss": 4.818,
+      "step": 73728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.878444708106787e-05,
+      "loss": 4.8204,
+      "step": 74240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.877606113355735e-05,
+      "loss": 4.8187,
+      "step": 74752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.876767518604683e-05,
+      "loss": 4.8119,
+      "step": 75264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.875930561734005e-05,
+      "loss": 4.8099,
+      "step": 75776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.875091966982953e-05,
+      "loss": 4.8102,
+      "step": 76288
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.7706427574157715,
+      "eval_runtime": 545.2364,
+      "eval_samples_per_second": 699.863,
+      "eval_steps_per_second": 21.871,
+      "step": 76319
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.874253372231901e-05,
+      "loss": 4.819,
+      "step": 76800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.873414777480849e-05,
+      "loss": 4.8106,
+      "step": 77312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.872576182729797e-05,
+      "loss": 4.7982,
+      "step": 77824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.871737587978745e-05,
+      "loss": 4.8001,
+      "step": 78336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.870898993227693e-05,
+      "loss": 4.7934,
+      "step": 78848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.87006039847664e-05,
+      "loss": 4.7726,
+      "step": 79360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.869221803725588e-05,
+      "loss": 4.785,
+      "step": 79872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.868383208974536e-05,
+      "loss": 4.7882,
+      "step": 80384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.867544614223484e-05,
+      "loss": 4.7692,
+      "step": 80896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.866706019472432e-05,
+      "loss": 4.785,
+      "step": 81408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.86586742472138e-05,
+      "loss": 4.7886,
+      "step": 81920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.865028829970329e-05,
+      "loss": 4.7729,
+      "step": 82432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.864190235219277e-05,
+      "loss": 4.7658,
+      "step": 82944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.863351640468225e-05,
+      "loss": 4.7677,
+      "step": 83456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.862513045717173e-05,
+      "loss": 4.767,
+      "step": 83968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.861674450966121e-05,
+      "loss": 4.7616,
+      "step": 84480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.860835856215069e-05,
+      "loss": 4.7565,
+      "step": 84992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.859997261464017e-05,
+      "loss": 4.756,
+      "step": 85504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.859158666712964e-05,
+      "loss": 4.7509,
+      "step": 86016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8583233477226585e-05,
+      "loss": 4.7549,
+      "step": 86528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8574847529716065e-05,
+      "loss": 4.7551,
+      "step": 87040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8566461582205545e-05,
+      "loss": 4.7543,
+      "step": 87552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8558075634695025e-05,
+      "loss": 4.7496,
+      "step": 88064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8549689687184505e-05,
+      "loss": 4.7474,
+      "step": 88576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.854132011847772e-05,
+      "loss": 4.7419,
+      "step": 89088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.85329341709672e-05,
+      "loss": 4.7387,
+      "step": 89600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.852454822345668e-05,
+      "loss": 4.7425,
+      "step": 90112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.851616227594616e-05,
+      "loss": 4.723,
+      "step": 90624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.850777632843564e-05,
+      "loss": 4.7354,
+      "step": 91136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8499390380925114e-05,
+      "loss": 4.7177,
+      "step": 91648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8491004433414594e-05,
+      "loss": 4.722,
+      "step": 92160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8482618485904074e-05,
+      "loss": 4.7319,
+      "step": 92672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.847424891719729e-05,
+      "loss": 4.7296,
+      "step": 93184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.846586296968676e-05,
+      "loss": 4.715,
+      "step": 93696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.845747702217624e-05,
+      "loss": 4.707,
+      "step": 94208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.844910745346946e-05,
+      "loss": 4.7239,
+      "step": 94720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.844072150595894e-05,
+      "loss": 4.7108,
+      "step": 95232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.843233555844842e-05,
+      "loss": 4.7093,
+      "step": 95744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.84239496109379e-05,
+      "loss": 4.7055,
+      "step": 96256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.841556366342738e-05,
+      "loss": 4.6932,
+      "step": 96768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.840717771591686e-05,
+      "loss": 4.6894,
+      "step": 97280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.839879176840634e-05,
+      "loss": 4.6913,
+      "step": 97792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.839040582089582e-05,
+      "loss": 4.6954,
+      "step": 98304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.83820198733853e-05,
+      "loss": 4.6896,
+      "step": 98816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.837363392587478e-05,
+      "loss": 4.6924,
+      "step": 99328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.836524797836426e-05,
+      "loss": 4.6844,
+      "step": 99840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.835686203085374e-05,
+      "loss": 4.6879,
+      "step": 100352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.834849246214695e-05,
+      "loss": 4.6993,
+      "step": 100864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.834010651463643e-05,
+      "loss": 4.6794,
+      "step": 101376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.833172056712591e-05,
+      "loss": 4.6801,
+      "step": 101888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.832333461961539e-05,
+      "loss": 4.6698,
+      "step": 102400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8314948672104874e-05,
+      "loss": 4.6902,
+      "step": 102912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.830657910339808e-05,
+      "loss": 4.6616,
+      "step": 103424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.829819315588756e-05,
+      "loss": 4.6838,
+      "step": 103936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.828982358718077e-05,
+      "loss": 4.6604,
+      "step": 104448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.828143763967025e-05,
+      "loss": 4.6699,
+      "step": 104960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.827305169215973e-05,
+      "loss": 4.6682,
+      "step": 105472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.826466574464921e-05,
+      "loss": 4.6641,
+      "step": 105984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.825627979713869e-05,
+      "loss": 4.6559,
+      "step": 106496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.824789384962817e-05,
+      "loss": 4.6559,
+      "step": 107008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.823950790211765e-05,
+      "loss": 4.6452,
+      "step": 107520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8231121954607125e-05,
+      "loss": 4.6602,
+      "step": 108032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.822273600709661e-05,
+      "loss": 4.6511,
+      "step": 108544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.821435005958609e-05,
+      "loss": 4.6454,
+      "step": 109056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.82059804908793e-05,
+      "loss": 4.6438,
+      "step": 109568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.819759454336878e-05,
+      "loss": 4.6443,
+      "step": 110080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.818920859585826e-05,
+      "loss": 4.6261,
+      "step": 110592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.818082264834774e-05,
+      "loss": 4.65,
+      "step": 111104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.817245307964095e-05,
+      "loss": 4.6354,
+      "step": 111616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.816406713213043e-05,
+      "loss": 4.6349,
+      "step": 112128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.815568118461991e-05,
+      "loss": 4.6435,
+      "step": 112640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.814729523710939e-05,
+      "loss": 4.6292,
+      "step": 113152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.813890928959887e-05,
+      "loss": 4.6431,
+      "step": 113664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.813053972089208e-05,
+      "loss": 4.6458,
+      "step": 114176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8122153773381565e-05,
+      "loss": 4.6363,
+      "step": 114688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8113767825871045e-05,
+      "loss": 4.6383,
+      "step": 115200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8105381878360525e-05,
+      "loss": 4.6205,
+      "step": 115712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8096995930850005e-05,
+      "loss": 4.6156,
+      "step": 116224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8088609983339485e-05,
+      "loss": 4.6269,
+      "step": 116736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8080240414632694e-05,
+      "loss": 4.6279,
+      "step": 117248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8071854467122174e-05,
+      "loss": 4.6204,
+      "step": 117760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8063468519611654e-05,
+      "loss": 4.6159,
+      "step": 118272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8055082572101134e-05,
+      "loss": 4.5988,
+      "step": 118784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8046696624590614e-05,
+      "loss": 4.61,
+      "step": 119296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8038310677080094e-05,
+      "loss": 4.6138,
+      "step": 119808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8029924729569574e-05,
+      "loss": 4.6127,
+      "step": 120320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8021538782059054e-05,
+      "loss": 4.6194,
+      "step": 120832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8013152834548534e-05,
+      "loss": 4.6174,
+      "step": 121344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800478326584174e-05,
+      "loss": 4.607,
+      "step": 121856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.799639731833123e-05,
+      "loss": 4.5994,
+      "step": 122368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.798802774962444e-05,
+      "loss": 4.5926,
+      "step": 122880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.797964180211392e-05,
+      "loss": 4.599,
+      "step": 123392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.79712558546034e-05,
+      "loss": 4.6017,
+      "step": 123904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.796286990709288e-05,
+      "loss": 4.5893,
+      "step": 124416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.795448395958236e-05,
+      "loss": 4.5903,
+      "step": 124928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.794611439087557e-05,
+      "loss": 4.5981,
+      "step": 125440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.793772844336505e-05,
+      "loss": 4.6013,
+      "step": 125952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.792934249585453e-05,
+      "loss": 4.5899,
+      "step": 126464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.792095654834401e-05,
+      "loss": 4.596,
+      "step": 126976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.791257060083349e-05,
+      "loss": 4.5978,
+      "step": 127488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.790418465332297e-05,
+      "loss": 4.5883,
+      "step": 128000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.789579870581245e-05,
+      "loss": 4.585,
+      "step": 128512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.788741275830193e-05,
+      "loss": 4.5951,
+      "step": 129024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7879043189595136e-05,
+      "loss": 4.5802,
+      "step": 129536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7870657242084616e-05,
+      "loss": 4.5931,
+      "step": 130048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.786228767337783e-05,
+      "loss": 4.573,
+      "step": 130560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.785390172586731e-05,
+      "loss": 4.5631,
+      "step": 131072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7845515778356785e-05,
+      "loss": 4.5704,
+      "step": 131584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7837129830846265e-05,
+      "loss": 4.5787,
+      "step": 132096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.782876026213948e-05,
+      "loss": 4.572,
+      "step": 132608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.782037431462896e-05,
+      "loss": 4.5589,
+      "step": 133120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7811988367118434e-05,
+      "loss": 4.559,
+      "step": 133632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.780360241960792e-05,
+      "loss": 4.5672,
+      "step": 134144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.77952164720974e-05,
+      "loss": 4.5783,
+      "step": 134656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.778683052458688e-05,
+      "loss": 4.565,
+      "step": 135168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.777844457707636e-05,
+      "loss": 4.5736,
+      "step": 135680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.777005862956584e-05,
+      "loss": 4.5694,
+      "step": 136192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.776168906085905e-05,
+      "loss": 4.5658,
+      "step": 136704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.775331949215226e-05,
+      "loss": 4.5613,
+      "step": 137216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.774493354464174e-05,
+      "loss": 4.5578,
+      "step": 137728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.773654759713122e-05,
+      "loss": 4.5541,
+      "step": 138240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.77281616496207e-05,
+      "loss": 4.5645,
+      "step": 138752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.771977570211018e-05,
+      "loss": 4.5491,
+      "step": 139264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.771138975459966e-05,
+      "loss": 4.5512,
+      "step": 139776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.770300380708914e-05,
+      "loss": 4.5553,
+      "step": 140288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.769461785957862e-05,
+      "loss": 4.5625,
+      "step": 140800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7686248290871835e-05,
+      "loss": 4.5395,
+      "step": 141312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7677862343361315e-05,
+      "loss": 4.543,
+      "step": 141824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7669476395850795e-05,
+      "loss": 4.5655,
+      "step": 142336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7661090448340274e-05,
+      "loss": 4.5387,
+      "step": 142848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7652720879633484e-05,
+      "loss": 4.5605,
+      "step": 143360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.764435131092669e-05,
+      "loss": 4.5427,
+      "step": 143872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.763596536341617e-05,
+      "loss": 4.5539,
+      "step": 144384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.762757941590565e-05,
+      "loss": 4.5414,
+      "step": 144896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.761919346839513e-05,
+      "loss": 4.537,
+      "step": 145408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.761080752088461e-05,
+      "loss": 4.5451,
+      "step": 145920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.760242157337409e-05,
+      "loss": 4.5409,
+      "step": 146432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.759405200466731e-05,
+      "loss": 4.5498,
+      "step": 146944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.758566605715679e-05,
+      "loss": 4.5441,
+      "step": 147456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.757728010964627e-05,
+      "loss": 4.5433,
+      "step": 147968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.756889416213575e-05,
+      "loss": 4.5408,
+      "step": 148480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.756052459342896e-05,
+      "loss": 4.5256,
+      "step": 148992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.755213864591844e-05,
+      "loss": 4.536,
+      "step": 149504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.754375269840792e-05,
+      "loss": 4.5226,
+      "step": 150016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.75353667508974e-05,
+      "loss": 4.5275,
+      "step": 150528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7526997182190606e-05,
+      "loss": 4.5307,
+      "step": 151040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7518611234680086e-05,
+      "loss": 4.518,
+      "step": 151552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7510225287169566e-05,
+      "loss": 4.5247,
+      "step": 152064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7501839339659046e-05,
+      "loss": 4.5218,
+      "step": 152576
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.488372802734375,
+      "eval_runtime": 546.2477,
+      "eval_samples_per_second": 698.568,
+      "eval_steps_per_second": 21.831,
+      "step": 152638
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7493453392148526e-05,
+      "loss": 4.5335,
+      "step": 153088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.748508382344174e-05,
+      "loss": 4.5308,
+      "step": 153600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.747669787593122e-05,
+      "loss": 4.5211,
+      "step": 154112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.74683119284207e-05,
+      "loss": 4.5176,
+      "step": 154624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.745992598091018e-05,
+      "loss": 4.5188,
+      "step": 155136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.745155641220339e-05,
+      "loss": 4.5028,
+      "step": 155648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.744317046469287e-05,
+      "loss": 4.5165,
+      "step": 156160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.743480089598608e-05,
+      "loss": 4.5148,
+      "step": 156672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.742641494847556e-05,
+      "loss": 4.4999,
+      "step": 157184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.741802900096504e-05,
+      "loss": 4.5196,
+      "step": 157696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.740964305345452e-05,
+      "loss": 4.5243,
+      "step": 158208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7401257105944e-05,
+      "loss": 4.5079,
+      "step": 158720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.739287115843348e-05,
+      "loss": 4.5009,
+      "step": 159232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7384485210922966e-05,
+      "loss": 4.5046,
+      "step": 159744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7376099263412446e-05,
+      "loss": 4.508,
+      "step": 160256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7367729694705656e-05,
+      "loss": 4.5042,
+      "step": 160768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7359343747195135e-05,
+      "loss": 4.4989,
+      "step": 161280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7350957799684615e-05,
+      "loss": 4.4989,
+      "step": 161792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7342571852174095e-05,
+      "loss": 4.4999,
+      "step": 162304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.733418590466357e-05,
+      "loss": 4.5027,
+      "step": 162816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7325816335956784e-05,
+      "loss": 4.5034,
+      "step": 163328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7317430388446264e-05,
+      "loss": 4.5055,
+      "step": 163840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.730904444093574e-05,
+      "loss": 4.5033,
+      "step": 164352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.730065849342522e-05,
+      "loss": 4.4952,
+      "step": 164864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7292288924718433e-05,
+      "loss": 4.5005,
+      "step": 165376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.728390297720791e-05,
+      "loss": 4.4946,
+      "step": 165888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.727551702969739e-05,
+      "loss": 4.4982,
+      "step": 166400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.726713108218687e-05,
+      "loss": 4.4837,
+      "step": 166912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.725876151348009e-05,
+      "loss": 4.4946,
+      "step": 167424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.725037556596957e-05,
+      "loss": 4.4793,
+      "step": 167936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.724198961845904e-05,
+      "loss": 4.4858,
+      "step": 168448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.723360367094852e-05,
+      "loss": 4.4955,
+      "step": 168960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7225217723438e-05,
+      "loss": 4.4939,
+      "step": 169472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.721684815473121e-05,
+      "loss": 4.4818,
+      "step": 169984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.720846220722069e-05,
+      "loss": 4.476,
+      "step": 170496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.720007625971017e-05,
+      "loss": 4.498,
+      "step": 171008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.719169031219965e-05,
+      "loss": 4.4778,
+      "step": 171520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.718330436468914e-05,
+      "loss": 4.4861,
+      "step": 172032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.717491841717862e-05,
+      "loss": 4.4812,
+      "step": 172544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.71665324696681e-05,
+      "loss": 4.4669,
+      "step": 173056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.715814652215758e-05,
+      "loss": 4.4645,
+      "step": 173568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.714977695345079e-05,
+      "loss": 4.4655,
+      "step": 174080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.714139100594027e-05,
+      "loss": 4.4806,
+      "step": 174592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7133021437233476e-05,
+      "loss": 4.4716,
+      "step": 175104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7124635489722956e-05,
+      "loss": 4.4708,
+      "step": 175616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7116249542212436e-05,
+      "loss": 4.4651,
+      "step": 176128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7107863594701916e-05,
+      "loss": 4.4686,
+      "step": 176640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7099477647191396e-05,
+      "loss": 4.4834,
+      "step": 177152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7091091699680876e-05,
+      "loss": 4.4632,
+      "step": 177664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.708272213097409e-05,
+      "loss": 4.4633,
+      "step": 178176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.707433618346357e-05,
+      "loss": 4.459,
+      "step": 178688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.706595023595305e-05,
+      "loss": 4.4801,
+      "step": 179200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.705756428844253e-05,
+      "loss": 4.4513,
+      "step": 179712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.704917834093201e-05,
+      "loss": 4.4739,
+      "step": 180224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.704079239342149e-05,
+      "loss": 4.4501,
+      "step": 180736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.70324228247147e-05,
+      "loss": 4.4635,
+      "step": 181248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.702403687720418e-05,
+      "loss": 4.4644,
+      "step": 181760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.701565092969366e-05,
+      "loss": 4.453,
+      "step": 182272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.700726498218314e-05,
+      "loss": 4.4572,
+      "step": 182784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.699887903467262e-05,
+      "loss": 4.4511,
+      "step": 183296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.69904930871621e-05,
+      "loss": 4.4369,
+      "step": 183808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6982107139651573e-05,
+      "loss": 4.4557,
+      "step": 184320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.697373757094479e-05,
+      "loss": 4.4539,
+      "step": 184832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6965351623434276e-05,
+      "loss": 4.4444,
+      "step": 185344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6956965675923756e-05,
+      "loss": 4.4472,
+      "step": 185856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.694857972841323e-05,
+      "loss": 4.4476,
+      "step": 186368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.694019378090271e-05,
+      "loss": 4.429,
+      "step": 186880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.693180783339219e-05,
+      "loss": 4.4563,
+      "step": 187392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.692342188588167e-05,
+      "loss": 4.437,
+      "step": 187904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.691503593837115e-05,
+      "loss": 4.441,
+      "step": 188416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.690666636966436e-05,
+      "loss": 4.4512,
+      "step": 188928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.689828042215384e-05,
+      "loss": 4.4379,
+      "step": 189440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.688989447464332e-05,
+      "loss": 4.4489,
+      "step": 189952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.68815085271328e-05,
+      "loss": 4.4582,
+      "step": 190464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.687312257962228e-05,
+      "loss": 4.4445,
+      "step": 190976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.686473663211176e-05,
+      "loss": 4.4521,
+      "step": 191488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6856367063404974e-05,
+      "loss": 4.4343,
+      "step": 192000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6847981115894454e-05,
+      "loss": 4.4262,
+      "step": 192512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6839595168383934e-05,
+      "loss": 4.4387,
+      "step": 193024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6831209220873414e-05,
+      "loss": 4.4461,
+      "step": 193536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6822823273362894e-05,
+      "loss": 4.437,
+      "step": 194048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6814437325852373e-05,
+      "loss": 4.4288,
+      "step": 194560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6806051378341853e-05,
+      "loss": 4.4202,
+      "step": 195072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.679768180963506e-05,
+      "loss": 4.4244,
+      "step": 195584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.678929586212454e-05,
+      "loss": 4.4274,
+      "step": 196096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.678090991461402e-05,
+      "loss": 4.431,
+      "step": 196608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.67725239671035e-05,
+      "loss": 4.4403,
+      "step": 197120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.676415439839671e-05,
+      "loss": 4.4375,
+      "step": 197632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.675576845088619e-05,
+      "loss": 4.4276,
+      "step": 198144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.674738250337568e-05,
+      "loss": 4.423,
+      "step": 198656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.673899655586516e-05,
+      "loss": 4.4165,
+      "step": 199168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.673061060835464e-05,
+      "loss": 4.4228,
+      "step": 199680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.672222466084412e-05,
+      "loss": 4.4209,
+      "step": 200192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.671383871333359e-05,
+      "loss": 4.4166,
+      "step": 200704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.670545276582307e-05,
+      "loss": 4.4163,
+      "step": 201216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.669706681831255e-05,
+      "loss": 4.426,
+      "step": 201728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.668868087080203e-05,
+      "loss": 4.424,
+      "step": 202240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.668031130209524e-05,
+      "loss": 4.4243,
+      "step": 202752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.667192535458472e-05,
+      "loss": 4.4239,
+      "step": 203264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.66635394070742e-05,
+      "loss": 4.4267,
+      "step": 203776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.665515345956368e-05,
+      "loss": 4.4198,
+      "step": 204288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6646783890856896e-05,
+      "loss": 4.4136,
+      "step": 204800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6638397943346376e-05,
+      "loss": 4.4292,
+      "step": 205312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.663002837463959e-05,
+      "loss": 4.4085,
+      "step": 205824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6621642427129065e-05,
+      "loss": 4.4311,
+      "step": 206336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6613256479618545e-05,
+      "loss": 4.4025,
+      "step": 206848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6604870532108025e-05,
+      "loss": 4.3963,
+      "step": 207360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6596484584597505e-05,
+      "loss": 4.4052,
+      "step": 207872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6588098637086985e-05,
+      "loss": 4.4194,
+      "step": 208384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6579712689576465e-05,
+      "loss": 4.4091,
+      "step": 208896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6571326742065945e-05,
+      "loss": 4.3951,
+      "step": 209408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6562940794555425e-05,
+      "loss": 4.3962,
+      "step": 209920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6554554847044905e-05,
+      "loss": 4.4031,
+      "step": 210432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6546168899534385e-05,
+      "loss": 4.418,
+      "step": 210944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6537782952023865e-05,
+      "loss": 4.4033,
+      "step": 211456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.652942976212081e-05,
+      "loss": 4.4126,
+      "step": 211968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.652104381461029e-05,
+      "loss": 4.4087,
+      "step": 212480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.651265786709977e-05,
+      "loss": 4.406,
+      "step": 212992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.650427191958925e-05,
+      "loss": 4.4093,
+      "step": 213504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.649588597207873e-05,
+      "loss": 4.3976,
+      "step": 214016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.648750002456821e-05,
+      "loss": 4.3992,
+      "step": 214528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.647911407705769e-05,
+      "loss": 4.4079,
+      "step": 215040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.647072812954717e-05,
+      "loss": 4.3909,
+      "step": 215552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.646234218203665e-05,
+      "loss": 4.3972,
+      "step": 216064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.645397261332986e-05,
+      "loss": 4.401,
+      "step": 216576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.644558666581934e-05,
+      "loss": 4.4045,
+      "step": 217088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.643720071830882e-05,
+      "loss": 4.3891,
+      "step": 217600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6428831149602034e-05,
+      "loss": 4.3887,
+      "step": 218112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6420445202091514e-05,
+      "loss": 4.4135,
+      "step": 218624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6412059254580994e-05,
+      "loss": 4.3912,
+      "step": 219136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6403673307070474e-05,
+      "loss": 4.4041,
+      "step": 219648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.639528735955995e-05,
+      "loss": 4.3949,
+      "step": 220160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.638690141204943e-05,
+      "loss": 4.4027,
+      "step": 220672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.637851546453891e-05,
+      "loss": 4.3912,
+      "step": 221184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.637012951702839e-05,
+      "loss": 4.3886,
+      "step": 221696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6361759948321596e-05,
+      "loss": 4.3994,
+      "step": 222208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6353374000811076e-05,
+      "loss": 4.3889,
+      "step": 222720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6344988053300556e-05,
+      "loss": 4.4018,
+      "step": 223232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.633661848459377e-05,
+      "loss": 4.399,
+      "step": 223744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.632823253708325e-05,
+      "loss": 4.3967,
+      "step": 224256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.631984658957273e-05,
+      "loss": 4.3914,
+      "step": 224768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.631146064206221e-05,
+      "loss": 4.3834,
+      "step": 225280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.630307469455169e-05,
+      "loss": 4.3895,
+      "step": 225792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.629468874704117e-05,
+      "loss": 4.3791,
+      "step": 226304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.628630279953065e-05,
+      "loss": 4.3838,
+      "step": 226816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.627791685202013e-05,
+      "loss": 4.3844,
+      "step": 227328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.626953090450961e-05,
+      "loss": 4.3776,
+      "step": 227840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.626114495699909e-05,
+      "loss": 4.3796,
+      "step": 228352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.62527753882923e-05,
+      "loss": 4.3799,
+      "step": 228864
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.354189872741699,
+      "eval_runtime": 546.5547,
+      "eval_samples_per_second": 698.175,
+      "eval_steps_per_second": 21.818,
+      "step": 228957
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.624438944078178e-05,
+      "loss": 4.3894,
+      "step": 229376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.623600349327126e-05,
+      "loss": 4.3938,
+      "step": 229888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.622761754576074e-05,
+      "loss": 4.3809,
+      "step": 230400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.621923159825022e-05,
+      "loss": 4.3769,
+      "step": 230912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.62108456507397e-05,
+      "loss": 4.3806,
+      "step": 231424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.620245970322919e-05,
+      "loss": 4.362,
+      "step": 231936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.619407375571867e-05,
+      "loss": 4.3777,
+      "step": 232448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.618568780820814e-05,
+      "loss": 4.379,
+      "step": 232960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.617730186069762e-05,
+      "loss": 4.3667,
+      "step": 233472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6168932291990836e-05,
+      "loss": 4.3811,
+      "step": 233984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.616054634448031e-05,
+      "loss": 4.3879,
+      "step": 234496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.615216039696979e-05,
+      "loss": 4.3686,
+      "step": 235008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.614377444945927e-05,
+      "loss": 4.3683,
+      "step": 235520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.613538850194875e-05,
+      "loss": 4.3707,
+      "step": 236032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.612700255443823e-05,
+      "loss": 4.3748,
+      "step": 236544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.611861660692771e-05,
+      "loss": 4.3688,
+      "step": 237056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.611023065941719e-05,
+      "loss": 4.3635,
+      "step": 237568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.610184471190667e-05,
+      "loss": 4.3654,
+      "step": 238080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6093475143199885e-05,
+      "loss": 4.3667,
+      "step": 238592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6085089195689365e-05,
+      "loss": 4.3738,
+      "step": 239104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6076703248178845e-05,
+      "loss": 4.3706,
+      "step": 239616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6068317300668325e-05,
+      "loss": 4.3753,
+      "step": 240128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6059931353157805e-05,
+      "loss": 4.3721,
+      "step": 240640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6051545405647285e-05,
+      "loss": 4.3666,
+      "step": 241152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6043159458136764e-05,
+      "loss": 4.3652,
+      "step": 241664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6034773510626244e-05,
+      "loss": 4.3653,
+      "step": 242176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6026387563115724e-05,
+      "loss": 4.3701,
+      "step": 242688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.60180016156052e-05,
+      "loss": 4.3575,
+      "step": 243200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6009632046898413e-05,
+      "loss": 4.3661,
+      "step": 243712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6001246099387893e-05,
+      "loss": 4.3523,
+      "step": 244224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.599286015187737e-05,
+      "loss": 4.3565,
+      "step": 244736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.598449058317059e-05,
+      "loss": 4.3703,
+      "step": 245248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.597610463566007e-05,
+      "loss": 4.3659,
+      "step": 245760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.596771868814955e-05,
+      "loss": 4.3581,
+      "step": 246272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.595933274063903e-05,
+      "loss": 4.352,
+      "step": 246784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.59509467931285e-05,
+      "loss": 4.3663,
+      "step": 247296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.594256084561798e-05,
+      "loss": 4.3538,
+      "step": 247808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.593417489810746e-05,
+      "loss": 4.3646,
+      "step": 248320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.592578895059694e-05,
+      "loss": 4.3548,
+      "step": 248832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.591740300308642e-05,
+      "loss": 4.3442,
+      "step": 249344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.59090170555759e-05,
+      "loss": 4.3437,
+      "step": 249856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.590064748686911e-05,
+      "loss": 4.3455,
+      "step": 250368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.589226153935859e-05,
+      "loss": 4.3555,
+      "step": 250880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.588387559184808e-05,
+      "loss": 4.3468,
+      "step": 251392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.587548964433756e-05,
+      "loss": 4.3488,
+      "step": 251904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.586712007563077e-05,
+      "loss": 4.3452,
+      "step": 252416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.585873412812025e-05,
+      "loss": 4.3494,
+      "step": 252928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.585034818060973e-05,
+      "loss": 4.3605,
+      "step": 253440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.584196223309921e-05,
+      "loss": 4.342,
+      "step": 253952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.583357628558869e-05,
+      "loss": 4.3423,
+      "step": 254464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.582519033807817e-05,
+      "loss": 4.3463,
+      "step": 254976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.581680439056765e-05,
+      "loss": 4.3567,
+      "step": 255488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5808418443057127e-05,
+      "loss": 4.3344,
+      "step": 256000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5800048874350336e-05,
+      "loss": 4.3516,
+      "step": 256512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5791662926839816e-05,
+      "loss": 4.3329,
+      "step": 257024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5783276979329296e-05,
+      "loss": 4.3415,
+      "step": 257536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5774891031818776e-05,
+      "loss": 4.3468,
+      "step": 258048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.576653784191572e-05,
+      "loss": 4.3334,
+      "step": 258560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.57581518944052e-05,
+      "loss": 4.347,
+      "step": 259072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.574976594689468e-05,
+      "loss": 4.3389,
+      "step": 259584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.574137999938416e-05,
+      "loss": 4.3219,
+      "step": 260096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.573299405187364e-05,
+      "loss": 4.3363,
+      "step": 260608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.572462448316685e-05,
+      "loss": 4.337,
+      "step": 261120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.571623853565633e-05,
+      "loss": 4.3356,
+      "step": 261632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.570785258814581e-05,
+      "loss": 4.3327,
+      "step": 262144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.569946664063529e-05,
+      "loss": 4.3332,
+      "step": 262656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.569108069312477e-05,
+      "loss": 4.3157,
+      "step": 263168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.568269474561425e-05,
+      "loss": 4.3427,
+      "step": 263680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5674325176907465e-05,
+      "loss": 4.3236,
+      "step": 264192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5665939229396945e-05,
+      "loss": 4.3279,
+      "step": 264704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5657569660690154e-05,
+      "loss": 4.3391,
+      "step": 265216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5649183713179634e-05,
+      "loss": 4.3266,
+      "step": 265728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5640797765669114e-05,
+      "loss": 4.3361,
+      "step": 266240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5632411818158594e-05,
+      "loss": 4.344,
+      "step": 266752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5624025870648074e-05,
+      "loss": 4.3349,
+      "step": 267264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.561565630194128e-05,
+      "loss": 4.3427,
+      "step": 267776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.560727035443076e-05,
+      "loss": 4.3235,
+      "step": 268288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.559888440692024e-05,
+      "loss": 4.3186,
+      "step": 268800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.559049845940972e-05,
+      "loss": 4.3268,
+      "step": 269312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.55821125118992e-05,
+      "loss": 4.3385,
+      "step": 269824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.557372656438868e-05,
+      "loss": 4.3271,
+      "step": 270336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.556534061687816e-05,
+      "loss": 4.3198,
+      "step": 270848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.555695466936764e-05,
+      "loss": 4.3105,
+      "step": 271360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.554858510066086e-05,
+      "loss": 4.3204,
+      "step": 271872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.554019915315033e-05,
+      "loss": 4.3183,
+      "step": 272384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.553181320563981e-05,
+      "loss": 4.3221,
+      "step": 272896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.552342725812929e-05,
+      "loss": 4.3288,
+      "step": 273408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.551504131061877e-05,
+      "loss": 4.334,
+      "step": 273920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.550665536310825e-05,
+      "loss": 4.3216,
+      "step": 274432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.549826941559773e-05,
+      "loss": 4.3195,
+      "step": 274944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.548989984689094e-05,
+      "loss": 4.309,
+      "step": 275456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.548151389938042e-05,
+      "loss": 4.3138,
+      "step": 275968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.54731279518699e-05,
+      "loss": 4.3189,
+      "step": 276480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.546474200435939e-05,
+      "loss": 4.3126,
+      "step": 276992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.545635605684887e-05,
+      "loss": 4.3085,
+      "step": 277504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.544797010933835e-05,
+      "loss": 4.3228,
+      "step": 278016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.543958416182783e-05,
+      "loss": 4.3159,
+      "step": 278528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.543119821431731e-05,
+      "loss": 4.3262,
+      "step": 279040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5422828645610516e-05,
+      "loss": 4.3164,
+      "step": 279552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5414442698099996e-05,
+      "loss": 4.3242,
+      "step": 280064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5406073129393205e-05,
+      "loss": 4.3154,
+      "step": 280576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5397687181882685e-05,
+      "loss": 4.31,
+      "step": 281088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5389301234372165e-05,
+      "loss": 4.328,
+      "step": 281600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5380915286861645e-05,
+      "loss": 4.3068,
+      "step": 282112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5372529339351125e-05,
+      "loss": 4.3303,
+      "step": 282624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5364143391840605e-05,
+      "loss": 4.3016,
+      "step": 283136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5355757444330085e-05,
+      "loss": 4.2976,
+      "step": 283648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.534737149681957e-05,
+      "loss": 4.3033,
+      "step": 284160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.533900192811278e-05,
+      "loss": 4.3188,
+      "step": 284672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.533063235940599e-05,
+      "loss": 4.3097,
+      "step": 285184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.532224641189547e-05,
+      "loss": 4.2946,
+      "step": 285696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.531386046438495e-05,
+      "loss": 4.2941,
+      "step": 286208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.530547451687443e-05,
+      "loss": 4.3027,
+      "step": 286720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.529710494816764e-05,
+      "loss": 4.3206,
+      "step": 287232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.528871900065712e-05,
+      "loss": 4.3046,
+      "step": 287744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.52803330531466e-05,
+      "loss": 4.3176,
+      "step": 288256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.527194710563608e-05,
+      "loss": 4.3086,
+      "step": 288768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.526356115812556e-05,
+      "loss": 4.3094,
+      "step": 289280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.525517521061504e-05,
+      "loss": 4.3105,
+      "step": 289792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.524678926310452e-05,
+      "loss": 4.3001,
+      "step": 290304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5238403315594e-05,
+      "loss": 4.3013,
+      "step": 290816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5230033746887215e-05,
+      "loss": 4.3081,
+      "step": 291328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5221647799376694e-05,
+      "loss": 4.2952,
+      "step": 291840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5213278230669904e-05,
+      "loss": 4.3016,
+      "step": 292352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5204892283159384e-05,
+      "loss": 4.3066,
+      "step": 292864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.519652271445259e-05,
+      "loss": 4.3043,
+      "step": 293376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.518813676694207e-05,
+      "loss": 4.2946,
+      "step": 293888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.517975081943155e-05,
+      "loss": 4.2942,
+      "step": 294400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.517136487192103e-05,
+      "loss": 4.3154,
+      "step": 294912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.516297892441051e-05,
+      "loss": 4.2976,
+      "step": 295424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.515459297689999e-05,
+      "loss": 4.3084,
+      "step": 295936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.514620702938947e-05,
+      "loss": 4.3015,
+      "step": 296448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.513782108187895e-05,
+      "loss": 4.3068,
+      "step": 296960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.512943513436843e-05,
+      "loss": 4.2982,
+      "step": 297472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.512104918685791e-05,
+      "loss": 4.2981,
+      "step": 297984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.511266323934739e-05,
+      "loss": 4.3029,
+      "step": 298496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.510427729183687e-05,
+      "loss": 4.2965,
+      "step": 299008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.509590772313008e-05,
+      "loss": 4.3116,
+      "step": 299520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.508752177561956e-05,
+      "loss": 4.3008,
+      "step": 300032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.507913582810904e-05,
+      "loss": 4.3056,
+      "step": 300544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.507076625940225e-05,
+      "loss": 4.302,
+      "step": 301056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.506238031189173e-05,
+      "loss": 4.2914,
+      "step": 301568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.505399436438121e-05,
+      "loss": 4.2963,
+      "step": 302080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.50456084168707e-05,
+      "loss": 4.2877,
+      "step": 302592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.503722246936018e-05,
+      "loss": 4.293,
+      "step": 303104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.502883652184966e-05,
+      "loss": 4.2916,
+      "step": 303616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.502045057433914e-05,
+      "loss": 4.291,
+      "step": 304128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.501206462682862e-05,
+      "loss": 4.2846,
+      "step": 304640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5003695058121826e-05,
+      "loss": 4.2923,
+      "step": 305152
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.270674228668213,
+      "eval_runtime": 545.1308,
+      "eval_samples_per_second": 699.999,
+      "eval_steps_per_second": 21.875,
+      "step": 305276
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4995309110611306e-05,
+      "loss": 4.2996,
+      "step": 305664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4986939541904515e-05,
+      "loss": 4.3004,
+      "step": 306176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4978553594393995e-05,
+      "loss": 4.2917,
+      "step": 306688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4970167646883475e-05,
+      "loss": 4.2845,
+      "step": 307200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4961781699372955e-05,
+      "loss": 4.2946,
+      "step": 307712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4953395751862435e-05,
+      "loss": 4.2724,
+      "step": 308224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4945009804351915e-05,
+      "loss": 4.2906,
+      "step": 308736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4936623856841395e-05,
+      "loss": 4.2865,
+      "step": 309248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.492823790933088e-05,
+      "loss": 4.28,
+      "step": 309760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.491986834062409e-05,
+      "loss": 4.2908,
+      "step": 310272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.491148239311357e-05,
+      "loss": 4.2999,
+      "step": 310784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.490309644560305e-05,
+      "loss": 4.2791,
+      "step": 311296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.489471049809253e-05,
+      "loss": 4.2842,
+      "step": 311808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4886324550582003e-05,
+      "loss": 4.2821,
+      "step": 312320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4877938603071483e-05,
+      "loss": 4.2884,
+      "step": 312832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.486955265556096e-05,
+      "loss": 4.2777,
+      "step": 313344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.486116670805044e-05,
+      "loss": 4.2773,
+      "step": 313856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.485279713934365e-05,
+      "loss": 4.2781,
+      "step": 314368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.484441119183313e-05,
+      "loss": 4.2789,
+      "step": 314880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.483602524432262e-05,
+      "loss": 4.2904,
+      "step": 315392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.48276392968121e-05,
+      "loss": 4.2828,
+      "step": 315904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.481925334930158e-05,
+      "loss": 4.289,
+      "step": 316416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.481086740179106e-05,
+      "loss": 4.2888,
+      "step": 316928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.480249783308427e-05,
+      "loss": 4.2807,
+      "step": 317440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.479411188557375e-05,
+      "loss": 4.2772,
+      "step": 317952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.478574231686696e-05,
+      "loss": 4.2789,
+      "step": 318464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.477735636935644e-05,
+      "loss": 4.2853,
+      "step": 318976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.476897042184592e-05,
+      "loss": 4.2729,
+      "step": 319488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.47605844743354e-05,
+      "loss": 4.2781,
+      "step": 320000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.475219852682488e-05,
+      "loss": 4.2747,
+      "step": 320512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.474381257931436e-05,
+      "loss": 4.2666,
+      "step": 321024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.473542663180384e-05,
+      "loss": 4.2829,
+      "step": 321536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.472705706309705e-05,
+      "loss": 4.2869,
+      "step": 322048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.471867111558653e-05,
+      "loss": 4.2762,
+      "step": 322560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.471028516807601e-05,
+      "loss": 4.2683,
+      "step": 323072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.470189922056549e-05,
+      "loss": 4.2774,
+      "step": 323584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.46935296518587e-05,
+      "loss": 4.2767,
+      "step": 324096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.468514370434818e-05,
+      "loss": 4.2769,
+      "step": 324608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.467675775683766e-05,
+      "loss": 4.2707,
+      "step": 325120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.466837180932714e-05,
+      "loss": 4.2622,
+      "step": 325632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.465998586181662e-05,
+      "loss": 4.2614,
+      "step": 326144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.46515999143061e-05,
+      "loss": 4.2622,
+      "step": 326656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.464323034559931e-05,
+      "loss": 4.2743,
+      "step": 327168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.463484439808879e-05,
+      "loss": 4.2677,
+      "step": 327680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.462645845057827e-05,
+      "loss": 4.2671,
+      "step": 328192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.461807250306775e-05,
+      "loss": 4.2647,
+      "step": 328704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.460968655555724e-05,
+      "loss": 4.2664,
+      "step": 329216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.460130060804672e-05,
+      "loss": 4.2827,
+      "step": 329728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.459291466053619e-05,
+      "loss": 4.2628,
+      "step": 330240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4584545091829406e-05,
+      "loss": 4.2615,
+      "step": 330752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4576159144318886e-05,
+      "loss": 4.2679,
+      "step": 331264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4567773196808366e-05,
+      "loss": 4.2713,
+      "step": 331776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.455938724929784e-05,
+      "loss": 4.2589,
+      "step": 332288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.455100130178732e-05,
+      "loss": 4.2677,
+      "step": 332800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.45426153542768e-05,
+      "loss": 4.2589,
+      "step": 333312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.453422940676628e-05,
+      "loss": 4.262,
+      "step": 333824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.452585983805949e-05,
+      "loss": 4.2673,
+      "step": 334336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4517473890548975e-05,
+      "loss": 4.2523,
+      "step": 334848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4509087943038455e-05,
+      "loss": 4.2702,
+      "step": 335360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4500701995527935e-05,
+      "loss": 4.2569,
+      "step": 335872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4492316048017415e-05,
+      "loss": 4.2456,
+      "step": 336384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4483930100506895e-05,
+      "loss": 4.2535,
+      "step": 336896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4475544152996375e-05,
+      "loss": 4.2603,
+      "step": 337408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4467158205485855e-05,
+      "loss": 4.2632,
+      "step": 337920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4458788636779064e-05,
+      "loss": 4.2538,
+      "step": 338432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4450402689268544e-05,
+      "loss": 4.2588,
+      "step": 338944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4442016741758024e-05,
+      "loss": 4.2344,
+      "step": 339456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.443364717305123e-05,
+      "loss": 4.2659,
+      "step": 339968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.442526122554071e-05,
+      "loss": 4.2481,
+      "step": 340480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.441687527803019e-05,
+      "loss": 4.2494,
+      "step": 340992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.440848933051967e-05,
+      "loss": 4.2633,
+      "step": 341504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.440010338300916e-05,
+      "loss": 4.2516,
+      "step": 342016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.439171743549864e-05,
+      "loss": 4.2566,
+      "step": 342528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.438334786679185e-05,
+      "loss": 4.2659,
+      "step": 343040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.437496191928133e-05,
+      "loss": 4.2594,
+      "step": 343552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.436657597177081e-05,
+      "loss": 4.2633,
+      "step": 344064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.435819002426029e-05,
+      "loss": 4.2492,
+      "step": 344576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.434980407674977e-05,
+      "loss": 4.2418,
+      "step": 345088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.434141812923925e-05,
+      "loss": 4.2535,
+      "step": 345600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.433303218172873e-05,
+      "loss": 4.261,
+      "step": 346112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.43246462342182e-05,
+      "loss": 4.2512,
+      "step": 346624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.431626028670768e-05,
+      "loss": 4.247,
+      "step": 347136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.43078907180009e-05,
+      "loss": 4.2344,
+      "step": 347648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.429950477049038e-05,
+      "loss": 4.2475,
+      "step": 348160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.429111882297986e-05,
+      "loss": 4.2416,
+      "step": 348672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.428273287546934e-05,
+      "loss": 4.2519,
+      "step": 349184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.427434692795882e-05,
+      "loss": 4.2501,
+      "step": 349696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.426599373805576e-05,
+      "loss": 4.2618,
+      "step": 350208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.425760779054524e-05,
+      "loss": 4.2449,
+      "step": 350720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.424922184303472e-05,
+      "loss": 4.2483,
+      "step": 351232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.42408358955242e-05,
+      "loss": 4.2342,
+      "step": 351744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4232449948013675e-05,
+      "loss": 4.2417,
+      "step": 352256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4224064000503155e-05,
+      "loss": 4.2453,
+      "step": 352768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4215678052992635e-05,
+      "loss": 4.2382,
+      "step": 353280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4207292105482115e-05,
+      "loss": 4.238,
+      "step": 353792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.419893891557907e-05,
+      "loss": 4.2513,
+      "step": 354304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.419055296806855e-05,
+      "loss": 4.2391,
+      "step": 354816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.418216702055803e-05,
+      "loss": 4.2594,
+      "step": 355328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.41737810730475e-05,
+      "loss": 4.2419,
+      "step": 355840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4165411504340716e-05,
+      "loss": 4.2506,
+      "step": 356352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4157025556830196e-05,
+      "loss": 4.2414,
+      "step": 356864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4148639609319676e-05,
+      "loss": 4.2389,
+      "step": 357376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.414025366180915e-05,
+      "loss": 4.2556,
+      "step": 357888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4131884093102365e-05,
+      "loss": 4.2361,
+      "step": 358400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4123498145591845e-05,
+      "loss": 4.2576,
+      "step": 358912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.411511219808132e-05,
+      "loss": 4.2332,
+      "step": 359424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4106726250570805e-05,
+      "loss": 4.2303,
+      "step": 359936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4098340303060284e-05,
+      "loss": 4.2295,
+      "step": 360448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4089954355549764e-05,
+      "loss": 4.2463,
+      "step": 360960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.408160116564671e-05,
+      "loss": 4.2427,
+      "step": 361472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.407321521813619e-05,
+      "loss": 4.2296,
+      "step": 361984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.406482927062567e-05,
+      "loss": 4.2225,
+      "step": 362496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.405644332311514e-05,
+      "loss": 4.2337,
+      "step": 363008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.404805737560462e-05,
+      "loss": 4.2507,
+      "step": 363520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.40396714280941e-05,
+      "loss": 4.2335,
+      "step": 364032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.403130185938732e-05,
+      "loss": 4.248,
+      "step": 364544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.402291591187679e-05,
+      "loss": 4.2387,
+      "step": 365056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.401452996436627e-05,
+      "loss": 4.2415,
+      "step": 365568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.400614401685576e-05,
+      "loss": 4.2428,
+      "step": 366080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.399775806934524e-05,
+      "loss": 4.2301,
+      "step": 366592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.398938850063845e-05,
+      "loss": 4.2313,
+      "step": 367104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.398100255312793e-05,
+      "loss": 4.2374,
+      "step": 367616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.397261660561741e-05,
+      "loss": 4.229,
+      "step": 368128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.396423065810689e-05,
+      "loss": 4.2329,
+      "step": 368640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.395584471059637e-05,
+      "loss": 4.2372,
+      "step": 369152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.394745876308585e-05,
+      "loss": 4.2362,
+      "step": 369664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.393907281557533e-05,
+      "loss": 4.2272,
+      "step": 370176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.393068686806481e-05,
+      "loss": 4.2286,
+      "step": 370688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.392230092055429e-05,
+      "loss": 4.2431,
+      "step": 371200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.391391497304377e-05,
+      "loss": 4.2277,
+      "step": 371712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.390552902553325e-05,
+      "loss": 4.2414,
+      "step": 372224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.389714307802273e-05,
+      "loss": 4.2358,
+      "step": 372736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3888773509315936e-05,
+      "loss": 4.2373,
+      "step": 373248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.388040394060915e-05,
+      "loss": 4.2323,
+      "step": 373760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.387201799309863e-05,
+      "loss": 4.2289,
+      "step": 374272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.386363204558811e-05,
+      "loss": 4.2358,
+      "step": 374784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.385524609807759e-05,
+      "loss": 4.2288,
+      "step": 375296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.384686015056707e-05,
+      "loss": 4.244,
+      "step": 375808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.383847420305655e-05,
+      "loss": 4.2376,
+      "step": 376320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.383008825554603e-05,
+      "loss": 4.2386,
+      "step": 376832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3821702308035505e-05,
+      "loss": 4.2361,
+      "step": 377344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.381333273932872e-05,
+      "loss": 4.2262,
+      "step": 377856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.38049467918182e-05,
+      "loss": 4.2285,
+      "step": 378368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.379656084430768e-05,
+      "loss": 4.2224,
+      "step": 378880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.378817489679716e-05,
+      "loss": 4.2261,
+      "step": 379392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.377978894928664e-05,
+      "loss": 4.2266,
+      "step": 379904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.377140300177612e-05,
+      "loss": 4.2252,
+      "step": 380416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.37630170542656e-05,
+      "loss": 4.2148,
+      "step": 380928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.375463110675508e-05,
+      "loss": 4.2284,
+      "step": 381440
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.213342666625977,
+      "eval_runtime": 543.7311,
+      "eval_samples_per_second": 701.801,
+      "eval_steps_per_second": 21.932,
+      "step": 381595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.374626153804829e-05,
+      "loss": 4.24,
+      "step": 381952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.373787559053777e-05,
+      "loss": 4.2331,
+      "step": 382464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.372948964302725e-05,
+      "loss": 4.2285,
+      "step": 382976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.372110369551673e-05,
+      "loss": 4.2234,
+      "step": 383488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.371271774800621e-05,
+      "loss": 4.2237,
+      "step": 384000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.370433180049569e-05,
+      "loss": 4.2124,
+      "step": 384512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.369594585298517e-05,
+      "loss": 4.2258,
+      "step": 385024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.368755990547465e-05,
+      "loss": 4.2155,
+      "step": 385536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.367917395796413e-05,
+      "loss": 4.2218,
+      "step": 386048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.367078801045361e-05,
+      "loss": 4.2282,
+      "step": 386560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.366240206294309e-05,
+      "loss": 4.2312,
+      "step": 387072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3654016115432576e-05,
+      "loss": 4.2147,
+      "step": 387584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.364563016792205e-05,
+      "loss": 4.2231,
+      "step": 388096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.363724422041153e-05,
+      "loss": 4.2184,
+      "step": 388608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.362885827290101e-05,
+      "loss": 4.2233,
+      "step": 389120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3620488704194225e-05,
+      "loss": 4.215,
+      "step": 389632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.36121027566837e-05,
+      "loss": 4.2129,
+      "step": 390144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.360371680917318e-05,
+      "loss": 4.2159,
+      "step": 390656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.359533086166266e-05,
+      "loss": 4.22,
+      "step": 391168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.358694491415214e-05,
+      "loss": 4.2275,
+      "step": 391680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.357855896664162e-05,
+      "loss": 4.2192,
+      "step": 392192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.35701730191311e-05,
+      "loss": 4.2302,
+      "step": 392704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.356178707162058e-05,
+      "loss": 4.2226,
+      "step": 393216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.355341750291379e-05,
+      "loss": 4.2216,
+      "step": 393728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.354503155540327e-05,
+      "loss": 4.2149,
+      "step": 394240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.353664560789275e-05,
+      "loss": 4.2163,
+      "step": 394752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.352825966038223e-05,
+      "loss": 4.2256,
+      "step": 395264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.351989009167544e-05,
+      "loss": 4.2125,
+      "step": 395776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.351150414416492e-05,
+      "loss": 4.2184,
+      "step": 396288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.35031181966544e-05,
+      "loss": 4.2121,
+      "step": 396800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.349474862794761e-05,
+      "loss": 4.2043,
+      "step": 397312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.348636268043709e-05,
+      "loss": 4.227,
+      "step": 397824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.347797673292657e-05,
+      "loss": 4.2204,
+      "step": 398336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.346959078541605e-05,
+      "loss": 4.2202,
+      "step": 398848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.346120483790553e-05,
+      "loss": 4.2049,
+      "step": 399360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.345281889039501e-05,
+      "loss": 4.2214,
+      "step": 399872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.344444932168823e-05,
+      "loss": 4.2121,
+      "step": 400384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.343606337417771e-05,
+      "loss": 4.2191,
+      "step": 400896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.342767742666719e-05,
+      "loss": 4.212,
+      "step": 401408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.341929147915667e-05,
+      "loss": 4.202,
+      "step": 401920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3410921910449876e-05,
+      "loss": 4.202,
+      "step": 402432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3402535962939356e-05,
+      "loss": 4.1998,
+      "step": 402944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3394150015428836e-05,
+      "loss": 4.2196,
+      "step": 403456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3385764067918316e-05,
+      "loss": 4.2074,
+      "step": 403968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3377378120407796e-05,
+      "loss": 4.2034,
+      "step": 404480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3368992172897276e-05,
+      "loss": 4.2104,
+      "step": 404992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3360606225386756e-05,
+      "loss": 4.2084,
+      "step": 405504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.335222027787623e-05,
+      "loss": 4.2211,
+      "step": 406016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3343850709169445e-05,
+      "loss": 4.201,
+      "step": 406528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.333546476165893e-05,
+      "loss": 4.2051,
+      "step": 407040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.332707881414841e-05,
+      "loss": 4.2094,
+      "step": 407552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3318692866637885e-05,
+      "loss": 4.2109,
+      "step": 408064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.33103232979311e-05,
+      "loss": 4.1996,
+      "step": 408576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.330193735042058e-05,
+      "loss": 4.2071,
+      "step": 409088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.329356778171379e-05,
+      "loss": 4.2019,
+      "step": 409600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.328518183420327e-05,
+      "loss": 4.2057,
+      "step": 410112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.327679588669275e-05,
+      "loss": 4.2045,
+      "step": 410624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.326840993918223e-05,
+      "loss": 4.2012,
+      "step": 411136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.32600239916717e-05,
+      "loss": 4.2125,
+      "step": 411648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.325165442296492e-05,
+      "loss": 4.1965,
+      "step": 412160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.32432684754544e-05,
+      "loss": 4.1929,
+      "step": 412672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3234882527943885e-05,
+      "loss": 4.1955,
+      "step": 413184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.322649658043336e-05,
+      "loss": 4.1959,
+      "step": 413696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.321811063292284e-05,
+      "loss": 4.2146,
+      "step": 414208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3209741064216054e-05,
+      "loss": 4.1951,
+      "step": 414720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.320135511670553e-05,
+      "loss": 4.2024,
+      "step": 415232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.319296916919501e-05,
+      "loss": 4.1818,
+      "step": 415744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.318458322168449e-05,
+      "loss": 4.2041,
+      "step": 416256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.317619727417397e-05,
+      "loss": 4.1917,
+      "step": 416768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3167827705467176e-05,
+      "loss": 4.1921,
+      "step": 417280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3159441757956656e-05,
+      "loss": 4.2067,
+      "step": 417792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3151055810446136e-05,
+      "loss": 4.1965,
+      "step": 418304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.314266986293562e-05,
+      "loss": 4.2019,
+      "step": 418816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.31342839154251e-05,
+      "loss": 4.2084,
+      "step": 419328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.312589796791458e-05,
+      "loss": 4.2016,
+      "step": 419840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.311751202040406e-05,
+      "loss": 4.2113,
+      "step": 420352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.310912607289354e-05,
+      "loss": 4.1952,
+      "step": 420864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.310074012538302e-05,
+      "loss": 4.1859,
+      "step": 421376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.309237055667623e-05,
+      "loss": 4.1971,
+      "step": 421888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.308398460916571e-05,
+      "loss": 4.2061,
+      "step": 422400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.307559866165519e-05,
+      "loss": 4.1949,
+      "step": 422912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.306721271414467e-05,
+      "loss": 4.1935,
+      "step": 423424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.305884314543788e-05,
+      "loss": 4.1796,
+      "step": 423936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.305045719792736e-05,
+      "loss": 4.1922,
+      "step": 424448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.304207125041684e-05,
+      "loss": 4.1847,
+      "step": 424960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.303368530290632e-05,
+      "loss": 4.1969,
+      "step": 425472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3025315734199536e-05,
+      "loss": 4.198,
+      "step": 425984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3016929786689016e-05,
+      "loss": 4.2035,
+      "step": 426496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3008560217982226e-05,
+      "loss": 4.1935,
+      "step": 427008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3000174270471705e-05,
+      "loss": 4.1946,
+      "step": 427520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2991788322961185e-05,
+      "loss": 4.1812,
+      "step": 428032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2983402375450665e-05,
+      "loss": 4.1865,
+      "step": 428544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2975016427940145e-05,
+      "loss": 4.1905,
+      "step": 429056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2966646859233354e-05,
+      "loss": 4.1873,
+      "step": 429568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2958260911722834e-05,
+      "loss": 4.184,
+      "step": 430080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2949874964212314e-05,
+      "loss": 4.1984,
+      "step": 430592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2941489016701794e-05,
+      "loss": 4.1849,
+      "step": 431104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2933103069191274e-05,
+      "loss": 4.2072,
+      "step": 431616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2924717121680754e-05,
+      "loss": 4.1826,
+      "step": 432128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.291633117417024e-05,
+      "loss": 4.1992,
+      "step": 432640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2907945226659714e-05,
+      "loss": 4.1884,
+      "step": 433152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.289957565795293e-05,
+      "loss": 4.1881,
+      "step": 433664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.289118971044241e-05,
+      "loss": 4.2054,
+      "step": 434176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.288282014173562e-05,
+      "loss": 4.1816,
+      "step": 434688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.28744341942251e-05,
+      "loss": 4.2024,
+      "step": 435200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.286604824671458e-05,
+      "loss": 4.1794,
+      "step": 435712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.285766229920406e-05,
+      "loss": 4.1779,
+      "step": 436224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.284927635169354e-05,
+      "loss": 4.1783,
+      "step": 436736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.284090678298675e-05,
+      "loss": 4.1908,
+      "step": 437248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.283252083547623e-05,
+      "loss": 4.1889,
+      "step": 437760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.282413488796571e-05,
+      "loss": 4.1801,
+      "step": 438272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.281574894045519e-05,
+      "loss": 4.1682,
+      "step": 438784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.280736299294467e-05,
+      "loss": 4.1809,
+      "step": 439296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.279897704543415e-05,
+      "loss": 4.1951,
+      "step": 439808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.279059109792363e-05,
+      "loss": 4.1879,
+      "step": 440320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.278220515041311e-05,
+      "loss": 4.1925,
+      "step": 440832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.277383558170632e-05,
+      "loss": 4.1901,
+      "step": 441344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.27654496341958e-05,
+      "loss": 4.189,
+      "step": 441856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.275706368668528e-05,
+      "loss": 4.1905,
+      "step": 442368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.274867773917476e-05,
+      "loss": 4.1819,
+      "step": 442880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.27403245492717e-05,
+      "loss": 4.1783,
+      "step": 443392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.273193860176118e-05,
+      "loss": 4.1858,
+      "step": 443904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.272355265425066e-05,
+      "loss": 4.1777,
+      "step": 444416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.271516670674014e-05,
+      "loss": 4.184,
+      "step": 444928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.270678075922962e-05,
+      "loss": 4.1842,
+      "step": 445440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.26983948117191e-05,
+      "loss": 4.1842,
+      "step": 445952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.269000886420858e-05,
+      "loss": 4.1806,
+      "step": 446464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.268162291669806e-05,
+      "loss": 4.1746,
+      "step": 446976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.267323696918754e-05,
+      "loss": 4.192,
+      "step": 447488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.266486740048075e-05,
+      "loss": 4.1778,
+      "step": 448000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.265648145297023e-05,
+      "loss": 4.1893,
+      "step": 448512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.264809550545971e-05,
+      "loss": 4.1905,
+      "step": 449024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.263970955794919e-05,
+      "loss": 4.1848,
+      "step": 449536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.26313399892424e-05,
+      "loss": 4.1857,
+      "step": 450048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2622954041731886e-05,
+      "loss": 4.1797,
+      "step": 450560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2614568094221366e-05,
+      "loss": 4.1811,
+      "step": 451072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2606182146710846e-05,
+      "loss": 4.1761,
+      "step": 451584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2597828956807784e-05,
+      "loss": 4.1948,
+      "step": 452096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2589443009297264e-05,
+      "loss": 4.1917,
+      "step": 452608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2581057061786744e-05,
+      "loss": 4.1861,
+      "step": 453120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2572671114276224e-05,
+      "loss": 4.1844,
+      "step": 453632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2564285166765704e-05,
+      "loss": 4.1792,
+      "step": 454144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.255591559805891e-05,
+      "loss": 4.1836,
+      "step": 454656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.254752965054839e-05,
+      "loss": 4.1756,
+      "step": 455168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.253916008184161e-05,
+      "loss": 4.1711,
+      "step": 455680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.253077413433109e-05,
+      "loss": 4.1824,
+      "step": 456192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.252238818682057e-05,
+      "loss": 4.1765,
+      "step": 456704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.251400223931005e-05,
+      "loss": 4.1653,
+      "step": 457216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.250561629179953e-05,
+      "loss": 4.1771,
+      "step": 457728
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.172706604003906,
+      "eval_runtime": 535.8869,
+      "eval_samples_per_second": 712.074,
+      "eval_steps_per_second": 22.253,
+      "step": 457914
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.249723034428901e-05,
+      "loss": 4.1869,
+      "step": 458240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.248884439677849e-05,
+      "loss": 4.1841,
+      "step": 458752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.24804748280717e-05,
+      "loss": 4.179,
+      "step": 459264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.247208888056118e-05,
+      "loss": 4.1751,
+      "step": 459776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.246370293305066e-05,
+      "loss": 4.1767,
+      "step": 460288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.245531698554014e-05,
+      "loss": 4.1657,
+      "step": 460800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.244693103802962e-05,
+      "loss": 4.1766,
+      "step": 461312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.24385450905191e-05,
+      "loss": 4.1696,
+      "step": 461824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.243015914300858e-05,
+      "loss": 4.1725,
+      "step": 462336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.242177319549806e-05,
+      "loss": 4.1788,
+      "step": 462848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.241338724798754e-05,
+      "loss": 4.1862,
+      "step": 463360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2405001300477024e-05,
+      "loss": 4.1638,
+      "step": 463872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.239663173177023e-05,
+      "loss": 4.1799,
+      "step": 464384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.238824578425971e-05,
+      "loss": 4.1694,
+      "step": 464896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.237985983674919e-05,
+      "loss": 4.1796,
+      "step": 465408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.237147388923867e-05,
+      "loss": 4.1654,
+      "step": 465920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.236310432053188e-05,
+      "loss": 4.1643,
+      "step": 466432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.235471837302136e-05,
+      "loss": 4.1713,
+      "step": 466944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.234633242551084e-05,
+      "loss": 4.1702,
+      "step": 467456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2337946478000315e-05,
+      "loss": 4.1794,
+      "step": 467968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2329560530489795e-05,
+      "loss": 4.174,
+      "step": 468480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2321174582979275e-05,
+      "loss": 4.183,
+      "step": 468992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.231278863546876e-05,
+      "loss": 4.1775,
+      "step": 469504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.230440268795824e-05,
+      "loss": 4.173,
+      "step": 470016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.229603311925145e-05,
+      "loss": 4.1681,
+      "step": 470528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.228764717174093e-05,
+      "loss": 4.1692,
+      "step": 471040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.227926122423041e-05,
+      "loss": 4.1788,
+      "step": 471552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.227087527671989e-05,
+      "loss": 4.1684,
+      "step": 472064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.226248932920937e-05,
+      "loss": 4.1681,
+      "step": 472576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.225410338169885e-05,
+      "loss": 4.1692,
+      "step": 473088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.224571743418833e-05,
+      "loss": 4.1611,
+      "step": 473600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.223733148667781e-05,
+      "loss": 4.177,
+      "step": 474112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.222894553916729e-05,
+      "loss": 4.1753,
+      "step": 474624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.22205759704605e-05,
+      "loss": 4.1753,
+      "step": 475136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.221219002294998e-05,
+      "loss": 4.158,
+      "step": 475648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2203820454243196e-05,
+      "loss": 4.1744,
+      "step": 476160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2195434506732676e-05,
+      "loss": 4.1667,
+      "step": 476672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2187048559222156e-05,
+      "loss": 4.1723,
+      "step": 477184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2178662611711635e-05,
+      "loss": 4.1682,
+      "step": 477696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2170276664201115e-05,
+      "loss": 4.1588,
+      "step": 478208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2161890716690595e-05,
+      "loss": 4.1526,
+      "step": 478720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2153504769180075e-05,
+      "loss": 4.1525,
+      "step": 479232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2145118821669555e-05,
+      "loss": 4.1769,
+      "step": 479744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2136749252962764e-05,
+      "loss": 4.1594,
+      "step": 480256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2128363305452244e-05,
+      "loss": 4.1554,
+      "step": 480768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.2119977357941724e-05,
+      "loss": 4.1678,
+      "step": 481280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.211160778923493e-05,
+      "loss": 4.1626,
+      "step": 481792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.210322184172441e-05,
+      "loss": 4.1722,
+      "step": 482304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.209483589421389e-05,
+      "loss": 4.158,
+      "step": 482816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.208644994670338e-05,
+      "loss": 4.1628,
+      "step": 483328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.207806399919286e-05,
+      "loss": 4.1613,
+      "step": 483840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.206967805168233e-05,
+      "loss": 4.1678,
+      "step": 484352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.206130848297555e-05,
+      "loss": 4.1581,
+      "step": 484864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.205292253546503e-05,
+      "loss": 4.1565,
+      "step": 485376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.204453658795451e-05,
+      "loss": 4.1598,
+      "step": 485888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.203615064044398e-05,
+      "loss": 4.1635,
+      "step": 486400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.202776469293346e-05,
+      "loss": 4.1559,
+      "step": 486912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.201937874542294e-05,
+      "loss": 4.1561,
+      "step": 487424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.201099279791242e-05,
+      "loss": 4.1678,
+      "step": 487936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.20026068504019e-05,
+      "loss": 4.1522,
+      "step": 488448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.199423728169512e-05,
+      "loss": 4.1487,
+      "step": 488960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.19858513341846e-05,
+      "loss": 4.151,
+      "step": 489472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.197748176547781e-05,
+      "loss": 4.1537,
+      "step": 489984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.196909581796729e-05,
+      "loss": 4.1705,
+      "step": 490496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.196070987045677e-05,
+      "loss": 4.151,
+      "step": 491008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.195232392294625e-05,
+      "loss": 4.1571,
+      "step": 491520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.194393797543573e-05,
+      "loss": 4.1379,
+      "step": 492032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1935568406728936e-05,
+      "loss": 4.158,
+      "step": 492544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1927182459218416e-05,
+      "loss": 4.1494,
+      "step": 493056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1918796511707896e-05,
+      "loss": 4.1517,
+      "step": 493568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1910410564197376e-05,
+      "loss": 4.1604,
+      "step": 494080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1902024616686856e-05,
+      "loss": 4.1521,
+      "step": 494592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1893638669176336e-05,
+      "loss": 4.1585,
+      "step": 495104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1885252721665816e-05,
+      "loss": 4.1643,
+      "step": 495616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.18768667741553e-05,
+      "loss": 4.1595,
+      "step": 496128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.186849720544851e-05,
+      "loss": 4.1636,
+      "step": 496640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.186011125793799e-05,
+      "loss": 4.1586,
+      "step": 497152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.185172531042747e-05,
+      "loss": 4.1412,
+      "step": 497664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.184333936291695e-05,
+      "loss": 4.1531,
+      "step": 498176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.183495341540643e-05,
+      "loss": 4.1595,
+      "step": 498688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.182656746789591e-05,
+      "loss": 4.156,
+      "step": 499200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.181819789918912e-05,
+      "loss": 4.1526,
+      "step": 499712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.18098119516786e-05,
+      "loss": 4.1356,
+      "step": 500224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.180142600416808e-05,
+      "loss": 4.151,
+      "step": 500736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.179304005665756e-05,
+      "loss": 4.1423,
+      "step": 501248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.178465410914704e-05,
+      "loss": 4.1586,
+      "step": 501760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.177626816163652e-05,
+      "loss": 4.1532,
+      "step": 502272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1767882214126e-05,
+      "loss": 4.1613,
+      "step": 502784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1759512645419216e-05,
+      "loss": 4.1483,
+      "step": 503296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1751126697908696e-05,
+      "loss": 4.1514,
+      "step": 503808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.174274075039817e-05,
+      "loss": 4.1412,
+      "step": 504320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.173435480288765e-05,
+      "loss": 4.1511,
+      "step": 504832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.172596885537713e-05,
+      "loss": 4.1429,
+      "step": 505344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.171758290786661e-05,
+      "loss": 4.1454,
+      "step": 505856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.170921333915982e-05,
+      "loss": 4.1393,
+      "step": 506368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.17008273916493e-05,
+      "loss": 4.1531,
+      "step": 506880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1692457822942514e-05,
+      "loss": 4.1425,
+      "step": 507392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.168407187543199e-05,
+      "loss": 4.1681,
+      "step": 507904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1675685927921474e-05,
+      "loss": 4.1421,
+      "step": 508416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1667299980410954e-05,
+      "loss": 4.1596,
+      "step": 508928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1658914032900434e-05,
+      "loss": 4.1466,
+      "step": 509440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1650528085389914e-05,
+      "loss": 4.1461,
+      "step": 509952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1642142137879393e-05,
+      "loss": 4.1587,
+      "step": 510464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1633756190368873e-05,
+      "loss": 4.1464,
+      "step": 510976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1625370242858353e-05,
+      "loss": 4.1584,
+      "step": 511488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.161698429534783e-05,
+      "loss": 4.1446,
+      "step": 512000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.160859834783731e-05,
+      "loss": 4.1324,
+      "step": 512512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.160021240032679e-05,
+      "loss": 4.1379,
+      "step": 513024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.159184283162e-05,
+      "loss": 4.1501,
+      "step": 513536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.158345688410948e-05,
+      "loss": 4.1491,
+      "step": 514048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.157507093659896e-05,
+      "loss": 4.1424,
+      "step": 514560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.156668498908844e-05,
+      "loss": 4.123,
+      "step": 515072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.155829904157792e-05,
+      "loss": 4.14,
+      "step": 515584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.154992947287114e-05,
+      "loss": 4.1545,
+      "step": 516096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.154154352536062e-05,
+      "loss": 4.1472,
+      "step": 516608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.15331575778501e-05,
+      "loss": 4.1524,
+      "step": 517120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.152477163033958e-05,
+      "loss": 4.1491,
+      "step": 517632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.151638568282906e-05,
+      "loss": 4.1476,
+      "step": 518144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.150801611412227e-05,
+      "loss": 4.1488,
+      "step": 518656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1499646545415476e-05,
+      "loss": 4.1444,
+      "step": 519168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1491260597904956e-05,
+      "loss": 4.1365,
+      "step": 519680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1482874650394436e-05,
+      "loss": 4.1438,
+      "step": 520192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1474488702883916e-05,
+      "loss": 4.1394,
+      "step": 520704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1466102755373396e-05,
+      "loss": 4.1422,
+      "step": 521216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1457716807862876e-05,
+      "loss": 4.1433,
+      "step": 521728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1449330860352356e-05,
+      "loss": 4.1438,
+      "step": 522240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1440944912841836e-05,
+      "loss": 4.1388,
+      "step": 522752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1432558965331316e-05,
+      "loss": 4.1375,
+      "step": 523264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1424173017820796e-05,
+      "loss": 4.1491,
+      "step": 523776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1415803449114005e-05,
+      "loss": 4.1368,
+      "step": 524288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.140743388040722e-05,
+      "loss": 4.1512,
+      "step": 524800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.13990479328967e-05,
+      "loss": 4.1508,
+      "step": 525312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1390661985386174e-05,
+      "loss": 4.144,
+      "step": 525824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1382276037875654e-05,
+      "loss": 4.1473,
+      "step": 526336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1373890090365134e-05,
+      "loss": 4.142,
+      "step": 526848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1365504142854614e-05,
+      "loss": 4.1394,
+      "step": 527360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1357118195344094e-05,
+      "loss": 4.1365,
+      "step": 527872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.134874862663731e-05,
+      "loss": 4.1554,
+      "step": 528384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.134036267912679e-05,
+      "loss": 4.1517,
+      "step": 528896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.133197673161627e-05,
+      "loss": 4.1499,
+      "step": 529408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.132360716290948e-05,
+      "loss": 4.1454,
+      "step": 529920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.131522121539896e-05,
+      "loss": 4.1372,
+      "step": 530432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.130683526788844e-05,
+      "loss": 4.1423,
+      "step": 530944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.129844932037792e-05,
+      "loss": 4.1373,
+      "step": 531456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.12900633728674e-05,
+      "loss": 4.1303,
+      "step": 531968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.128167742535688e-05,
+      "loss": 4.147,
+      "step": 532480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.127329147784636e-05,
+      "loss": 4.136,
+      "step": 532992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.126492190913957e-05,
+      "loss": 4.1255,
+      "step": 533504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.125653596162905e-05,
+      "loss": 4.1395,
+      "step": 534016
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.141369819641113,
+      "eval_runtime": 535.6248,
+      "eval_samples_per_second": 712.422,
+      "eval_steps_per_second": 22.264,
+      "step": 534233
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1248150014118534e-05,
+      "loss": 4.1526,
+      "step": 534528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1239764066608014e-05,
+      "loss": 4.1424,
+      "step": 535040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1231378119097494e-05,
+      "loss": 4.1462,
+      "step": 535552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1222992171586974e-05,
+      "loss": 4.1357,
+      "step": 536064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1214606224076454e-05,
+      "loss": 4.1349,
+      "step": 536576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1206220276565934e-05,
+      "loss": 4.1318,
+      "step": 537088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1197834329055414e-05,
+      "loss": 4.1365,
+      "step": 537600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.118944838154489e-05,
+      "loss": 4.1313,
+      "step": 538112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.118106243403437e-05,
+      "loss": 4.1339,
+      "step": 538624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.117267648652385e-05,
+      "loss": 4.1371,
+      "step": 539136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.116429053901333e-05,
+      "loss": 4.1493,
+      "step": 539648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.115590459150281e-05,
+      "loss": 4.1297,
+      "step": 540160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.114751864399229e-05,
+      "loss": 4.141,
+      "step": 540672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.113913269648177e-05,
+      "loss": 4.1278,
+      "step": 541184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1130746748971247e-05,
+      "loss": 4.1423,
+      "step": 541696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1122360801460727e-05,
+      "loss": 4.128,
+      "step": 542208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.111397485395021e-05,
+      "loss": 4.1282,
+      "step": 542720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.110558890643969e-05,
+      "loss": 4.1317,
+      "step": 543232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.109720295892917e-05,
+      "loss": 4.1329,
+      "step": 543744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.108881701141865e-05,
+      "loss": 4.1399,
+      "step": 544256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.108044744271186e-05,
+      "loss": 4.1336,
+      "step": 544768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.107206149520134e-05,
+      "loss": 4.1477,
+      "step": 545280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.106367554769082e-05,
+      "loss": 4.1355,
+      "step": 545792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.10552896001803e-05,
+      "loss": 4.1364,
+      "step": 546304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.104692003147351e-05,
+      "loss": 4.1339,
+      "step": 546816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.103853408396299e-05,
+      "loss": 4.1322,
+      "step": 547328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.103014813645247e-05,
+      "loss": 4.1367,
+      "step": 547840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.102177856774568e-05,
+      "loss": 4.1319,
+      "step": 548352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.101339262023517e-05,
+      "loss": 4.1323,
+      "step": 548864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.100500667272465e-05,
+      "loss": 4.1292,
+      "step": 549376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.099662072521413e-05,
+      "loss": 4.1224,
+      "step": 549888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0988251156507336e-05,
+      "loss": 4.1358,
+      "step": 550400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0979865208996816e-05,
+      "loss": 4.1404,
+      "step": 550912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0971479261486296e-05,
+      "loss": 4.1343,
+      "step": 551424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0963093313975776e-05,
+      "loss": 4.1232,
+      "step": 551936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0954723745268985e-05,
+      "loss": 4.1357,
+      "step": 552448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0946354176562194e-05,
+      "loss": 4.13,
+      "step": 552960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0937968229051674e-05,
+      "loss": 4.1338,
+      "step": 553472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0929582281541154e-05,
+      "loss": 4.1294,
+      "step": 553984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0921196334030634e-05,
+      "loss": 4.1264,
+      "step": 554496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.091281038652012e-05,
+      "loss": 4.1146,
+      "step": 555008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.09044244390096e-05,
+      "loss": 4.1148,
+      "step": 555520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.089603849149908e-05,
+      "loss": 4.1359,
+      "step": 556032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0887652543988554e-05,
+      "loss": 4.1228,
+      "step": 556544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.087928297528177e-05,
+      "loss": 4.1193,
+      "step": 557056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.087089702777125e-05,
+      "loss": 4.1316,
+      "step": 557568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.086252745906446e-05,
+      "loss": 4.1217,
+      "step": 558080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.085414151155394e-05,
+      "loss": 4.1351,
+      "step": 558592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.084575556404342e-05,
+      "loss": 4.1213,
+      "step": 559104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.08373696165329e-05,
+      "loss": 4.1289,
+      "step": 559616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.082900004782611e-05,
+      "loss": 4.1226,
+      "step": 560128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.082061410031559e-05,
+      "loss": 4.1294,
+      "step": 560640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0812228152805074e-05,
+      "loss": 4.1236,
+      "step": 561152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0803842205294554e-05,
+      "loss": 4.1173,
+      "step": 561664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.079545625778403e-05,
+      "loss": 4.1262,
+      "step": 562176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.078708668907724e-05,
+      "loss": 4.1246,
+      "step": 562688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.077870074156672e-05,
+      "loss": 4.1216,
+      "step": 563200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0770314794056196e-05,
+      "loss": 4.1201,
+      "step": 563712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0761928846545676e-05,
+      "loss": 4.1324,
+      "step": 564224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0753542899035156e-05,
+      "loss": 4.1121,
+      "step": 564736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0745156951524636e-05,
+      "loss": 4.1167,
+      "step": 565248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0736771004014116e-05,
+      "loss": 4.1138,
+      "step": 565760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0728385056503596e-05,
+      "loss": 4.1206,
+      "step": 566272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0719999108993076e-05,
+      "loss": 4.1316,
+      "step": 566784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0711613161482556e-05,
+      "loss": 4.1155,
+      "step": 567296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0703227213972036e-05,
+      "loss": 4.1211,
+      "step": 567808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.069484126646152e-05,
+      "loss": 4.1044,
+      "step": 568320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.068647169775473e-05,
+      "loss": 4.1222,
+      "step": 568832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.067808575024421e-05,
+      "loss": 4.1127,
+      "step": 569344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.066969980273369e-05,
+      "loss": 4.1185,
+      "step": 569856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.066131385522317e-05,
+      "loss": 4.1232,
+      "step": 570368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.065292790771265e-05,
+      "loss": 4.1179,
+      "step": 570880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.064455833900586e-05,
+      "loss": 4.12,
+      "step": 571392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.063617239149534e-05,
+      "loss": 4.1311,
+      "step": 571904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.062778644398482e-05,
+      "loss": 4.1227,
+      "step": 572416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.061941687527803e-05,
+      "loss": 4.1274,
+      "step": 572928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.061103092776751e-05,
+      "loss": 4.122,
+      "step": 573440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.060264498025699e-05,
+      "loss": 4.1077,
+      "step": 573952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0594259032746476e-05,
+      "loss": 4.1167,
+      "step": 574464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0585873085235956e-05,
+      "loss": 4.1238,
+      "step": 574976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0577487137725436e-05,
+      "loss": 4.1225,
+      "step": 575488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.056910119021491e-05,
+      "loss": 4.1165,
+      "step": 576000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.056071524270439e-05,
+      "loss": 4.1037,
+      "step": 576512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0552345673997605e-05,
+      "loss": 4.112,
+      "step": 577024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0543976105290814e-05,
+      "loss": 4.109,
+      "step": 577536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0535590157780294e-05,
+      "loss": 4.1212,
+      "step": 578048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0527204210269774e-05,
+      "loss": 4.117,
+      "step": 578560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0518818262759254e-05,
+      "loss": 4.1241,
+      "step": 579072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0510432315248734e-05,
+      "loss": 4.1178,
+      "step": 579584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0502046367738214e-05,
+      "loss": 4.1125,
+      "step": 580096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0493660420227694e-05,
+      "loss": 4.1103,
+      "step": 580608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0485274472717174e-05,
+      "loss": 4.1147,
+      "step": 581120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.047690490401038e-05,
+      "loss": 4.1053,
+      "step": 581632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.046851895649986e-05,
+      "loss": 4.1121,
+      "step": 582144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.046013300898934e-05,
+      "loss": 4.1084,
+      "step": 582656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.045176344028256e-05,
+      "loss": 4.1173,
+      "step": 583168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.044337749277203e-05,
+      "loss": 4.1087,
+      "step": 583680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.043499154526151e-05,
+      "loss": 4.1337,
+      "step": 584192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.042660559775099e-05,
+      "loss": 4.1052,
+      "step": 584704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.041821965024047e-05,
+      "loss": 4.1222,
+      "step": 585216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.040983370272995e-05,
+      "loss": 4.1168,
+      "step": 585728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.040144775521943e-05,
+      "loss": 4.1123,
+      "step": 586240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.039307818651265e-05,
+      "loss": 4.1186,
+      "step": 586752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.038469223900213e-05,
+      "loss": 4.115,
+      "step": 587264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.037630629149161e-05,
+      "loss": 4.122,
+      "step": 587776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.036792034398109e-05,
+      "loss": 4.1102,
+      "step": 588288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.035953439647057e-05,
+      "loss": 4.1,
+      "step": 588800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.035114844896005e-05,
+      "loss": 4.105,
+      "step": 589312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.034276250144953e-05,
+      "loss": 4.1148,
+      "step": 589824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.033439293274274e-05,
+      "loss": 4.1169,
+      "step": 590336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0326023364035946e-05,
+      "loss": 4.1063,
+      "step": 590848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0317637416525426e-05,
+      "loss": 4.0869,
+      "step": 591360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0309251469014906e-05,
+      "loss": 4.108,
+      "step": 591872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0300865521504386e-05,
+      "loss": 4.1219,
+      "step": 592384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0292479573993866e-05,
+      "loss": 4.1124,
+      "step": 592896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.028411000528708e-05,
+      "loss": 4.122,
+      "step": 593408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.027572405777656e-05,
+      "loss": 4.115,
+      "step": 593920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.026733811026604e-05,
+      "loss": 4.1087,
+      "step": 594432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.025895216275552e-05,
+      "loss": 4.1153,
+      "step": 594944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.025058259404873e-05,
+      "loss": 4.1114,
+      "step": 595456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.024219664653821e-05,
+      "loss": 4.1025,
+      "step": 595968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.023381069902769e-05,
+      "loss": 4.1101,
+      "step": 596480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.022542475151717e-05,
+      "loss": 4.1073,
+      "step": 596992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.021703880400665e-05,
+      "loss": 4.1074,
+      "step": 597504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.020865285649613e-05,
+      "loss": 4.1058,
+      "step": 598016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.020026690898561e-05,
+      "loss": 4.1156,
+      "step": 598528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.019188096147509e-05,
+      "loss": 4.1063,
+      "step": 599040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0183527771572035e-05,
+      "loss": 4.1005,
+      "step": 599552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0175141824061515e-05,
+      "loss": 4.1193,
+      "step": 600064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0166755876550995e-05,
+      "loss": 4.0989,
+      "step": 600576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0158369929040475e-05,
+      "loss": 4.1167,
+      "step": 601088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.015001673913741e-05,
+      "loss": 4.1195,
+      "step": 601600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.014163079162689e-05,
+      "loss": 4.1122,
+      "step": 602112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.013324484411637e-05,
+      "loss": 4.1106,
+      "step": 602624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.012485889660585e-05,
+      "loss": 4.1077,
+      "step": 603136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.011647294909533e-05,
+      "loss": 4.1073,
+      "step": 603648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.010808700158481e-05,
+      "loss": 4.1051,
+      "step": 604160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.009970105407429e-05,
+      "loss": 4.1232,
+      "step": 604672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.009131510656377e-05,
+      "loss": 4.1156,
+      "step": 605184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.008292915905325e-05,
+      "loss": 4.118,
+      "step": 605696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.007455959034647e-05,
+      "loss": 4.1096,
+      "step": 606208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.006617364283595e-05,
+      "loss": 4.1051,
+      "step": 606720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.005778769532543e-05,
+      "loss": 4.1105,
+      "step": 607232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.004940174781491e-05,
+      "loss": 4.1011,
+      "step": 607744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.004101580030439e-05,
+      "loss": 4.1015,
+      "step": 608256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.003262985279387e-05,
+      "loss": 4.1123,
+      "step": 608768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.002424390528334e-05,
+      "loss": 4.1048,
+      "step": 609280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.001585795777282e-05,
+      "loss": 4.0919,
+      "step": 609792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000748838906604e-05,
+      "loss": 4.1119,
+      "step": 610304
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.116526126861572,
+      "eval_runtime": 539.0275,
+      "eval_samples_per_second": 707.925,
+      "eval_steps_per_second": 22.123,
+      "step": 610552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.999911882035925e-05,
+      "loss": 4.1073,
+      "step": 610816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.999073287284873e-05,
+      "loss": 4.1177,
+      "step": 611328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9982346925338207e-05,
+      "loss": 4.112,
+      "step": 611840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.997396097782769e-05,
+      "loss": 4.1022,
+      "step": 612352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9965575030317167e-05,
+      "loss": 4.1025,
+      "step": 612864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9957189082806646e-05,
+      "loss": 4.0993,
+      "step": 613376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9948803135296126e-05,
+      "loss": 4.1036,
+      "step": 613888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9940417187785606e-05,
+      "loss": 4.0977,
+      "step": 614400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9932047619078815e-05,
+      "loss": 4.0997,
+      "step": 614912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9923661671568295e-05,
+      "loss": 4.1068,
+      "step": 615424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.991529210286151e-05,
+      "loss": 4.1144,
+      "step": 615936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9906906155350984e-05,
+      "loss": 4.0997,
+      "step": 616448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9898520207840464e-05,
+      "loss": 4.1078,
+      "step": 616960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9890134260329944e-05,
+      "loss": 4.0955,
+      "step": 617472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.988174831281943e-05,
+      "loss": 4.116,
+      "step": 617984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.987336236530891e-05,
+      "loss": 4.0925,
+      "step": 618496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.986499279660212e-05,
+      "loss": 4.0968,
+      "step": 619008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.98566068490916e-05,
+      "loss": 4.1013,
+      "step": 619520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.984822090158108e-05,
+      "loss": 4.1019,
+      "step": 620032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.983983495407056e-05,
+      "loss": 4.1074,
+      "step": 620544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.983144900656004e-05,
+      "loss": 4.1014,
+      "step": 621056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.982306305904952e-05,
+      "loss": 4.1166,
+      "step": 621568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9814677111539e-05,
+      "loss": 4.1044,
+      "step": 622080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.980629116402848e-05,
+      "loss": 4.106,
+      "step": 622592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.979792159532169e-05,
+      "loss": 4.0993,
+      "step": 623104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.97895520266149e-05,
+      "loss": 4.107,
+      "step": 623616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9781166079104385e-05,
+      "loss": 4.1047,
+      "step": 624128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9772780131593865e-05,
+      "loss": 4.1005,
+      "step": 624640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9764394184083345e-05,
+      "loss": 4.1039,
+      "step": 625152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9756008236572825e-05,
+      "loss": 4.0996,
+      "step": 625664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9747622289062305e-05,
+      "loss": 4.0889,
+      "step": 626176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9739236341551785e-05,
+      "loss": 4.1064,
+      "step": 626688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9730866772844994e-05,
+      "loss": 4.1091,
+      "step": 627200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9722480825334474e-05,
+      "loss": 4.1027,
+      "step": 627712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9714094877823954e-05,
+      "loss": 4.0924,
+      "step": 628224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9705708930313434e-05,
+      "loss": 4.1068,
+      "step": 628736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.969733936160664e-05,
+      "loss": 4.0987,
+      "step": 629248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.968895341409612e-05,
+      "loss": 4.1015,
+      "step": 629760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.96805674665856e-05,
+      "loss": 4.1032,
+      "step": 630272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.967218151907508e-05,
+      "loss": 4.0924,
+      "step": 630784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.966379557156457e-05,
+      "loss": 4.0881,
+      "step": 631296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.965540962405405e-05,
+      "loss": 4.0834,
+      "step": 631808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.964702367654353e-05,
+      "loss": 4.1068,
+      "step": 632320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9638637729033e-05,
+      "loss": 4.0934,
+      "step": 632832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.963025178152248e-05,
+      "loss": 4.0881,
+      "step": 633344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.962186583401196e-05,
+      "loss": 4.1041,
+      "step": 633856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.961347988650144e-05,
+      "loss": 4.087,
+      "step": 634368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.960509393899092e-05,
+      "loss": 4.1093,
+      "step": 634880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.959672437028413e-05,
+      "loss": 4.0943,
+      "step": 635392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.958833842277361e-05,
+      "loss": 4.1005,
+      "step": 635904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.957996885406682e-05,
+      "loss": 4.0951,
+      "step": 636416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.957158290655631e-05,
+      "loss": 4.0947,
+      "step": 636928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.956319695904579e-05,
+      "loss": 4.0982,
+      "step": 637440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.955481101153527e-05,
+      "loss": 4.0824,
+      "step": 637952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9546441442828476e-05,
+      "loss": 4.0976,
+      "step": 638464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9538055495317956e-05,
+      "loss": 4.0924,
+      "step": 638976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9529669547807436e-05,
+      "loss": 4.0908,
+      "step": 639488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9521283600296916e-05,
+      "loss": 4.0929,
+      "step": 640000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9512897652786396e-05,
+      "loss": 4.1025,
+      "step": 640512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9504528084079605e-05,
+      "loss": 4.0799,
+      "step": 641024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9496142136569085e-05,
+      "loss": 4.0904,
+      "step": 641536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9487756189058565e-05,
+      "loss": 4.0868,
+      "step": 642048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9479370241548045e-05,
+      "loss": 4.0912,
+      "step": 642560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9470984294037525e-05,
+      "loss": 4.0985,
+      "step": 643072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9462598346527005e-05,
+      "loss": 4.088,
+      "step": 643584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9454212399016485e-05,
+      "loss": 4.0973,
+      "step": 644096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.944582645150597e-05,
+      "loss": 4.0693,
+      "step": 644608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.943747326160291e-05,
+      "loss": 4.0973,
+      "step": 645120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.942908731409239e-05,
+      "loss": 4.0829,
+      "step": 645632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.942070136658187e-05,
+      "loss": 4.0865,
+      "step": 646144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.941231541907135e-05,
+      "loss": 4.0922,
+      "step": 646656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.940392947156083e-05,
+      "loss": 4.093,
+      "step": 647168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.939555990285404e-05,
+      "loss": 4.09,
+      "step": 647680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.938717395534352e-05,
+      "loss": 4.1049,
+      "step": 648192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9378788007833e-05,
+      "loss": 4.0911,
+      "step": 648704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.937040206032248e-05,
+      "loss": 4.0995,
+      "step": 649216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.936201611281196e-05,
+      "loss": 4.0956,
+      "step": 649728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9353646544105174e-05,
+      "loss": 4.0818,
+      "step": 650240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9345260596594654e-05,
+      "loss": 4.0835,
+      "step": 650752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9336874649084134e-05,
+      "loss": 4.0937,
+      "step": 651264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9328488701573614e-05,
+      "loss": 4.0973,
+      "step": 651776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9320102754063094e-05,
+      "loss": 4.0932,
+      "step": 652288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9311716806552574e-05,
+      "loss": 4.071,
+      "step": 652800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9303330859042054e-05,
+      "loss": 4.0817,
+      "step": 653312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9294944911531534e-05,
+      "loss": 4.0829,
+      "step": 653824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.928657534282474e-05,
+      "loss": 4.0899,
+      "step": 654336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.927818939531422e-05,
+      "loss": 4.0915,
+      "step": 654848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.92698034478037e-05,
+      "loss": 4.0941,
+      "step": 655360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.926141750029318e-05,
+      "loss": 4.0878,
+      "step": 655872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.925303155278266e-05,
+      "loss": 4.0863,
+      "step": 656384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.924464560527214e-05,
+      "loss": 4.0819,
+      "step": 656896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.923625965776162e-05,
+      "loss": 4.0844,
+      "step": 657408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.92278737102511e-05,
+      "loss": 4.074,
+      "step": 657920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.921950414154431e-05,
+      "loss": 4.0906,
+      "step": 658432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.921111819403379e-05,
+      "loss": 4.082,
+      "step": 658944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.920273224652327e-05,
+      "loss": 4.0864,
+      "step": 659456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.919436267781648e-05,
+      "loss": 4.0809,
+      "step": 659968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.918597673030596e-05,
+      "loss": 4.103,
+      "step": 660480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.917759078279544e-05,
+      "loss": 4.0792,
+      "step": 660992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.916920483528492e-05,
+      "loss": 4.0945,
+      "step": 661504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.91608188877744e-05,
+      "loss": 4.0908,
+      "step": 662016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.915243294026388e-05,
+      "loss": 4.0862,
+      "step": 662528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9144063371557096e-05,
+      "loss": 4.0878,
+      "step": 663040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.913569380285031e-05,
+      "loss": 4.0903,
+      "step": 663552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9127307855339786e-05,
+      "loss": 4.0933,
+      "step": 664064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9118921907829266e-05,
+      "loss": 4.081,
+      "step": 664576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9110535960318745e-05,
+      "loss": 4.0746,
+      "step": 665088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9102150012808225e-05,
+      "loss": 4.0779,
+      "step": 665600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9093764065297705e-05,
+      "loss": 4.0862,
+      "step": 666112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9085378117787185e-05,
+      "loss": 4.0857,
+      "step": 666624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9076992170276665e-05,
+      "loss": 4.0841,
+      "step": 667136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9068606222766145e-05,
+      "loss": 4.0595,
+      "step": 667648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9060220275255625e-05,
+      "loss": 4.0836,
+      "step": 668160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9051834327745105e-05,
+      "loss": 4.089,
+      "step": 668672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9043448380234585e-05,
+      "loss": 4.0857,
+      "step": 669184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.90350788115278e-05,
+      "loss": 4.0932,
+      "step": 669696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.902670924282101e-05,
+      "loss": 4.0886,
+      "step": 670208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.901832329531049e-05,
+      "loss": 4.0783,
+      "step": 670720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.900993734779997e-05,
+      "loss": 4.0901,
+      "step": 671232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.900155140028945e-05,
+      "loss": 4.0875,
+      "step": 671744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.899316545277893e-05,
+      "loss": 4.0742,
+      "step": 672256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.898477950526841e-05,
+      "loss": 4.0842,
+      "step": 672768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.897639355775789e-05,
+      "loss": 4.0796,
+      "step": 673280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.896800761024737e-05,
+      "loss": 4.0793,
+      "step": 673792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.895963804154058e-05,
+      "loss": 4.0798,
+      "step": 674304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.895125209403006e-05,
+      "loss": 4.0849,
+      "step": 674816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.894286614651954e-05,
+      "loss": 4.0854,
+      "step": 675328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.893448019900902e-05,
+      "loss": 4.0719,
+      "step": 675840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.89260942514985e-05,
+      "loss": 4.0884,
+      "step": 676352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8917724682791715e-05,
+      "loss": 4.075,
+      "step": 676864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8909338735281195e-05,
+      "loss": 4.0891,
+      "step": 677376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8900969166574404e-05,
+      "loss": 4.0967,
+      "step": 677888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8892583219063884e-05,
+      "loss": 4.0819,
+      "step": 678400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8884197271553364e-05,
+      "loss": 4.0853,
+      "step": 678912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8875811324042843e-05,
+      "loss": 4.0782,
+      "step": 679424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.886742537653232e-05,
+      "loss": 4.0819,
+      "step": 679936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8859039429021797e-05,
+      "loss": 4.0787,
+      "step": 680448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8850653481511277e-05,
+      "loss": 4.094,
+      "step": 680960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8842267534000757e-05,
+      "loss": 4.0946,
+      "step": 681472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8833881586490236e-05,
+      "loss": 4.0857,
+      "step": 681984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.882551201778345e-05,
+      "loss": 4.0825,
+      "step": 682496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.881712607027293e-05,
+      "loss": 4.0842,
+      "step": 683008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.880874012276241e-05,
+      "loss": 4.0808,
+      "step": 683520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.880035417525189e-05,
+      "loss": 4.0781,
+      "step": 684032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.87919846065451e-05,
+      "loss": 4.0741,
+      "step": 684544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.878359865903458e-05,
+      "loss": 4.0847,
+      "step": 685056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.877521271152406e-05,
+      "loss": 4.0792,
+      "step": 685568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.876682676401354e-05,
+      "loss": 4.0689,
+      "step": 686080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.875845719530675e-05,
+      "loss": 4.0787,
+      "step": 686592
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.097525119781494,
+      "eval_runtime": 536.3073,
+      "eval_samples_per_second": 711.516,
+      "eval_steps_per_second": 22.235,
+      "step": 686871
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8750087626599966e-05,
+      "loss": 4.0873,
+      "step": 687104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.874170167908944e-05,
+      "loss": 4.0887,
+      "step": 687616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8733315731578926e-05,
+      "loss": 4.0877,
+      "step": 688128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8724929784068406e-05,
+      "loss": 4.0718,
+      "step": 688640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8716543836557886e-05,
+      "loss": 4.0823,
+      "step": 689152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8708157889047366e-05,
+      "loss": 4.0729,
+      "step": 689664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8699771941536846e-05,
+      "loss": 4.0748,
+      "step": 690176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8691385994026326e-05,
+      "loss": 4.0741,
+      "step": 690688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8683000046515806e-05,
+      "loss": 4.0713,
+      "step": 691200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8674614099005286e-05,
+      "loss": 4.0843,
+      "step": 691712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8666244530298495e-05,
+      "loss": 4.0847,
+      "step": 692224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8657858582787975e-05,
+      "loss": 4.0726,
+      "step": 692736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8649472635277455e-05,
+      "loss": 4.0854,
+      "step": 693248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8641086687766935e-05,
+      "loss": 4.0689,
+      "step": 693760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8632700740256415e-05,
+      "loss": 4.0833,
+      "step": 694272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8624314792745895e-05,
+      "loss": 4.0671,
+      "step": 694784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8615928845235375e-05,
+      "loss": 4.0725,
+      "step": 695296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8607542897724855e-05,
+      "loss": 4.0773,
+      "step": 695808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8599156950214334e-05,
+      "loss": 4.0742,
+      "step": 696320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8590771002703814e-05,
+      "loss": 4.0846,
+      "step": 696832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8582385055193294e-05,
+      "loss": 4.0696,
+      "step": 697344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8573999107682774e-05,
+      "loss": 4.0948,
+      "step": 697856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8565613160172254e-05,
+      "loss": 4.074,
+      "step": 698368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8557227212661734e-05,
+      "loss": 4.0794,
+      "step": 698880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8548841265151214e-05,
+      "loss": 4.0762,
+      "step": 699392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.854047169644442e-05,
+      "loss": 4.0813,
+      "step": 699904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.85320857489339e-05,
+      "loss": 4.0782,
+      "step": 700416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.852369980142338e-05,
+      "loss": 4.0714,
+      "step": 700928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.851531385391286e-05,
+      "loss": 4.0775,
+      "step": 701440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.850692790640234e-05,
+      "loss": 4.0764,
+      "step": 701952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.849855833769556e-05,
+      "loss": 4.0583,
+      "step": 702464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.849017239018504e-05,
+      "loss": 4.0821,
+      "step": 702976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.848178644267452e-05,
+      "loss": 4.0844,
+      "step": 703488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8473400495164e-05,
+      "loss": 4.0787,
+      "step": 704000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846501454765348e-05,
+      "loss": 4.067,
+      "step": 704512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.845664497894669e-05,
+      "loss": 4.0818,
+      "step": 705024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.84482754102399e-05,
+      "loss": 4.0754,
+      "step": 705536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.843988946272938e-05,
+      "loss": 4.0747,
+      "step": 706048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.843150351521886e-05,
+      "loss": 4.0756,
+      "step": 706560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.842311756770834e-05,
+      "loss": 4.0651,
+      "step": 707072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.841473162019782e-05,
+      "loss": 4.0651,
+      "step": 707584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.84063456726873e-05,
+      "loss": 4.0577,
+      "step": 708096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.839795972517678e-05,
+      "loss": 4.0804,
+      "step": 708608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.838957377766626e-05,
+      "loss": 4.069,
+      "step": 709120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8381187830155743e-05,
+      "loss": 4.0627,
+      "step": 709632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8372801882645223e-05,
+      "loss": 4.0759,
+      "step": 710144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.836443231393843e-05,
+      "loss": 4.0622,
+      "step": 710656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.835604636642791e-05,
+      "loss": 4.0797,
+      "step": 711168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.834766041891739e-05,
+      "loss": 4.0736,
+      "step": 711680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8339274471406866e-05,
+      "loss": 4.0725,
+      "step": 712192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8330888523896346e-05,
+      "loss": 4.0711,
+      "step": 712704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8322502576385826e-05,
+      "loss": 4.0652,
+      "step": 713216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8314116628875305e-05,
+      "loss": 4.0744,
+      "step": 713728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8305747060168515e-05,
+      "loss": 4.0559,
+      "step": 714240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8297361112657995e-05,
+      "loss": 4.0747,
+      "step": 714752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.828897516514748e-05,
+      "loss": 4.0639,
+      "step": 715264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.828058921763696e-05,
+      "loss": 4.0681,
+      "step": 715776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.827220327012644e-05,
+      "loss": 4.0638,
+      "step": 716288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8263850080223386e-05,
+      "loss": 4.0773,
+      "step": 716800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8255464132712866e-05,
+      "loss": 4.0616,
+      "step": 717312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.824707818520234e-05,
+      "loss": 4.0594,
+      "step": 717824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.823869223769182e-05,
+      "loss": 4.0657,
+      "step": 718336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.82303062901813e-05,
+      "loss": 4.06,
+      "step": 718848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8221936721474515e-05,
+      "loss": 4.0727,
+      "step": 719360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.821355077396399e-05,
+      "loss": 4.0695,
+      "step": 719872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.820516482645347e-05,
+      "loss": 4.0707,
+      "step": 720384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.819677887894295e-05,
+      "loss": 4.046,
+      "step": 720896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8188392931432435e-05,
+      "loss": 4.0711,
+      "step": 721408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8180006983921915e-05,
+      "loss": 4.0609,
+      "step": 721920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8171621036411395e-05,
+      "loss": 4.0612,
+      "step": 722432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8163235088900875e-05,
+      "loss": 4.0641,
+      "step": 722944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8154865520194084e-05,
+      "loss": 4.0667,
+      "step": 723456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8146479572683564e-05,
+      "loss": 4.064,
+      "step": 723968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8138093625173044e-05,
+      "loss": 4.082,
+      "step": 724480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8129707677662524e-05,
+      "loss": 4.0669,
+      "step": 724992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.812133810895573e-05,
+      "loss": 4.0731,
+      "step": 725504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.811295216144521e-05,
+      "loss": 4.0679,
+      "step": 726016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.810456621393469e-05,
+      "loss": 4.0589,
+      "step": 726528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.809618026642417e-05,
+      "loss": 4.059,
+      "step": 727040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.808781069771739e-05,
+      "loss": 4.0664,
+      "step": 727552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.807942475020687e-05,
+      "loss": 4.07,
+      "step": 728064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.807103880269635e-05,
+      "loss": 4.0744,
+      "step": 728576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.806265285518583e-05,
+      "loss": 4.0451,
+      "step": 729088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.805426690767531e-05,
+      "loss": 4.0538,
+      "step": 729600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.804588096016479e-05,
+      "loss": 4.0586,
+      "step": 730112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8037527770261727e-05,
+      "loss": 4.0665,
+      "step": 730624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8029141822751207e-05,
+      "loss": 4.0641,
+      "step": 731136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8020755875240687e-05,
+      "loss": 4.0733,
+      "step": 731648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8012369927730166e-05,
+      "loss": 4.0632,
+      "step": 732160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8003983980219646e-05,
+      "loss": 4.0622,
+      "step": 732672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7995598032709126e-05,
+      "loss": 4.0586,
+      "step": 733184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7987212085198606e-05,
+      "loss": 4.0606,
+      "step": 733696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7978826137688086e-05,
+      "loss": 4.0494,
+      "step": 734208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.79704565689813e-05,
+      "loss": 4.0629,
+      "step": 734720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.796207062147078e-05,
+      "loss": 4.0618,
+      "step": 735232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.795368467396026e-05,
+      "loss": 4.0597,
+      "step": 735744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.794529872644974e-05,
+      "loss": 4.0589,
+      "step": 736256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.793691277893922e-05,
+      "loss": 4.0735,
+      "step": 736768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.792854321023243e-05,
+      "loss": 4.0574,
+      "step": 737280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.792015726272191e-05,
+      "loss": 4.0697,
+      "step": 737792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.791177131521139e-05,
+      "loss": 4.0654,
+      "step": 738304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.790338536770087e-05,
+      "loss": 4.0653,
+      "step": 738816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.789501579899408e-05,
+      "loss": 4.0591,
+      "step": 739328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7886646230287296e-05,
+      "loss": 4.0666,
+      "step": 739840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7878260282776776e-05,
+      "loss": 4.0691,
+      "step": 740352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7869874335266256e-05,
+      "loss": 4.0557,
+      "step": 740864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7861488387755736e-05,
+      "loss": 4.0534,
+      "step": 741376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7853102440245216e-05,
+      "loss": 4.0514,
+      "step": 741888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7844716492734696e-05,
+      "loss": 4.0591,
+      "step": 742400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7836346924027905e-05,
+      "loss": 4.0627,
+      "step": 742912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7827960976517385e-05,
+      "loss": 4.0597,
+      "step": 743424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7819575029006865e-05,
+      "loss": 4.0414,
+      "step": 743936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7811189081496345e-05,
+      "loss": 4.0543,
+      "step": 744448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7802803133985825e-05,
+      "loss": 4.0621,
+      "step": 744960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.77944171864753e-05,
+      "loss": 4.0651,
+      "step": 745472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.778603123896478e-05,
+      "loss": 4.0722,
+      "step": 745984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7777645291454264e-05,
+      "loss": 4.0629,
+      "step": 746496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7769275722747474e-05,
+      "loss": 4.0566,
+      "step": 747008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7760889775236954e-05,
+      "loss": 4.0619,
+      "step": 747520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7752503827726433e-05,
+      "loss": 4.066,
+      "step": 748032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.774413425901965e-05,
+      "loss": 4.0533,
+      "step": 748544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.773576469031286e-05,
+      "loss": 4.0547,
+      "step": 749056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.772737874280234e-05,
+      "loss": 4.0603,
+      "step": 749568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.771899279529182e-05,
+      "loss": 4.0493,
+      "step": 750080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.77106068477813e-05,
+      "loss": 4.0612,
+      "step": 750592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.770222090027077e-05,
+      "loss": 4.058,
+      "step": 751104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.769383495276025e-05,
+      "loss": 4.0627,
+      "step": 751616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.768544900524973e-05,
+      "loss": 4.0487,
+      "step": 752128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.767706305773921e-05,
+      "loss": 4.0623,
+      "step": 752640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.76686771102287e-05,
+      "loss": 4.0518,
+      "step": 753152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.766029116271818e-05,
+      "loss": 4.0668,
+      "step": 753664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.765190521520766e-05,
+      "loss": 4.0715,
+      "step": 754176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.764351926769714e-05,
+      "loss": 4.0582,
+      "step": 754688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.763514969899035e-05,
+      "loss": 4.0602,
+      "step": 755200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.762676375147983e-05,
+      "loss": 4.0608,
+      "step": 755712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7618394182773036e-05,
+      "loss": 4.0579,
+      "step": 756224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7610008235262516e-05,
+      "loss": 4.056,
+      "step": 756736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7601622287751996e-05,
+      "loss": 4.0696,
+      "step": 757248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7593236340241476e-05,
+      "loss": 4.0656,
+      "step": 757760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7584850392730956e-05,
+      "loss": 4.0654,
+      "step": 758272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7576464445220436e-05,
+      "loss": 4.0606,
+      "step": 758784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.756809487651365e-05,
+      "loss": 4.0654,
+      "step": 759296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.755970892900313e-05,
+      "loss": 4.0534,
+      "step": 759808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.755132298149261e-05,
+      "loss": 4.053,
+      "step": 760320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.754293703398209e-05,
+      "loss": 4.0523,
+      "step": 760832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.753455108647157e-05,
+      "loss": 4.062,
+      "step": 761344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.752616513896105e-05,
+      "loss": 4.056,
+      "step": 761856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.751777919145053e-05,
+      "loss": 4.0411,
+      "step": 762368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.750940962274374e-05,
+      "loss": 4.0592,
+      "step": 762880
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.081372261047363,
+      "eval_runtime": 537.7049,
+      "eval_samples_per_second": 709.666,
+      "eval_steps_per_second": 22.178,
+      "step": 763190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.750102367523322e-05,
+      "loss": 4.0626,
+      "step": 763392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.74926377277227e-05,
+      "loss": 4.0633,
+      "step": 763904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.748425178021218e-05,
+      "loss": 4.0663,
+      "step": 764416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7475865832701654e-05,
+      "loss": 4.0463,
+      "step": 764928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7467479885191134e-05,
+      "loss": 4.0626,
+      "step": 765440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.745911031648435e-05,
+      "loss": 4.0527,
+      "step": 765952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7450724368973836e-05,
+      "loss": 4.0513,
+      "step": 766464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.744233842146331e-05,
+      "loss": 4.0482,
+      "step": 766976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.743395247395279e-05,
+      "loss": 4.0506,
+      "step": 767488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.742556652644227e-05,
+      "loss": 4.0612,
+      "step": 768000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7417196957735485e-05,
+      "loss": 4.0593,
+      "step": 768512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.740881101022496e-05,
+      "loss": 4.0508,
+      "step": 769024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.740042506271444e-05,
+      "loss": 4.0612,
+      "step": 769536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.739203911520392e-05,
+      "loss": 4.0451,
+      "step": 770048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.73836531676934e-05,
+      "loss": 4.0628,
+      "step": 770560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.737526722018288e-05,
+      "loss": 4.044,
+      "step": 771072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.736689765147609e-05,
+      "loss": 4.0484,
+      "step": 771584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7358511703965574e-05,
+      "loss": 4.0551,
+      "step": 772096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7350125756455054e-05,
+      "loss": 4.0539,
+      "step": 772608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7341739808944534e-05,
+      "loss": 4.0586,
+      "step": 773120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7333353861434014e-05,
+      "loss": 4.0453,
+      "step": 773632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7324967913923494e-05,
+      "loss": 4.0726,
+      "step": 774144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7316581966412974e-05,
+      "loss": 4.0533,
+      "step": 774656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.730821239770618e-05,
+      "loss": 4.0587,
+      "step": 775168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.729982645019566e-05,
+      "loss": 4.0511,
+      "step": 775680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.729144050268514e-05,
+      "loss": 4.0615,
+      "step": 776192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.728305455517462e-05,
+      "loss": 4.0546,
+      "step": 776704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.72746686076641e-05,
+      "loss": 4.0488,
+      "step": 777216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.726628266015358e-05,
+      "loss": 4.0557,
+      "step": 777728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.725789671264306e-05,
+      "loss": 4.0537,
+      "step": 778240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.724951076513254e-05,
+      "loss": 4.0376,
+      "step": 778752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.724114119642576e-05,
+      "loss": 4.0572,
+      "step": 779264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.723277162771897e-05,
+      "loss": 4.0599,
+      "step": 779776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.722438568020845e-05,
+      "loss": 4.0553,
+      "step": 780288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.721599973269793e-05,
+      "loss": 4.0446,
+      "step": 780800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.720761378518741e-05,
+      "loss": 4.0635,
+      "step": 781312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7199244216480616e-05,
+      "loss": 4.0501,
+      "step": 781824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7190858268970096e-05,
+      "loss": 4.0574,
+      "step": 782336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7182472321459576e-05,
+      "loss": 4.0527,
+      "step": 782848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7174086373949056e-05,
+      "loss": 4.0433,
+      "step": 783360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7165700426438536e-05,
+      "loss": 4.0465,
+      "step": 783872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7157314478928016e-05,
+      "loss": 4.0352,
+      "step": 784384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7148928531417496e-05,
+      "loss": 4.0553,
+      "step": 784896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7140542583906976e-05,
+      "loss": 4.0463,
+      "step": 785408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7132156636396456e-05,
+      "loss": 4.0408,
+      "step": 785920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7123770688885936e-05,
+      "loss": 4.0531,
+      "step": 786432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7115401120179145e-05,
+      "loss": 4.0407,
+      "step": 786944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7107015172668625e-05,
+      "loss": 4.0567,
+      "step": 787456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7098629225158105e-05,
+      "loss": 4.0506,
+      "step": 787968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7090243277647585e-05,
+      "loss": 4.0473,
+      "step": 788480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7081857330137065e-05,
+      "loss": 4.0491,
+      "step": 788992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7073471382626545e-05,
+      "loss": 4.0449,
+      "step": 789504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7065085435116025e-05,
+      "loss": 4.0499,
+      "step": 790016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7056699487605505e-05,
+      "loss": 4.0358,
+      "step": 790528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7048329918898714e-05,
+      "loss": 4.0501,
+      "step": 791040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.703996035019193e-05,
+      "loss": 4.0451,
+      "step": 791552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.703157440268141e-05,
+      "loss": 4.0452,
+      "step": 792064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.702318845517089e-05,
+      "loss": 4.0428,
+      "step": 792576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.701480250766037e-05,
+      "loss": 4.0527,
+      "step": 793088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.700643293895358e-05,
+      "loss": 4.0409,
+      "step": 793600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.699804699144306e-05,
+      "loss": 4.0382,
+      "step": 794112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.698966104393254e-05,
+      "loss": 4.0447,
+      "step": 794624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.698127509642202e-05,
+      "loss": 4.0373,
+      "step": 795136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.69728891489115e-05,
+      "loss": 4.049,
+      "step": 795648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.696450320140098e-05,
+      "loss": 4.0475,
+      "step": 796160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.695611725389046e-05,
+      "loss": 4.0473,
+      "step": 796672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.694773130637994e-05,
+      "loss": 4.0257,
+      "step": 797184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.693934535886942e-05,
+      "loss": 4.0502,
+      "step": 797696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.6930992168966363e-05,
+      "loss": 4.0385,
+      "step": 798208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.6922606221455843e-05,
+      "loss": 4.0394,
+      "step": 798720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.691422027394532e-05,
+      "loss": 4.0395,
+      "step": 799232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.69058343264348e-05,
+      "loss": 4.0486,
+      "step": 799744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.689744837892428e-05,
+      "loss": 4.0396,
+      "step": 800256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.688906243141376e-05,
+      "loss": 4.0603,
+      "step": 800768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.688069286270697e-05,
+      "loss": 4.0473,
+      "step": 801280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.687230691519645e-05,
+      "loss": 4.0476,
+      "step": 801792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.686392096768593e-05,
+      "loss": 4.0472,
+      "step": 802304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.685553502017541e-05,
+      "loss": 4.0365,
+      "step": 802816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.684714907266489e-05,
+      "loss": 4.0383,
+      "step": 803328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.683876312515437e-05,
+      "loss": 4.0466,
+      "step": 803840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.683039355644758e-05,
+      "loss": 4.0463,
+      "step": 804352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.682200760893707e-05,
+      "loss": 4.0518,
+      "step": 804864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.681363804023028e-05,
+      "loss": 4.0248,
+      "step": 805376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.680525209271976e-05,
+      "loss": 4.0324,
+      "step": 805888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.679686614520924e-05,
+      "loss": 4.0347,
+      "step": 806400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.678848019769872e-05,
+      "loss": 4.0475,
+      "step": 806912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.67800942501882e-05,
+      "loss": 4.0402,
+      "step": 807424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.677170830267768e-05,
+      "loss": 4.0501,
+      "step": 807936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.676332235516715e-05,
+      "loss": 4.0438,
+      "step": 808448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.675493640765663e-05,
+      "loss": 4.0431,
+      "step": 808960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.674655046014611e-05,
+      "loss": 4.0331,
+      "step": 809472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.673816451263559e-05,
+      "loss": 4.041,
+      "step": 809984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.672977856512507e-05,
+      "loss": 4.031,
+      "step": 810496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.672139261761455e-05,
+      "loss": 4.0394,
+      "step": 811008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6713006670104037e-05,
+      "loss": 4.0382,
+      "step": 811520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6704637101397246e-05,
+      "loss": 4.04,
+      "step": 812032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6696251153886726e-05,
+      "loss": 4.0402,
+      "step": 812544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6687865206376206e-05,
+      "loss": 4.0489,
+      "step": 813056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6679479258865685e-05,
+      "loss": 4.042,
+      "step": 813568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6671093311355165e-05,
+      "loss": 4.0473,
+      "step": 814080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6662723742648375e-05,
+      "loss": 4.0405,
+      "step": 814592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6654354173941584e-05,
+      "loss": 4.0445,
+      "step": 815104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6645968226431064e-05,
+      "loss": 4.0402,
+      "step": 815616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6637582278920544e-05,
+      "loss": 4.0476,
+      "step": 816128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6629196331410023e-05,
+      "loss": 4.0455,
+      "step": 816640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6620810383899503e-05,
+      "loss": 4.0332,
+      "step": 817152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6612424436388983e-05,
+      "loss": 4.0322,
+      "step": 817664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.660403848887847e-05,
+      "loss": 4.0282,
+      "step": 818176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.659566892017168e-05,
+      "loss": 4.0412,
+      "step": 818688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.658728297266116e-05,
+      "loss": 4.041,
+      "step": 819200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.657889702515064e-05,
+      "loss": 4.0445,
+      "step": 819712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.657051107764012e-05,
+      "loss": 4.0185,
+      "step": 820224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.65621251301296e-05,
+      "loss": 4.0268,
+      "step": 820736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.655373918261908e-05,
+      "loss": 4.0436,
+      "step": 821248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.654535323510856e-05,
+      "loss": 4.0445,
+      "step": 821760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.653696728759804e-05,
+      "loss": 4.0517,
+      "step": 822272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.652859771889125e-05,
+      "loss": 4.0454,
+      "step": 822784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.652021177138073e-05,
+      "loss": 4.031,
+      "step": 823296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.651182582387021e-05,
+      "loss": 4.0421,
+      "step": 823808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.650343987635969e-05,
+      "loss": 4.0413,
+      "step": 824320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6495070307652904e-05,
+      "loss": 4.0325,
+      "step": 824832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6486684360142384e-05,
+      "loss": 4.0318,
+      "step": 825344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.647831479143559e-05,
+      "loss": 4.0386,
+      "step": 825856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.646992884392507e-05,
+      "loss": 4.0337,
+      "step": 826368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.646154289641455e-05,
+      "loss": 4.0379,
+      "step": 826880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.645315694890403e-05,
+      "loss": 4.0343,
+      "step": 827392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.644477100139351e-05,
+      "loss": 4.0452,
+      "step": 827904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.643640143268672e-05,
+      "loss": 4.0298,
+      "step": 828416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.64280154851762e-05,
+      "loss": 4.037,
+      "step": 828928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.641962953766568e-05,
+      "loss": 4.0367,
+      "step": 829440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.641124359015516e-05,
+      "loss": 4.0485,
+      "step": 829952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.640285764264464e-05,
+      "loss": 4.0473,
+      "step": 830464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.639447169513412e-05,
+      "loss": 4.0361,
+      "step": 830976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.63860857476236e-05,
+      "loss": 4.0392,
+      "step": 831488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.637769980011308e-05,
+      "loss": 4.0414,
+      "step": 832000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.636933023140629e-05,
+      "loss": 4.0366,
+      "step": 832512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.636094428389577e-05,
+      "loss": 4.034,
+      "step": 833024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.635255833638525e-05,
+      "loss": 4.048,
+      "step": 833536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.634417238887473e-05,
+      "loss": 4.0446,
+      "step": 834048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.633580282016794e-05,
+      "loss": 4.0446,
+      "step": 834560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.632741687265742e-05,
+      "loss": 4.0395,
+      "step": 835072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.63190309251469e-05,
+      "loss": 4.0432,
+      "step": 835584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.631064497763638e-05,
+      "loss": 4.0297,
+      "step": 836096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.630225903012586e-05,
+      "loss": 4.033,
+      "step": 836608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6293873082615346e-05,
+      "loss": 4.0339,
+      "step": 837120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6285487135104826e-05,
+      "loss": 4.0393,
+      "step": 837632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6277101187594306e-05,
+      "loss": 4.0349,
+      "step": 838144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6268731618887515e-05,
+      "loss": 4.023,
+      "step": 838656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6260362050180724e-05,
+      "loss": 4.0379,
+      "step": 839168
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.067577362060547,
+      "eval_runtime": 550.3917,
+      "eval_samples_per_second": 693.308,
+      "eval_steps_per_second": 21.666,
+      "step": 839509
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6251976102670204e-05,
+      "loss": 4.0535,
+      "step": 839680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6243590155159684e-05,
+      "loss": 4.0438,
+      "step": 840192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6235204207649164e-05,
+      "loss": 4.0424,
+      "step": 840704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6226818260138644e-05,
+      "loss": 4.0295,
+      "step": 841216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6218432312628124e-05,
+      "loss": 4.0398,
+      "step": 841728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6210046365117604e-05,
+      "loss": 4.0339,
+      "step": 842240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6201660417607084e-05,
+      "loss": 4.0278,
+      "step": 842752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6193274470096564e-05,
+      "loss": 4.0294,
+      "step": 843264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6184888522586044e-05,
+      "loss": 4.0349,
+      "step": 843776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.617650257507553e-05,
+      "loss": 4.0372,
+      "step": 844288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6168116627565004e-05,
+      "loss": 4.0375,
+      "step": 844800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6159730680054484e-05,
+      "loss": 4.0323,
+      "step": 845312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6151344732543964e-05,
+      "loss": 4.0389,
+      "step": 845824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6142958785033444e-05,
+      "loss": 4.0312,
+      "step": 846336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6134572837522923e-05,
+      "loss": 4.0379,
+      "step": 846848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6126186890012403e-05,
+      "loss": 4.0266,
+      "step": 847360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.611780094250188e-05,
+      "loss": 4.0269,
+      "step": 847872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.610941499499136e-05,
+      "loss": 4.0361,
+      "step": 848384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.610104542628457e-05,
+      "loss": 4.0308,
+      "step": 848896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.609265947877405e-05,
+      "loss": 4.0386,
+      "step": 849408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.608427353126353e-05,
+      "loss": 4.0284,
+      "step": 849920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.607588758375301e-05,
+      "loss": 4.0514,
+      "step": 850432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.606750163624249e-05,
+      "loss": 4.0329,
+      "step": 850944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.605911568873198e-05,
+      "loss": 4.0377,
+      "step": 851456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.605072974122146e-05,
+      "loss": 4.0331,
+      "step": 851968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.604236017251467e-05,
+      "loss": 4.0415,
+      "step": 852480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.603399060380788e-05,
+      "loss": 4.0335,
+      "step": 852992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.602560465629736e-05,
+      "loss": 4.0271,
+      "step": 853504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.601721870878684e-05,
+      "loss": 4.0358,
+      "step": 854016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.600883276127632e-05,
+      "loss": 4.0357,
+      "step": 854528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.60004468137658e-05,
+      "loss": 4.0197,
+      "step": 855040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.599206086625528e-05,
+      "loss": 4.0323,
+      "step": 855552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5983691297548486e-05,
+      "loss": 4.0433,
+      "step": 856064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5975305350037966e-05,
+      "loss": 4.0399,
+      "step": 856576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5966919402527446e-05,
+      "loss": 4.0224,
+      "step": 857088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.595853345501693e-05,
+      "loss": 4.038,
+      "step": 857600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.595016388631014e-05,
+      "loss": 4.0336,
+      "step": 858112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.594177793879962e-05,
+      "loss": 4.034,
+      "step": 858624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.59333919912891e-05,
+      "loss": 4.0304,
+      "step": 859136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.592500604377858e-05,
+      "loss": 4.0247,
+      "step": 859648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.591662009626806e-05,
+      "loss": 4.0272,
+      "step": 860160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5908234148757535e-05,
+      "loss": 4.014,
+      "step": 860672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5899848201247015e-05,
+      "loss": 4.0324,
+      "step": 861184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5891462253736495e-05,
+      "loss": 4.0297,
+      "step": 861696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5883076306225975e-05,
+      "loss": 4.0234,
+      "step": 862208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5874690358715455e-05,
+      "loss": 4.0311,
+      "step": 862720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.586632079000867e-05,
+      "loss": 4.0211,
+      "step": 863232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.585793484249815e-05,
+      "loss": 4.0413,
+      "step": 863744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.584954889498763e-05,
+      "loss": 4.0322,
+      "step": 864256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.584116294747711e-05,
+      "loss": 4.0203,
+      "step": 864768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.583279337877032e-05,
+      "loss": 4.0271,
+      "step": 865280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58244074312598e-05,
+      "loss": 4.0264,
+      "step": 865792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.581602148374928e-05,
+      "loss": 4.0362,
+      "step": 866304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.580763553623876e-05,
+      "loss": 4.0123,
+      "step": 866816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.579924958872824e-05,
+      "loss": 4.0358,
+      "step": 867328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.579088002002145e-05,
+      "loss": 4.0248,
+      "step": 867840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.578249407251093e-05,
+      "loss": 4.0263,
+      "step": 868352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.577410812500041e-05,
+      "loss": 4.0263,
+      "step": 868864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.576572217748989e-05,
+      "loss": 4.035,
+      "step": 869376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.575733622997937e-05,
+      "loss": 4.0198,
+      "step": 869888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5748950282468855e-05,
+      "loss": 4.0205,
+      "step": 870400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5740564334958335e-05,
+      "loss": 4.024,
+      "step": 870912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5732178387447815e-05,
+      "loss": 4.0181,
+      "step": 871424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5723808818741024e-05,
+      "loss": 4.0302,
+      "step": 871936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5715422871230504e-05,
+      "loss": 4.0278,
+      "step": 872448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.570705330252371e-05,
+      "loss": 4.0336,
+      "step": 872960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.569866735501319e-05,
+      "loss": 4.0029,
+      "step": 873472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.569028140750267e-05,
+      "loss": 4.0281,
+      "step": 873984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.568191183879588e-05,
+      "loss": 4.0244,
+      "step": 874496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.567352589128536e-05,
+      "loss": 4.017,
+      "step": 875008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.566513994377484e-05,
+      "loss": 4.021,
+      "step": 875520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.565675399626432e-05,
+      "loss": 4.0304,
+      "step": 876032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.564836804875381e-05,
+      "loss": 4.0177,
+      "step": 876544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.563998210124329e-05,
+      "loss": 4.0413,
+      "step": 877056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.563159615373277e-05,
+      "loss": 4.029,
+      "step": 877568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.562321020622225e-05,
+      "loss": 4.0313,
+      "step": 878080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.561482425871172e-05,
+      "loss": 4.027,
+      "step": 878592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.560645469000494e-05,
+      "loss": 4.0197,
+      "step": 879104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5598085121298147e-05,
+      "loss": 4.0194,
+      "step": 879616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5589699173787627e-05,
+      "loss": 4.0277,
+      "step": 880128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5581313226277106e-05,
+      "loss": 4.0309,
+      "step": 880640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5572927278766586e-05,
+      "loss": 4.0279,
+      "step": 881152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5564541331256066e-05,
+      "loss": 4.01,
+      "step": 881664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5556155383745546e-05,
+      "loss": 4.0142,
+      "step": 882176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5547769436235026e-05,
+      "loss": 4.0116,
+      "step": 882688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5539383488724506e-05,
+      "loss": 4.031,
+      "step": 883200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.553101392001772e-05,
+      "loss": 4.0226,
+      "step": 883712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5522627972507195e-05,
+      "loss": 4.0319,
+      "step": 884224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.551425840380041e-05,
+      "loss": 4.0261,
+      "step": 884736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.550587245628989e-05,
+      "loss": 4.0221,
+      "step": 885248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.549748650877937e-05,
+      "loss": 4.0154,
+      "step": 885760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5489100561268844e-05,
+      "loss": 4.0194,
+      "step": 886272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5480714613758324e-05,
+      "loss": 4.0159,
+      "step": 886784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5472328666247804e-05,
+      "loss": 4.0166,
+      "step": 887296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5463942718737284e-05,
+      "loss": 4.0158,
+      "step": 887808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5455556771226764e-05,
+      "loss": 4.0255,
+      "step": 888320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5447170823716244e-05,
+      "loss": 4.0221,
+      "step": 888832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5438784876205724e-05,
+      "loss": 4.03,
+      "step": 889344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.543039892869521e-05,
+      "loss": 4.0257,
+      "step": 889856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.542201298118469e-05,
+      "loss": 4.0261,
+      "step": 890368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.541362703367417e-05,
+      "loss": 4.023,
+      "step": 890880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.540525746496738e-05,
+      "loss": 4.028,
+      "step": 891392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.539687151745686e-05,
+      "loss": 4.0207,
+      "step": 891904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.538848556994634e-05,
+      "loss": 4.0299,
+      "step": 892416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.538011600123955e-05,
+      "loss": 4.022,
+      "step": 892928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.537173005372903e-05,
+      "loss": 4.0223,
+      "step": 893440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.536334410621851e-05,
+      "loss": 4.0166,
+      "step": 893952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.535497453751172e-05,
+      "loss": 4.0045,
+      "step": 894464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.53465885900012e-05,
+      "loss": 4.0238,
+      "step": 894976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.533820264249068e-05,
+      "loss": 4.0256,
+      "step": 895488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5329816694980164e-05,
+      "loss": 4.0215,
+      "step": 896000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5321430747469644e-05,
+      "loss": 4.002,
+      "step": 896512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5313044799959124e-05,
+      "loss": 4.0073,
+      "step": 897024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5304658852448604e-05,
+      "loss": 4.0248,
+      "step": 897536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5296272904938084e-05,
+      "loss": 4.0279,
+      "step": 898048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.528788695742756e-05,
+      "loss": 4.0334,
+      "step": 898560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.527950100991704e-05,
+      "loss": 4.0268,
+      "step": 899072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.527111506240652e-05,
+      "loss": 4.0193,
+      "step": 899584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.526274549369973e-05,
+      "loss": 4.0214,
+      "step": 900096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5254359546189206e-05,
+      "loss": 4.0243,
+      "step": 900608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5245973598678686e-05,
+      "loss": 4.0145,
+      "step": 901120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5237587651168166e-05,
+      "loss": 4.0146,
+      "step": 901632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5229201703657646e-05,
+      "loss": 4.0243,
+      "step": 902144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.522083213495086e-05,
+      "loss": 4.0106,
+      "step": 902656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.521244618744034e-05,
+      "loss": 4.023,
+      "step": 903168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.520407661873356e-05,
+      "loss": 4.0158,
+      "step": 903680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.519569067122303e-05,
+      "loss": 4.0277,
+      "step": 904192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.518730472371251e-05,
+      "loss": 4.0102,
+      "step": 904704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.517891877620199e-05,
+      "loss": 4.0214,
+      "step": 905216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.517053282869147e-05,
+      "loss": 4.0161,
+      "step": 905728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.516214688118095e-05,
+      "loss": 4.0296,
+      "step": 906240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.515377731247416e-05,
+      "loss": 4.0283,
+      "step": 906752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.514539136496364e-05,
+      "loss": 4.0253,
+      "step": 907264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.513700541745312e-05,
+      "loss": 4.017,
+      "step": 907776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.51286194699426e-05,
+      "loss": 4.0227,
+      "step": 908288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.512023352243209e-05,
+      "loss": 4.0198,
+      "step": 908800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5111847574921567e-05,
+      "loss": 4.0169,
+      "step": 909312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5103478006214776e-05,
+      "loss": 4.0313,
+      "step": 909824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5095092058704256e-05,
+      "loss": 4.0257,
+      "step": 910336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5086722489997465e-05,
+      "loss": 4.0296,
+      "step": 910848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5078336542486945e-05,
+      "loss": 4.027,
+      "step": 911360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5069950594976425e-05,
+      "loss": 4.0256,
+      "step": 911872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5061564647465905e-05,
+      "loss": 4.0139,
+      "step": 912384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5053178699955385e-05,
+      "loss": 4.016,
+      "step": 912896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5044809131248594e-05,
+      "loss": 4.0171,
+      "step": 913408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5036423183738074e-05,
+      "loss": 4.0226,
+      "step": 913920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5028037236227554e-05,
+      "loss": 4.0174,
+      "step": 914432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.501965128871704e-05,
+      "loss": 4.0039,
+      "step": 914944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.501126534120652e-05,
+      "loss": 4.0211,
+      "step": 915456
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.056208610534668,
+      "eval_runtime": 551.0765,
+      "eval_samples_per_second": 692.447,
+      "eval_steps_per_second": 21.639,
+      "step": 915828
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5002879393696e-05,
+      "loss": 4.0202,
+      "step": 915968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.499449344618548e-05,
+      "loss": 4.0263,
+      "step": 916480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.498610749867496e-05,
+      "loss": 4.0241,
+      "step": 916992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.497772155116444e-05,
+      "loss": 4.0153,
+      "step": 917504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.496935198245765e-05,
+      "loss": 4.0231,
+      "step": 918016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.496096603494713e-05,
+      "loss": 4.0176,
+      "step": 918528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.495258008743661e-05,
+      "loss": 4.0104,
+      "step": 919040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.494419413992609e-05,
+      "loss": 4.0133,
+      "step": 919552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.493580819241557e-05,
+      "loss": 4.0157,
+      "step": 920064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.492742224490504e-05,
+      "loss": 4.0179,
+      "step": 920576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.491905267619826e-05,
+      "loss": 4.0215,
+      "step": 921088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.491066672868774e-05,
+      "loss": 4.0175,
+      "step": 921600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.490228078117722e-05,
+      "loss": 4.0208,
+      "step": 922112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.48938948336667e-05,
+      "loss": 4.0131,
+      "step": 922624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.488550888615618e-05,
+      "loss": 4.0193,
+      "step": 923136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.487712293864566e-05,
+      "loss": 4.0102,
+      "step": 923648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.486873699113514e-05,
+      "loss": 4.013,
+      "step": 924160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.486036742242835e-05,
+      "loss": 4.015,
+      "step": 924672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.485198147491783e-05,
+      "loss": 4.0132,
+      "step": 925184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.484359552740731e-05,
+      "loss": 4.023,
+      "step": 925696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.483520957989679e-05,
+      "loss": 4.0093,
+      "step": 926208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.482682363238627e-05,
+      "loss": 4.0311,
+      "step": 926720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4818454063679476e-05,
+      "loss": 4.0178,
+      "step": 927232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4810068116168956e-05,
+      "loss": 4.0231,
+      "step": 927744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.480168216865844e-05,
+      "loss": 4.0128,
+      "step": 928256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.479329622114792e-05,
+      "loss": 4.0247,
+      "step": 928768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.47849102736374e-05,
+      "loss": 4.0224,
+      "step": 929280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.477654070493061e-05,
+      "loss": 4.0131,
+      "step": 929792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.476815475742009e-05,
+      "loss": 4.0149,
+      "step": 930304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.475976880990957e-05,
+      "loss": 4.0214,
+      "step": 930816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.475138286239905e-05,
+      "loss": 4.0058,
+      "step": 931328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.474299691488853e-05,
+      "loss": 4.0142,
+      "step": 931840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.473461096737801e-05,
+      "loss": 4.0245,
+      "step": 932352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.472622501986749e-05,
+      "loss": 4.0227,
+      "step": 932864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.471783907235697e-05,
+      "loss": 4.0065,
+      "step": 933376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.470945312484645e-05,
+      "loss": 4.0179,
+      "step": 933888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4701099934943396e-05,
+      "loss": 4.0215,
+      "step": 934400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4692713987432876e-05,
+      "loss": 4.0202,
+      "step": 934912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4684328039922356e-05,
+      "loss": 4.0126,
+      "step": 935424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4675942092411836e-05,
+      "loss": 4.0144,
+      "step": 935936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4667556144901316e-05,
+      "loss": 4.01,
+      "step": 936448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4659186576194525e-05,
+      "loss": 3.9938,
+      "step": 936960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4650800628684005e-05,
+      "loss": 4.0206,
+      "step": 937472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4642414681173485e-05,
+      "loss": 4.0111,
+      "step": 937984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4634028733662965e-05,
+      "loss": 4.0042,
+      "step": 938496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4625642786152445e-05,
+      "loss": 4.0158,
+      "step": 939008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4617256838641925e-05,
+      "loss": 4.0054,
+      "step": 939520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.46088708911314e-05,
+      "loss": 4.024,
+      "step": 940032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.460048494362088e-05,
+      "loss": 4.0173,
+      "step": 940544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4592115374914094e-05,
+      "loss": 4.0023,
+      "step": 941056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.458374580620731e-05,
+      "loss": 4.013,
+      "step": 941568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.457535985869679e-05,
+      "loss": 4.0105,
+      "step": 942080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.456697391118627e-05,
+      "loss": 4.019,
+      "step": 942592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.455858796367575e-05,
+      "loss": 3.9957,
+      "step": 943104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.455020201616523e-05,
+      "loss": 4.0206,
+      "step": 943616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.45418160686547e-05,
+      "loss": 4.003,
+      "step": 944128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.453343012114418e-05,
+      "loss": 4.0108,
+      "step": 944640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.452504417363366e-05,
+      "loss": 4.0132,
+      "step": 945152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.451665822612314e-05,
+      "loss": 4.0173,
+      "step": 945664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.450830503622009e-05,
+      "loss": 4.007,
+      "step": 946176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.449991908870957e-05,
+      "loss": 3.9998,
+      "step": 946688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.449153314119905e-05,
+      "loss": 4.0082,
+      "step": 947200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.448314719368853e-05,
+      "loss": 4.0041,
+      "step": 947712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.447476124617801e-05,
+      "loss": 4.0121,
+      "step": 948224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.446637529866749e-05,
+      "loss": 4.0152,
+      "step": 948736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.445798935115697e-05,
+      "loss": 4.0115,
+      "step": 949248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.444960340364645e-05,
+      "loss": 3.9917,
+      "step": 949760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4441233834939656e-05,
+      "loss": 4.0092,
+      "step": 950272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4432847887429136e-05,
+      "loss": 4.0082,
+      "step": 950784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4424461939918616e-05,
+      "loss": 3.9993,
+      "step": 951296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4416092371211825e-05,
+      "loss": 4.0063,
+      "step": 951808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4407706423701305e-05,
+      "loss": 4.0157,
+      "step": 952320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4399320476190785e-05,
+      "loss": 4.0058,
+      "step": 952832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.439093452868027e-05,
+      "loss": 4.0168,
+      "step": 953344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.438256495997348e-05,
+      "loss": 4.0165,
+      "step": 953856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.437417901246296e-05,
+      "loss": 4.0177,
+      "step": 954368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.436579306495244e-05,
+      "loss": 4.0141,
+      "step": 954880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.435740711744192e-05,
+      "loss": 4.0004,
+      "step": 955392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.43490211699314e-05,
+      "loss": 4.0026,
+      "step": 955904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.434063522242088e-05,
+      "loss": 4.013,
+      "step": 956416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.433224927491036e-05,
+      "loss": 4.0156,
+      "step": 956928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.432386332739984e-05,
+      "loss": 4.0113,
+      "step": 957440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.431549375869305e-05,
+      "loss": 3.99,
+      "step": 957952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.430710781118253e-05,
+      "loss": 4.0048,
+      "step": 958464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.429872186367201e-05,
+      "loss": 3.9946,
+      "step": 958976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.429035229496522e-05,
+      "loss": 4.0134,
+      "step": 959488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4281966347454706e-05,
+      "loss": 4.0069,
+      "step": 960000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4273580399944186e-05,
+      "loss": 4.0181,
+      "step": 960512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4265194452433666e-05,
+      "loss": 4.0069,
+      "step": 961024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4256808504923146e-05,
+      "loss": 4.0072,
+      "step": 961536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.4248422557412625e-05,
+      "loss": 3.9985,
+      "step": 962048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.4240036609902105e-05,
+      "loss": 4.0051,
+      "step": 962560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.4231650662391585e-05,
+      "loss": 4.0044,
+      "step": 963072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.4223264714881065e-05,
+      "loss": 4.001,
+      "step": 963584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.421487876737054e-05,
+      "loss": 3.9993,
+      "step": 964096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.420649281986002e-05,
+      "loss": 4.0069,
+      "step": 964608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.41981068723495e-05,
+      "loss": 4.0066,
+      "step": 965120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.418973730364271e-05,
+      "loss": 4.0149,
+      "step": 965632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.418135135613219e-05,
+      "loss": 4.0092,
+      "step": 966144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.4172965408621674e-05,
+      "loss": 4.0099,
+      "step": 966656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.4164579461111154e-05,
+      "loss": 4.0088,
+      "step": 967168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.415620989240436e-05,
+      "loss": 4.0074,
+      "step": 967680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.414782394489384e-05,
+      "loss": 4.0071,
+      "step": 968192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.413943799738332e-05,
+      "loss": 4.0158,
+      "step": 968704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.41310520498728e-05,
+      "loss": 4.0054,
+      "step": 969216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.412266610236228e-05,
+      "loss": 4.0094,
+      "step": 969728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.411428015485176e-05,
+      "loss": 3.9965,
+      "step": 970240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.410589420734124e-05,
+      "loss": 3.9944,
+      "step": 970752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.409750825983072e-05,
+      "loss": 4.0046,
+      "step": 971264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.408913869112393e-05,
+      "loss": 4.0114,
+      "step": 971776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.408075274361341e-05,
+      "loss": 4.0064,
+      "step": 972288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.407236679610289e-05,
+      "loss": 3.9872,
+      "step": 972800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.406398084859237e-05,
+      "loss": 3.9884,
+      "step": 973312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.405561127988559e-05,
+      "loss": 4.014,
+      "step": 973824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.404722533237507e-05,
+      "loss": 4.0138,
+      "step": 974336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.403885576366828e-05,
+      "loss": 4.0139,
+      "step": 974848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.403046981615776e-05,
+      "loss": 4.01,
+      "step": 975360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.402208386864724e-05,
+      "loss": 4.0082,
+      "step": 975872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.401369792113672e-05,
+      "loss": 4.0102,
+      "step": 976384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.40053119736262e-05,
+      "loss": 4.0056,
+      "step": 976896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.399692602611568e-05,
+      "loss": 3.9966,
+      "step": 977408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3988540078605157e-05,
+      "loss": 3.9985,
+      "step": 977920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3980154131094637e-05,
+      "loss": 4.0103,
+      "step": 978432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3971768183584117e-05,
+      "loss": 3.9944,
+      "step": 978944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3963398614877326e-05,
+      "loss": 4.0041,
+      "step": 979456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.395501266736681e-05,
+      "loss": 4.004,
+      "step": 979968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.394662671985629e-05,
+      "loss": 4.0111,
+      "step": 980480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.393824077234577e-05,
+      "loss": 3.9956,
+      "step": 980992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.392987120363898e-05,
+      "loss": 4.0073,
+      "step": 981504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.392148525612846e-05,
+      "loss": 4.004,
+      "step": 982016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.391309930861794e-05,
+      "loss": 4.0096,
+      "step": 982528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.390471336110742e-05,
+      "loss": 4.0118,
+      "step": 983040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3896327413596894e-05,
+      "loss": 4.0119,
+      "step": 983552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.388795784489011e-05,
+      "loss": 4.0059,
+      "step": 984064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.387957189737959e-05,
+      "loss": 4.0045,
+      "step": 984576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.387118594986907e-05,
+      "loss": 4.0061,
+      "step": 985088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.386280000235855e-05,
+      "loss": 3.9969,
+      "step": 985600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.385443043365176e-05,
+      "loss": 4.0176,
+      "step": 986112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3846060864944975e-05,
+      "loss": 4.0121,
+      "step": 986624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3837674917434455e-05,
+      "loss": 4.0161,
+      "step": 987136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3829288969923935e-05,
+      "loss": 4.0081,
+      "step": 987648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3820903022413415e-05,
+      "loss": 4.0123,
+      "step": 988160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3812533453706624e-05,
+      "loss": 3.9963,
+      "step": 988672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3804147506196104e-05,
+      "loss": 4.0034,
+      "step": 989184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3795761558685584e-05,
+      "loss": 4.0009,
+      "step": 989696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3787375611175064e-05,
+      "loss": 4.0091,
+      "step": 990208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3778989663664544e-05,
+      "loss": 4.0041,
+      "step": 990720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.377062009495775e-05,
+      "loss": 3.9887,
+      "step": 991232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.376223414744723e-05,
+      "loss": 4.0007,
+      "step": 991744
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.047861576080322,
+      "eval_runtime": 547.9436,
+      "eval_samples_per_second": 696.406,
+      "eval_steps_per_second": 21.763,
+      "step": 992147
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.375384819993671e-05,
+      "loss": 4.0031,
+      "step": 992256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.37454622524262e-05,
+      "loss": 4.0123,
+      "step": 992768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.373707630491567e-05,
+      "loss": 4.0087,
+      "step": 993280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.372869035740515e-05,
+      "loss": 3.9931,
+      "step": 993792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.372030440989463e-05,
+      "loss": 4.0023,
+      "step": 994304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.371191846238411e-05,
+      "loss": 4.0104,
+      "step": 994816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.370353251487359e-05,
+      "loss": 4.0061,
+      "step": 995328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.369514656736307e-05,
+      "loss": 4.0018,
+      "step": 995840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.368676061985255e-05,
+      "loss": 3.9984,
+      "step": 996352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.367837467234203e-05,
+      "loss": 4.0009,
+      "step": 996864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.366998872483151e-05,
+      "loss": 4.0095,
+      "step": 997376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.366160277732099e-05,
+      "loss": 3.9952,
+      "step": 997888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.365321682981047e-05,
+      "loss": 4.0041,
+      "step": 998400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.364483088229995e-05,
+      "loss": 4.004,
+      "step": 998912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.363644493478943e-05,
+      "loss": 4.0045,
+      "step": 999424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.362805898727891e-05,
+      "loss": 3.9925,
+      "step": 999936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.361967303976839e-05,
+      "loss": 4.0108,
+      "step": 1000448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.361130347106161e-05,
+      "loss": 4.0,
+      "step": 1000960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.360291752355109e-05,
+      "loss": 4.0038,
+      "step": 1001472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.359453157604056e-05,
+      "loss": 4.0097,
+      "step": 1001984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.358614562853004e-05,
+      "loss": 4.0182,
+      "step": 1002496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.357775968101952e-05,
+      "loss": 4.0073,
+      "step": 1003008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3569373733509e-05,
+      "loss": 3.9918,
+      "step": 1003520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.356098778599848e-05,
+      "loss": 4.0075,
+      "step": 1004032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.355260183848796e-05,
+      "loss": 3.9954,
+      "step": 1004544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.354423226978117e-05,
+      "loss": 4.0003,
+      "step": 1005056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3535862701074386e-05,
+      "loss": 3.9996,
+      "step": 1005568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3527476753563866e-05,
+      "loss": 4.0055,
+      "step": 1006080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3519090806053346e-05,
+      "loss": 3.9973,
+      "step": 1006592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3510704858542826e-05,
+      "loss": 4.0142,
+      "step": 1007104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3502318911032306e-05,
+      "loss": 4.0061,
+      "step": 1007616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3493932963521786e-05,
+      "loss": 4.0053,
+      "step": 1008128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3485563394814995e-05,
+      "loss": 4.0036,
+      "step": 1008640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3477177447304475e-05,
+      "loss": 4.0139,
+      "step": 1009152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3468791499793955e-05,
+      "loss": 3.9974,
+      "step": 1009664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3460405552283435e-05,
+      "loss": 4.0105,
+      "step": 1010176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3452019604772915e-05,
+      "loss": 4.0011,
+      "step": 1010688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3443633657262395e-05,
+      "loss": 4.0066,
+      "step": 1011200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3435247709751875e-05,
+      "loss": 3.9966,
+      "step": 1011712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3426861762241354e-05,
+      "loss": 4.0135,
+      "step": 1012224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.341849219353457e-05,
+      "loss": 4.0136,
+      "step": 1012736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.341010624602405e-05,
+      "loss": 3.9909,
+      "step": 1013248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.340172029851353e-05,
+      "loss": 4.0002,
+      "step": 1013760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.339335072980674e-05,
+      "loss": 3.9945,
+      "step": 1014272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.338496478229622e-05,
+      "loss": 3.9914,
+      "step": 1014784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.33765788347857e-05,
+      "loss": 3.9831,
+      "step": 1015296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.336819288727518e-05,
+      "loss": 4.0041,
+      "step": 1015808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.335980693976466e-05,
+      "loss": 3.9974,
+      "step": 1016320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.335142099225414e-05,
+      "loss": 3.9985,
+      "step": 1016832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.334303504474362e-05,
+      "loss": 4.0034,
+      "step": 1017344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.333464909723309e-05,
+      "loss": 3.9823,
+      "step": 1017856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3326295907330044e-05,
+      "loss": 3.9986,
+      "step": 1018368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3317909959819524e-05,
+      "loss": 3.9999,
+      "step": 1018880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3309524012309004e-05,
+      "loss": 3.9807,
+      "step": 1019392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3301138064798484e-05,
+      "loss": 3.9922,
+      "step": 1019904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.329276849609169e-05,
+      "loss": 4.0013,
+      "step": 1020416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.328438254858117e-05,
+      "loss": 4.0103,
+      "step": 1020928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.327599660107065e-05,
+      "loss": 3.9843,
+      "step": 1021440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.326761065356013e-05,
+      "loss": 4.0027,
+      "step": 1021952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.325922470604961e-05,
+      "loss": 3.9986,
+      "step": 1022464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.325083875853909e-05,
+      "loss": 3.9877,
+      "step": 1022976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3242452811028566e-05,
+      "loss": 3.9977,
+      "step": 1023488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3234066863518046e-05,
+      "loss": 3.984,
+      "step": 1024000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3225680916007526e-05,
+      "loss": 3.9888,
+      "step": 1024512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.321731134730074e-05,
+      "loss": 4.0088,
+      "step": 1025024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.320892539979022e-05,
+      "loss": 4.0074,
+      "step": 1025536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.32005394522797e-05,
+      "loss": 3.9926,
+      "step": 1026048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.319215350476918e-05,
+      "loss": 3.9998,
+      "step": 1026560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.318378393606239e-05,
+      "loss": 3.9977,
+      "step": 1027072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.317539798855187e-05,
+      "loss": 3.9876,
+      "step": 1027584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.316701204104135e-05,
+      "loss": 3.997,
+      "step": 1028096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.315862609353083e-05,
+      "loss": 4.0016,
+      "step": 1028608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.315024014602031e-05,
+      "loss": 3.9934,
+      "step": 1029120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.314185419850979e-05,
+      "loss": 3.9875,
+      "step": 1029632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3133484629803e-05,
+      "loss": 3.9996,
+      "step": 1030144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.312509868229248e-05,
+      "loss": 3.9926,
+      "step": 1030656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.311671273478196e-05,
+      "loss": 3.9956,
+      "step": 1031168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3108326787271446e-05,
+      "loss": 4.0037,
+      "step": 1031680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3099940839760926e-05,
+      "loss": 3.9936,
+      "step": 1032192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3091587649857864e-05,
+      "loss": 3.9999,
+      "step": 1032704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3083201702347344e-05,
+      "loss": 3.9949,
+      "step": 1033216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3074815754836824e-05,
+      "loss": 3.9979,
+      "step": 1033728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3066429807326304e-05,
+      "loss": 3.9857,
+      "step": 1034240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3058043859815784e-05,
+      "loss": 3.9905,
+      "step": 1034752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3049657912305264e-05,
+      "loss": 3.9886,
+      "step": 1035264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3041271964794744e-05,
+      "loss": 3.9932,
+      "step": 1035776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3032886017284224e-05,
+      "loss": 3.988,
+      "step": 1036288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3024500069773704e-05,
+      "loss": 3.9923,
+      "step": 1036800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.301613050106691e-05,
+      "loss": 3.9904,
+      "step": 1037312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.300776093236013e-05,
+      "loss": 3.9953,
+      "step": 1037824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.299937498484961e-05,
+      "loss": 3.9909,
+      "step": 1038336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.299098903733909e-05,
+      "loss": 3.9989,
+      "step": 1038848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.298260308982857e-05,
+      "loss": 3.9925,
+      "step": 1039360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.297421714231805e-05,
+      "loss": 3.9932,
+      "step": 1039872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.296584757361126e-05,
+      "loss": 3.9897,
+      "step": 1040384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.295746162610074e-05,
+      "loss": 3.9906,
+      "step": 1040896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.294907567859022e-05,
+      "loss": 3.9928,
+      "step": 1041408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.29406897310797e-05,
+      "loss": 3.9912,
+      "step": 1041920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.293230378356918e-05,
+      "loss": 3.9848,
+      "step": 1042432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.292391783605866e-05,
+      "loss": 4.0055,
+      "step": 1042944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.291553188854814e-05,
+      "loss": 3.9953,
+      "step": 1043456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2907162319841354e-05,
+      "loss": 4.0045,
+      "step": 1043968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2898776372330834e-05,
+      "loss": 3.9889,
+      "step": 1044480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2890390424820314e-05,
+      "loss": 3.9981,
+      "step": 1044992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2882004477309793e-05,
+      "loss": 3.9898,
+      "step": 1045504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2873634908603e-05,
+      "loss": 3.9854,
+      "step": 1046016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.286524896109248e-05,
+      "loss": 3.9925,
+      "step": 1046528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.285687939238569e-05,
+      "loss": 3.9967,
+      "step": 1047040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.284849344487517e-05,
+      "loss": 3.9814,
+      "step": 1047552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.284010749736465e-05,
+      "loss": 3.9746,
+      "step": 1048064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.283172154985413e-05,
+      "loss": 4.0038,
+      "step": 1048576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.282333560234361e-05,
+      "loss": 3.9969,
+      "step": 1049088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.281494965483309e-05,
+      "loss": 3.9809,
+      "step": 1049600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.280656370732257e-05,
+      "loss": 3.9901,
+      "step": 1050112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.279817775981205e-05,
+      "loss": 3.9899,
+      "step": 1050624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.278979181230153e-05,
+      "loss": 3.9913,
+      "step": 1051136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.278140586479101e-05,
+      "loss": 3.9868,
+      "step": 1051648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.277303629608423e-05,
+      "loss": 3.9883,
+      "step": 1052160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.27646503485737e-05,
+      "loss": 3.993,
+      "step": 1052672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.275626440106318e-05,
+      "loss": 3.993,
+      "step": 1053184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.274787845355266e-05,
+      "loss": 3.9872,
+      "step": 1053696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2739508884845876e-05,
+      "loss": 4.0049,
+      "step": 1054208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.273112293733535e-05,
+      "loss": 3.9853,
+      "step": 1054720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.272273698982483e-05,
+      "loss": 3.9933,
+      "step": 1055232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.271435104231431e-05,
+      "loss": 3.9997,
+      "step": 1055744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.270596509480379e-05,
+      "loss": 3.9897,
+      "step": 1056256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.269757914729327e-05,
+      "loss": 3.9964,
+      "step": 1056768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2689193199782756e-05,
+      "loss": 3.9828,
+      "step": 1057280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2680807252272236e-05,
+      "loss": 4.0002,
+      "step": 1057792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2672421304761716e-05,
+      "loss": 3.994,
+      "step": 1058304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2664035357251196e-05,
+      "loss": 4.0035,
+      "step": 1058816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2655649409740676e-05,
+      "loss": 3.9894,
+      "step": 1059328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2647279841033885e-05,
+      "loss": 4.0033,
+      "step": 1059840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2638893893523365e-05,
+      "loss": 3.9888,
+      "step": 1060352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2630507946012845e-05,
+      "loss": 3.9975,
+      "step": 1060864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2622121998502325e-05,
+      "loss": 3.9875,
+      "step": 1061376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2613752429795534e-05,
+      "loss": 3.9901,
+      "step": 1061888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.260538286108874e-05,
+      "loss": 4.0016,
+      "step": 1062400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.259699691357822e-05,
+      "loss": 3.9912,
+      "step": 1062912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.258861096606771e-05,
+      "loss": 3.9885,
+      "step": 1063424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.258022501855719e-05,
+      "loss": 3.9929,
+      "step": 1063936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.257183907104667e-05,
+      "loss": 3.9916,
+      "step": 1064448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.256345312353615e-05,
+      "loss": 3.9971,
+      "step": 1064960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.255506717602563e-05,
+      "loss": 3.9913,
+      "step": 1065472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.254668122851511e-05,
+      "loss": 3.9924,
+      "step": 1065984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.253831165980832e-05,
+      "loss": 3.9922,
+      "step": 1066496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.25299257122978e-05,
+      "loss": 3.9976,
+      "step": 1067008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.252153976478728e-05,
+      "loss": 3.9928,
+      "step": 1067520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.251315381727676e-05,
+      "loss": 3.9787,
+      "step": 1068032
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.038179874420166,
+      "eval_runtime": 297.1512,
+      "eval_samples_per_second": 1284.164,
+      "eval_steps_per_second": 40.131,
+      "step": 1068467
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.250478424856997e-05,
+      "loss": 3.9963,
+      "step": 1068544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.249639830105945e-05,
+      "loss": 3.9978,
+      "step": 1069056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.248802873235266e-05,
+      "loss": 3.9947,
+      "step": 1069568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.247964278484214e-05,
+      "loss": 3.9818,
+      "step": 1070080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.247125683733162e-05,
+      "loss": 3.9868,
+      "step": 1070592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.246288726862483e-05,
+      "loss": 3.9928,
+      "step": 1071104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.245450132111431e-05,
+      "loss": 3.9943,
+      "step": 1071616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.244611537360379e-05,
+      "loss": 3.9888,
+      "step": 1072128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.243772942609327e-05,
+      "loss": 3.9805,
+      "step": 1072640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.242934347858275e-05,
+      "loss": 3.9911,
+      "step": 1073152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.242095753107223e-05,
+      "loss": 3.9904,
+      "step": 1073664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.241257158356171e-05,
+      "loss": 3.9825,
+      "step": 1074176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2404185636051185e-05,
+      "loss": 3.9925,
+      "step": 1074688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2395799688540665e-05,
+      "loss": 3.9898,
+      "step": 1075200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2387413741030145e-05,
+      "loss": 3.9902,
+      "step": 1075712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.237902779351963e-05,
+      "loss": 3.9764,
+      "step": 1076224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.237064184600911e-05,
+      "loss": 3.9987,
+      "step": 1076736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.236225589849859e-05,
+      "loss": 3.9846,
+      "step": 1077248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.235386995098807e-05,
+      "loss": 3.9869,
+      "step": 1077760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.234548400347755e-05,
+      "loss": 3.9961,
+      "step": 1078272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.233711443477076e-05,
+      "loss": 4.0028,
+      "step": 1078784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.232872848726024e-05,
+      "loss": 3.9942,
+      "step": 1079296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.232035891855345e-05,
+      "loss": 3.9791,
+      "step": 1079808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.231197297104293e-05,
+      "loss": 3.9952,
+      "step": 1080320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.230358702353241e-05,
+      "loss": 3.9832,
+      "step": 1080832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.229520107602189e-05,
+      "loss": 3.982,
+      "step": 1081344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.228681512851137e-05,
+      "loss": 3.9867,
+      "step": 1081856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.227842918100085e-05,
+      "loss": 3.9932,
+      "step": 1082368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.227004323349033e-05,
+      "loss": 3.981,
+      "step": 1082880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2261673664783545e-05,
+      "loss": 3.9982,
+      "step": 1083392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2253287717273025e-05,
+      "loss": 3.9937,
+      "step": 1083904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2244901769762505e-05,
+      "loss": 3.9903,
+      "step": 1084416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2236532201055714e-05,
+      "loss": 3.9904,
+      "step": 1084928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2228146253545194e-05,
+      "loss": 3.9969,
+      "step": 1085440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2219760306034674e-05,
+      "loss": 3.9865,
+      "step": 1085952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2211374358524154e-05,
+      "loss": 3.9945,
+      "step": 1086464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2202988411013634e-05,
+      "loss": 3.9872,
+      "step": 1086976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2194602463503114e-05,
+      "loss": 3.9894,
+      "step": 1087488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2186216515992594e-05,
+      "loss": 3.9827,
+      "step": 1088000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.21778469472858e-05,
+      "loss": 3.9984,
+      "step": 1088512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.216946099977528e-05,
+      "loss": 3.9998,
+      "step": 1089024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.216107505226476e-05,
+      "loss": 3.9782,
+      "step": 1089536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.215268910475425e-05,
+      "loss": 3.9891,
+      "step": 1090048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.214430315724372e-05,
+      "loss": 3.9779,
+      "step": 1090560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.21359172097332e-05,
+      "loss": 3.9788,
+      "step": 1091072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.212753126222268e-05,
+      "loss": 3.9708,
+      "step": 1091584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.211914531471216e-05,
+      "loss": 3.9856,
+      "step": 1092096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.211075936720164e-05,
+      "loss": 3.9842,
+      "step": 1092608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.210237341969112e-05,
+      "loss": 3.982,
+      "step": 1093120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.20939874721806e-05,
+      "loss": 3.9948,
+      "step": 1093632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.208560152467008e-05,
+      "loss": 3.9624,
+      "step": 1094144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.207724833476702e-05,
+      "loss": 3.9872,
+      "step": 1094656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.20688623872565e-05,
+      "loss": 3.9803,
+      "step": 1095168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.206047643974599e-05,
+      "loss": 3.9682,
+      "step": 1095680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.205209049223547e-05,
+      "loss": 3.9807,
+      "step": 1096192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.204370454472495e-05,
+      "loss": 3.983,
+      "step": 1096704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.203531859721443e-05,
+      "loss": 3.9965,
+      "step": 1097216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.202693264970391e-05,
+      "loss": 3.9737,
+      "step": 1097728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2018563080997116e-05,
+      "loss": 3.9844,
+      "step": 1098240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2010177133486596e-05,
+      "loss": 3.9856,
+      "step": 1098752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.2001791185976076e-05,
+      "loss": 3.9716,
+      "step": 1099264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1993405238465556e-05,
+      "loss": 3.9858,
+      "step": 1099776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1985019290955036e-05,
+      "loss": 3.972,
+      "step": 1100288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1976649722248245e-05,
+      "loss": 3.9754,
+      "step": 1100800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1968263774737725e-05,
+      "loss": 3.9948,
+      "step": 1101312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1959877827227205e-05,
+      "loss": 3.9913,
+      "step": 1101824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1951491879716685e-05,
+      "loss": 3.9791,
+      "step": 1102336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.194310593220617e-05,
+      "loss": 3.9855,
+      "step": 1102848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.193471998469565e-05,
+      "loss": 3.9866,
+      "step": 1103360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.192633403718513e-05,
+      "loss": 3.9762,
+      "step": 1103872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.191794808967461e-05,
+      "loss": 3.982,
+      "step": 1104384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.190957852096782e-05,
+      "loss": 3.9869,
+      "step": 1104896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.19011925734573e-05,
+      "loss": 3.9778,
+      "step": 1105408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.189280662594678e-05,
+      "loss": 3.9728,
+      "step": 1105920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.188442067843626e-05,
+      "loss": 3.9856,
+      "step": 1106432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1876034730925734e-05,
+      "loss": 3.9776,
+      "step": 1106944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1867648783415214e-05,
+      "loss": 3.9833,
+      "step": 1107456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1859262835904694e-05,
+      "loss": 3.9943,
+      "step": 1107968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1850876888394174e-05,
+      "loss": 3.9752,
+      "step": 1108480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.184250731968739e-05,
+      "loss": 3.987,
+      "step": 1108992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.183412137217687e-05,
+      "loss": 3.9832,
+      "step": 1109504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1825751803470086e-05,
+      "loss": 3.9809,
+      "step": 1110016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.181736585595956e-05,
+      "loss": 3.9744,
+      "step": 1110528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.180897990844904e-05,
+      "loss": 3.9751,
+      "step": 1111040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.180059396093852e-05,
+      "loss": 3.9762,
+      "step": 1111552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.1792208013428e-05,
+      "loss": 3.9769,
+      "step": 1112064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.178382206591748e-05,
+      "loss": 3.9754,
+      "step": 1112576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.177543611840696e-05,
+      "loss": 3.9793,
+      "step": 1113088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.176706654970017e-05,
+      "loss": 3.9775,
+      "step": 1113600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.175868060218965e-05,
+      "loss": 3.9799,
+      "step": 1114112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.175029465467913e-05,
+      "loss": 3.9795,
+      "step": 1114624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.174192508597234e-05,
+      "loss": 3.9794,
+      "step": 1115136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.173353913846182e-05,
+      "loss": 3.9779,
+      "step": 1115648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.17251531909513e-05,
+      "loss": 3.9803,
+      "step": 1116160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.171676724344078e-05,
+      "loss": 3.9756,
+      "step": 1116672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.170838129593026e-05,
+      "loss": 3.9786,
+      "step": 1117184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.169999534841974e-05,
+      "loss": 3.9809,
+      "step": 1117696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.169160940090922e-05,
+      "loss": 3.9757,
+      "step": 1118208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.16832234533987e-05,
+      "loss": 3.9706,
+      "step": 1118720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.167483750588818e-05,
+      "loss": 3.9903,
+      "step": 1119232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.166645155837766e-05,
+      "loss": 3.9877,
+      "step": 1119744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.165806561086714e-05,
+      "loss": 3.9852,
+      "step": 1120256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1649679663356616e-05,
+      "loss": 3.9785,
+      "step": 1120768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.164131009464983e-05,
+      "loss": 3.9828,
+      "step": 1121280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.163294052594304e-05,
+      "loss": 3.977,
+      "step": 1121792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.162455457843253e-05,
+      "loss": 3.9728,
+      "step": 1122304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.161616863092201e-05,
+      "loss": 3.9785,
+      "step": 1122816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.160778268341149e-05,
+      "loss": 3.9842,
+      "step": 1123328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.159939673590097e-05,
+      "loss": 3.9648,
+      "step": 1123840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.159101078839045e-05,
+      "loss": 3.9666,
+      "step": 1124352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.158262484087992e-05,
+      "loss": 3.986,
+      "step": 1124864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.157425527217314e-05,
+      "loss": 3.9789,
+      "step": 1125376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.156586932466262e-05,
+      "loss": 3.9692,
+      "step": 1125888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.155748337715209e-05,
+      "loss": 3.9752,
+      "step": 1126400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.154909742964157e-05,
+      "loss": 3.9783,
+      "step": 1126912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.154071148213105e-05,
+      "loss": 3.979,
+      "step": 1127424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1532341913424266e-05,
+      "loss": 3.9734,
+      "step": 1127936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1523955965913746e-05,
+      "loss": 3.976,
+      "step": 1128448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1515570018403226e-05,
+      "loss": 3.9815,
+      "step": 1128960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1507184070892705e-05,
+      "loss": 3.9799,
+      "step": 1129472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1498798123382185e-05,
+      "loss": 3.972,
+      "step": 1129984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1490412175871665e-05,
+      "loss": 3.9922,
+      "step": 1130496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1482026228361145e-05,
+      "loss": 3.9731,
+      "step": 1131008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1473656659654354e-05,
+      "loss": 3.9866,
+      "step": 1131520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1465270712143834e-05,
+      "loss": 3.9823,
+      "step": 1132032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1456884764633314e-05,
+      "loss": 3.9729,
+      "step": 1132544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1448498817122794e-05,
+      "loss": 3.9815,
+      "step": 1133056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1440112869612274e-05,
+      "loss": 3.9699,
+      "step": 1133568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1431726922101754e-05,
+      "loss": 3.9833,
+      "step": 1134080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1423340974591234e-05,
+      "loss": 3.9803,
+      "step": 1134592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.141497140588445e-05,
+      "loss": 3.9935,
+      "step": 1135104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.140658545837393e-05,
+      "loss": 3.9776,
+      "step": 1135616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.139819951086341e-05,
+      "loss": 3.9882,
+      "step": 1136128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.138981356335289e-05,
+      "loss": 3.9759,
+      "step": 1136640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.13814439946461e-05,
+      "loss": 3.9859,
+      "step": 1137152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.137305804713558e-05,
+      "loss": 3.9728,
+      "step": 1137664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.136467209962506e-05,
+      "loss": 3.9759,
+      "step": 1138176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.135628615211454e-05,
+      "loss": 3.9884,
+      "step": 1138688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.134790020460402e-05,
+      "loss": 3.984,
+      "step": 1139200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.13395142570935e-05,
+      "loss": 3.9715,
+      "step": 1139712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.133112830958298e-05,
+      "loss": 3.9787,
+      "step": 1140224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.132274236207245e-05,
+      "loss": 3.9834,
+      "step": 1140736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.131435641456193e-05,
+      "loss": 3.9816,
+      "step": 1141248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.130598684585515e-05,
+      "loss": 3.9781,
+      "step": 1141760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1297600898344634e-05,
+      "loss": 3.9818,
+      "step": 1142272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.128921495083411e-05,
+      "loss": 3.9797,
+      "step": 1142784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.128082900332359e-05,
+      "loss": 3.9809,
+      "step": 1143296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.127244305581307e-05,
+      "loss": 3.9828,
+      "step": 1143808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.126405710830255e-05,
+      "loss": 3.9644,
+      "step": 1144320
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.0305256843566895,
+      "eval_runtime": 296.0522,
+      "eval_samples_per_second": 1288.931,
+      "eval_steps_per_second": 40.28,
+      "step": 1144787
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.125568753959576e-05,
+      "loss": 3.9661,
+      "step": 1144832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1247301592085237e-05,
+      "loss": 3.9772,
+      "step": 1145344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1238915644574717e-05,
+      "loss": 3.971,
+      "step": 1145856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1230529697064196e-05,
+      "loss": 3.9869,
+      "step": 1146368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1222143749553676e-05,
+      "loss": 3.9808,
+      "step": 1146880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1213757802043156e-05,
+      "loss": 3.9888,
+      "step": 1147392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1205371854532636e-05,
+      "loss": 3.9757,
+      "step": 1147904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1196985907022116e-05,
+      "loss": 3.9739,
+      "step": 1148416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1188599959511596e-05,
+      "loss": 3.9683,
+      "step": 1148928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.118021401200108e-05,
+      "loss": 3.9808,
+      "step": 1149440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.117182806449056e-05,
+      "loss": 3.9804,
+      "step": 1149952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.116344211698004e-05,
+      "loss": 3.979,
+      "step": 1150464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.115505616946952e-05,
+      "loss": 3.9848,
+      "step": 1150976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.114668660076273e-05,
+      "loss": 3.9692,
+      "step": 1151488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.113830065325221e-05,
+      "loss": 3.9739,
+      "step": 1152000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.112991470574169e-05,
+      "loss": 3.965,
+      "step": 1152512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1121528758231165e-05,
+      "loss": 3.9747,
+      "step": 1153024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1113142810720645e-05,
+      "loss": 3.9803,
+      "step": 1153536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1104756863210125e-05,
+      "loss": 3.9756,
+      "step": 1154048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1096370915699605e-05,
+      "loss": 3.974,
+      "step": 1154560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1087984968189085e-05,
+      "loss": 3.9885,
+      "step": 1155072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1079599020678565e-05,
+      "loss": 3.9779,
+      "step": 1155584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.107121307316805e-05,
+      "loss": 3.98,
+      "step": 1156096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.106282712565753e-05,
+      "loss": 3.9751,
+      "step": 1156608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.105444117814701e-05,
+      "loss": 3.993,
+      "step": 1157120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.104605523063649e-05,
+      "loss": 3.9705,
+      "step": 1157632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.10376856619297e-05,
+      "loss": 3.9746,
+      "step": 1158144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.102929971441918e-05,
+      "loss": 3.9699,
+      "step": 1158656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.102091376690866e-05,
+      "loss": 3.9727,
+      "step": 1159168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.101252781939814e-05,
+      "loss": 3.9708,
+      "step": 1159680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.100415825069135e-05,
+      "loss": 3.9754,
+      "step": 1160192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.099577230318083e-05,
+      "loss": 3.9799,
+      "step": 1160704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.098738635567031e-05,
+      "loss": 3.98,
+      "step": 1161216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.097900040815979e-05,
+      "loss": 3.9813,
+      "step": 1161728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.097061446064927e-05,
+      "loss": 3.9747,
+      "step": 1162240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0962261270746214e-05,
+      "loss": 3.9852,
+      "step": 1162752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0953875323235694e-05,
+      "loss": 3.9811,
+      "step": 1163264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0945489375725174e-05,
+      "loss": 3.9664,
+      "step": 1163776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0937103428214654e-05,
+      "loss": 3.9713,
+      "step": 1164288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0928717480704134e-05,
+      "loss": 3.9581,
+      "step": 1164800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0920331533193614e-05,
+      "loss": 3.9694,
+      "step": 1165312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0911945585683094e-05,
+      "loss": 3.9661,
+      "step": 1165824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0903559638172574e-05,
+      "loss": 3.9848,
+      "step": 1166336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.089519006946578e-05,
+      "loss": 3.9649,
+      "step": 1166848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.088680412195526e-05,
+      "loss": 3.966,
+      "step": 1167360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.087841817444474e-05,
+      "loss": 3.9714,
+      "step": 1167872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.087003222693422e-05,
+      "loss": 3.9753,
+      "step": 1168384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.086166265822744e-05,
+      "loss": 3.9718,
+      "step": 1168896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.085327671071692e-05,
+      "loss": 3.9542,
+      "step": 1169408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.08448907632064e-05,
+      "loss": 3.9715,
+      "step": 1169920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.083650481569588e-05,
+      "loss": 3.988,
+      "step": 1170432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.082811886818536e-05,
+      "loss": 3.9788,
+      "step": 1170944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.081973292067483e-05,
+      "loss": 3.9745,
+      "step": 1171456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.081134697316431e-05,
+      "loss": 3.9647,
+      "step": 1171968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.080296102565379e-05,
+      "loss": 3.9789,
+      "step": 1172480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.079457507814327e-05,
+      "loss": 3.9606,
+      "step": 1172992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.078620550943648e-05,
+      "loss": 3.9828,
+      "step": 1173504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.07778359407297e-05,
+      "loss": 3.9555,
+      "step": 1174016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.076944999321918e-05,
+      "loss": 3.9889,
+      "step": 1174528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0761064045708657e-05,
+      "loss": 3.9775,
+      "step": 1175040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0752678098198137e-05,
+      "loss": 3.9571,
+      "step": 1175552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0744292150687617e-05,
+      "loss": 3.9665,
+      "step": 1176064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0735906203177096e-05,
+      "loss": 3.9787,
+      "step": 1176576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0727520255666576e-05,
+      "loss": 3.9499,
+      "step": 1177088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0719134308156056e-05,
+      "loss": 3.964,
+      "step": 1177600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0710748360645536e-05,
+      "loss": 3.9831,
+      "step": 1178112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0702378791938745e-05,
+      "loss": 3.9716,
+      "step": 1178624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0693992844428225e-05,
+      "loss": 3.965,
+      "step": 1179136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0685606896917705e-05,
+      "loss": 3.9537,
+      "step": 1179648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0677237328210914e-05,
+      "loss": 3.966,
+      "step": 1180160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0668851380700394e-05,
+      "loss": 3.9608,
+      "step": 1180672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.066046543318988e-05,
+      "loss": 3.9813,
+      "step": 1181184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.065207948567936e-05,
+      "loss": 3.9612,
+      "step": 1181696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.064369353816884e-05,
+      "loss": 3.976,
+      "step": 1182208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.063530759065832e-05,
+      "loss": 3.9778,
+      "step": 1182720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.06269216431478e-05,
+      "loss": 3.9741,
+      "step": 1183232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.061853569563728e-05,
+      "loss": 3.9673,
+      "step": 1183744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.061016612693049e-05,
+      "loss": 3.9681,
+      "step": 1184256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.060178017941997e-05,
+      "loss": 3.9722,
+      "step": 1184768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.059341061071318e-05,
+      "loss": 3.9731,
+      "step": 1185280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.058502466320266e-05,
+      "loss": 3.9653,
+      "step": 1185792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.057663871569214e-05,
+      "loss": 3.968,
+      "step": 1186304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.056825276818162e-05,
+      "loss": 3.9736,
+      "step": 1186816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.05598668206711e-05,
+      "loss": 3.9741,
+      "step": 1187328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.055148087316058e-05,
+      "loss": 3.9531,
+      "step": 1187840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0543111304453795e-05,
+      "loss": 3.969,
+      "step": 1188352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0534725356943275e-05,
+      "loss": 3.969,
+      "step": 1188864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0526339409432755e-05,
+      "loss": 3.9711,
+      "step": 1189376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0517953461922235e-05,
+      "loss": 3.9699,
+      "step": 1189888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0509567514411715e-05,
+      "loss": 3.9664,
+      "step": 1190400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0501181566901188e-05,
+      "loss": 3.9646,
+      "step": 1190912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0492795619390668e-05,
+      "loss": 3.9602,
+      "step": 1191424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0484409671880148e-05,
+      "loss": 3.9635,
+      "step": 1191936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0476023724369628e-05,
+      "loss": 3.9791,
+      "step": 1192448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0467670534466576e-05,
+      "loss": 3.967,
+      "step": 1192960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0459284586956056e-05,
+      "loss": 3.9758,
+      "step": 1193472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0450898639445536e-05,
+      "loss": 3.964,
+      "step": 1193984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0442512691935016e-05,
+      "loss": 3.9628,
+      "step": 1194496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.043412674442449e-05,
+      "loss": 3.962,
+      "step": 1195008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0425757175717705e-05,
+      "loss": 3.9767,
+      "step": 1195520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0417371228207185e-05,
+      "loss": 3.9688,
+      "step": 1196032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.040898528069666e-05,
+      "loss": 3.9717,
+      "step": 1196544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.040059933318614e-05,
+      "loss": 3.9685,
+      "step": 1197056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.039221338567562e-05,
+      "loss": 3.96,
+      "step": 1197568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.03838274381651e-05,
+      "loss": 3.9797,
+      "step": 1198080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0375457869458314e-05,
+      "loss": 3.9577,
+      "step": 1198592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0367071921947794e-05,
+      "loss": 3.9616,
+      "step": 1199104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0358685974437274e-05,
+      "loss": 3.9705,
+      "step": 1199616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0350300026926754e-05,
+      "loss": 3.9654,
+      "step": 1200128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0341914079416234e-05,
+      "loss": 3.9608,
+      "step": 1200640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0333528131905714e-05,
+      "loss": 3.9681,
+      "step": 1201152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0325142184395194e-05,
+      "loss": 3.956,
+      "step": 1201664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0316756236884673e-05,
+      "loss": 3.9706,
+      "step": 1202176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0308386668177886e-05,
+      "loss": 3.9621,
+      "step": 1202688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0300000720667366e-05,
+      "loss": 3.9675,
+      "step": 1203200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0291614773156846e-05,
+      "loss": 3.9733,
+      "step": 1203712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0283228825646326e-05,
+      "loss": 3.983,
+      "step": 1204224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0274842878135806e-05,
+      "loss": 3.971,
+      "step": 1204736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0266473309429015e-05,
+      "loss": 3.9627,
+      "step": 1205248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0258087361918498e-05,
+      "loss": 3.9629,
+      "step": 1205760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0249701414407978e-05,
+      "loss": 3.9662,
+      "step": 1206272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0241315466897458e-05,
+      "loss": 3.9648,
+      "step": 1206784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0232962276994396e-05,
+      "loss": 3.9674,
+      "step": 1207296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0224576329483876e-05,
+      "loss": 3.9613,
+      "step": 1207808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.021619038197336e-05,
+      "loss": 3.9721,
+      "step": 1208320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.020780443446284e-05,
+      "loss": 3.9648,
+      "step": 1208832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.019941848695232e-05,
+      "loss": 3.9717,
+      "step": 1209344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.01910325394418e-05,
+      "loss": 3.9634,
+      "step": 1209856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.018264659193128e-05,
+      "loss": 3.9645,
+      "step": 1210368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.017426064442076e-05,
+      "loss": 3.9596,
+      "step": 1210880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.016589107571397e-05,
+      "loss": 3.9784,
+      "step": 1211392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0157505128203452e-05,
+      "loss": 3.9768,
+      "step": 1211904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0149119180692932e-05,
+      "loss": 3.9707,
+      "step": 1212416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0140733233182412e-05,
+      "loss": 3.9627,
+      "step": 1212928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0132347285671892e-05,
+      "loss": 3.9671,
+      "step": 1213440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0123961338161372e-05,
+      "loss": 3.9701,
+      "step": 1213952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.011557539065085e-05,
+      "loss": 3.9693,
+      "step": 1214464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0107189443140328e-05,
+      "loss": 3.9721,
+      "step": 1214976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0098819874433544e-05,
+      "loss": 3.9756,
+      "step": 1215488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0090433926923024e-05,
+      "loss": 3.9706,
+      "step": 1216000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0082047979412497e-05,
+      "loss": 3.9762,
+      "step": 1216512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0073662031901977e-05,
+      "loss": 3.9632,
+      "step": 1217024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0065276084391457e-05,
+      "loss": 3.9745,
+      "step": 1217536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0056906515684673e-05,
+      "loss": 3.9648,
+      "step": 1218048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.004852056817415e-05,
+      "loss": 3.9598,
+      "step": 1218560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.004013462066363e-05,
+      "loss": 3.9718,
+      "step": 1219072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.003174867315311e-05,
+      "loss": 3.9559,
+      "step": 1219584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.002336272564259e-05,
+      "loss": 3.9636,
+      "step": 1220096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.001497677813207e-05,
+      "loss": 3.9696,
+      "step": 1220608
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.024211406707764,
+      "eval_runtime": 292.3271,
+      "eval_samples_per_second": 1305.356,
+      "eval_steps_per_second": 40.793,
+      "step": 1221107
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.000659083062155e-05,
+      "loss": 3.9635,
+      "step": 1221120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.999820488311103e-05,
+      "loss": 3.9652,
+      "step": 1221632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9989835314404242e-05,
+      "loss": 3.9603,
+      "step": 1222144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9981449366893722e-05,
+      "loss": 3.9748,
+      "step": 1222656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9973063419383202e-05,
+      "loss": 3.9702,
+      "step": 1223168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.996467747187268e-05,
+      "loss": 3.9801,
+      "step": 1223680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.995629152436216e-05,
+      "loss": 3.9625,
+      "step": 1224192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9947921955655374e-05,
+      "loss": 3.9603,
+      "step": 1224704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9939536008144854e-05,
+      "loss": 3.9577,
+      "step": 1225216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9931150060634334e-05,
+      "loss": 3.9716,
+      "step": 1225728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9922764113123814e-05,
+      "loss": 3.9682,
+      "step": 1226240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9914394544417023e-05,
+      "loss": 3.9664,
+      "step": 1226752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9906024975710236e-05,
+      "loss": 3.9686,
+      "step": 1227264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9897655407003445e-05,
+      "loss": 3.9596,
+      "step": 1227776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9889269459492925e-05,
+      "loss": 3.9646,
+      "step": 1228288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9880883511982405e-05,
+      "loss": 3.9566,
+      "step": 1228800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9872497564471885e-05,
+      "loss": 3.9565,
+      "step": 1229312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9864111616961364e-05,
+      "loss": 3.9677,
+      "step": 1229824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9855725669450844e-05,
+      "loss": 3.9615,
+      "step": 1230336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9847339721940328e-05,
+      "loss": 3.9651,
+      "step": 1230848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9838953774429808e-05,
+      "loss": 3.9826,
+      "step": 1231360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9830567826919288e-05,
+      "loss": 3.9631,
+      "step": 1231872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9822181879408768e-05,
+      "loss": 3.9712,
+      "step": 1232384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9813795931898248e-05,
+      "loss": 3.9632,
+      "step": 1232896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9805426363191457e-05,
+      "loss": 3.9783,
+      "step": 1233408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9797040415680937e-05,
+      "loss": 3.9611,
+      "step": 1233920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9788654468170417e-05,
+      "loss": 3.9636,
+      "step": 1234432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.97802685206599e-05,
+      "loss": 3.957,
+      "step": 1234944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.977188257314938e-05,
+      "loss": 3.9647,
+      "step": 1235456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.976349662563886e-05,
+      "loss": 3.9596,
+      "step": 1235968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9755110678128333e-05,
+      "loss": 3.9632,
+      "step": 1236480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9746724730617813e-05,
+      "loss": 3.9686,
+      "step": 1236992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.973835516191103e-05,
+      "loss": 3.9671,
+      "step": 1237504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.972996921440051e-05,
+      "loss": 3.9696,
+      "step": 1238016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.972159964569372e-05,
+      "loss": 3.9682,
+      "step": 1238528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.971323007698693e-05,
+      "loss": 3.9673,
+      "step": 1239040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.970484412947641e-05,
+      "loss": 3.9735,
+      "step": 1239552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.969645818196589e-05,
+      "loss": 3.9528,
+      "step": 1240064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.968807223445537e-05,
+      "loss": 3.9634,
+      "step": 1240576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9679686286944854e-05,
+      "loss": 3.9453,
+      "step": 1241088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9671300339434334e-05,
+      "loss": 3.9585,
+      "step": 1241600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9662914391923807e-05,
+      "loss": 3.955,
+      "step": 1242112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9654528444413287e-05,
+      "loss": 3.9695,
+      "step": 1242624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9646142496902767e-05,
+      "loss": 3.9572,
+      "step": 1243136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9637772928195983e-05,
+      "loss": 3.9525,
+      "step": 1243648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.962938698068546e-05,
+      "loss": 3.9643,
+      "step": 1244160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.962100103317494e-05,
+      "loss": 3.9634,
+      "step": 1244672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.961261508566442e-05,
+      "loss": 3.9578,
+      "step": 1245184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9604245516957628e-05,
+      "loss": 3.9452,
+      "step": 1245696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9595859569447108e-05,
+      "loss": 3.9564,
+      "step": 1246208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.958747362193659e-05,
+      "loss": 3.9745,
+      "step": 1246720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.957908767442607e-05,
+      "loss": 3.9701,
+      "step": 1247232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.957070172691555e-05,
+      "loss": 3.9587,
+      "step": 1247744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.956233215820876e-05,
+      "loss": 3.9586,
+      "step": 1248256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.955394621069824e-05,
+      "loss": 3.9658,
+      "step": 1248768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.954556026318772e-05,
+      "loss": 3.9479,
+      "step": 1249280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.95371743156772e-05,
+      "loss": 3.9704,
+      "step": 1249792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9528788368166684e-05,
+      "loss": 3.9454,
+      "step": 1250304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9520402420656164e-05,
+      "loss": 3.9744,
+      "step": 1250816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9512016473145644e-05,
+      "loss": 3.9686,
+      "step": 1251328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9503630525635123e-05,
+      "loss": 3.948,
+      "step": 1251840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9495277335732062e-05,
+      "loss": 3.9529,
+      "step": 1252352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9486891388221545e-05,
+      "loss": 3.9673,
+      "step": 1252864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9478521819514754e-05,
+      "loss": 3.9403,
+      "step": 1253376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9470135872004234e-05,
+      "loss": 3.9517,
+      "step": 1253888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9461749924493714e-05,
+      "loss": 3.971,
+      "step": 1254400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9453363976983194e-05,
+      "loss": 3.9624,
+      "step": 1254912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9444978029472674e-05,
+      "loss": 3.9552,
+      "step": 1255424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9436592081962154e-05,
+      "loss": 3.9446,
+      "step": 1255936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9428206134451637e-05,
+      "loss": 3.9532,
+      "step": 1256448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9419820186941117e-05,
+      "loss": 3.9456,
+      "step": 1256960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9411434239430597e-05,
+      "loss": 3.9696,
+      "step": 1257472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9403048291920077e-05,
+      "loss": 3.9499,
+      "step": 1257984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9394662344409557e-05,
+      "loss": 3.9638,
+      "step": 1258496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9386292775702766e-05,
+      "loss": 3.9646,
+      "step": 1259008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.937792320699598e-05,
+      "loss": 3.9682,
+      "step": 1259520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.936953725948546e-05,
+      "loss": 3.955,
+      "step": 1260032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.936115131197494e-05,
+      "loss": 3.9586,
+      "step": 1260544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.935276536446442e-05,
+      "loss": 3.9652,
+      "step": 1261056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.93443794169539e-05,
+      "loss": 3.9611,
+      "step": 1261568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.933599346944338e-05,
+      "loss": 3.957,
+      "step": 1262080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.932760752193286e-05,
+      "loss": 3.9531,
+      "step": 1262592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.931922157442234e-05,
+      "loss": 3.9638,
+      "step": 1263104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.931085200571555e-05,
+      "loss": 3.9624,
+      "step": 1263616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.930248243700876e-05,
+      "loss": 3.9465,
+      "step": 1264128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.929409648949824e-05,
+      "loss": 3.9526,
+      "step": 1264640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.928571054198772e-05,
+      "loss": 3.9574,
+      "step": 1265152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.92773245944772e-05,
+      "loss": 3.9606,
+      "step": 1265664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9268955025770412e-05,
+      "loss": 3.9573,
+      "step": 1266176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.9260569078259892e-05,
+      "loss": 3.9583,
+      "step": 1266688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9252183130749372e-05,
+      "loss": 3.9529,
+      "step": 1267200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9243797183238852e-05,
+      "loss": 3.9455,
+      "step": 1267712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9235411235728332e-05,
+      "loss": 3.952,
+      "step": 1268224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9227025288217812e-05,
+      "loss": 3.9642,
+      "step": 1268736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9218655719511025e-05,
+      "loss": 3.9561,
+      "step": 1269248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9210269772000505e-05,
+      "loss": 3.9643,
+      "step": 1269760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9201883824489984e-05,
+      "loss": 3.9576,
+      "step": 1270272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9193497876979464e-05,
+      "loss": 3.9488,
+      "step": 1270784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9185111929468938e-05,
+      "loss": 3.9517,
+      "step": 1271296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.917672598195842e-05,
+      "loss": 3.9646,
+      "step": 1271808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9168356413251637e-05,
+      "loss": 3.9564,
+      "step": 1272320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9159970465741117e-05,
+      "loss": 3.9662,
+      "step": 1272832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.915158451823059e-05,
+      "loss": 3.9523,
+      "step": 1273344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.914319857072007e-05,
+      "loss": 3.9481,
+      "step": 1273856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.913481262320955e-05,
+      "loss": 3.9702,
+      "step": 1274368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9126443054502762e-05,
+      "loss": 3.9441,
+      "step": 1274880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9118057106992242e-05,
+      "loss": 3.9511,
+      "step": 1275392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9109671159481722e-05,
+      "loss": 3.9592,
+      "step": 1275904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9101285211971202e-05,
+      "loss": 3.9556,
+      "step": 1276416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9092899264460682e-05,
+      "loss": 3.9502,
+      "step": 1276928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9084513316950162e-05,
+      "loss": 3.9503,
+      "step": 1277440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9076143748243375e-05,
+      "loss": 3.9487,
+      "step": 1277952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9067757800732855e-05,
+      "loss": 3.9585,
+      "step": 1278464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9059371853222335e-05,
+      "loss": 3.9525,
+      "step": 1278976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9050985905711815e-05,
+      "loss": 3.9533,
+      "step": 1279488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9042599958201294e-05,
+      "loss": 3.965,
+      "step": 1280000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9034214010690774e-05,
+      "loss": 3.9721,
+      "step": 1280512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9025844441983984e-05,
+      "loss": 3.9603,
+      "step": 1281024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9017458494473463e-05,
+      "loss": 3.9552,
+      "step": 1281536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9009072546962947e-05,
+      "loss": 3.9498,
+      "step": 1282048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.9000702978256156e-05,
+      "loss": 3.9518,
+      "step": 1282560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8992317030745636e-05,
+      "loss": 3.9574,
+      "step": 1283072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8983931083235116e-05,
+      "loss": 3.957,
+      "step": 1283584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8975545135724596e-05,
+      "loss": 3.9509,
+      "step": 1284096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8967159188214076e-05,
+      "loss": 3.9581,
+      "step": 1284608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8958773240703556e-05,
+      "loss": 3.9561,
+      "step": 1285120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.895038729319304e-05,
+      "loss": 3.9567,
+      "step": 1285632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8942017724486248e-05,
+      "loss": 3.9528,
+      "step": 1286144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8933631776975728e-05,
+      "loss": 3.9536,
+      "step": 1286656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8925245829465208e-05,
+      "loss": 3.9509,
+      "step": 1287168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8916859881954688e-05,
+      "loss": 3.9642,
+      "step": 1287680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8908473934444168e-05,
+      "loss": 3.9666,
+      "step": 1288192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8900087986933648e-05,
+      "loss": 3.9647,
+      "step": 1288704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8891702039423124e-05,
+      "loss": 3.9464,
+      "step": 1289216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8883316091912604e-05,
+      "loss": 3.9601,
+      "step": 1289728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8874930144402084e-05,
+      "loss": 3.9569,
+      "step": 1290240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8866544196891564e-05,
+      "loss": 3.9604,
+      "step": 1290752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8858158249381044e-05,
+      "loss": 3.9528,
+      "step": 1291264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8849788680674257e-05,
+      "loss": 3.97,
+      "step": 1291776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8841402733163737e-05,
+      "loss": 3.9543,
+      "step": 1292288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8833033164456953e-05,
+      "loss": 3.9712,
+      "step": 1292800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8824647216946426e-05,
+      "loss": 3.9485,
+      "step": 1293312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8816261269435906e-05,
+      "loss": 3.9664,
+      "step": 1293824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8807875321925386e-05,
+      "loss": 3.9535,
+      "step": 1294336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.879948937441487e-05,
+      "loss": 3.9489,
+      "step": 1294848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.879110342690435e-05,
+      "loss": 3.959,
+      "step": 1295360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.878271747939383e-05,
+      "loss": 3.9472,
+      "step": 1295872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.877433153188331e-05,
+      "loss": 3.9518,
+      "step": 1296384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8765961963176518e-05,
+      "loss": 3.9565,
+      "step": 1296896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8757576015665998e-05,
+      "loss": 3.9515,
+      "step": 1297408
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.018600940704346,
+      "eval_runtime": 295.6113,
+      "eval_samples_per_second": 1290.854,
+      "eval_steps_per_second": 40.34,
+      "step": 1297427
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8749190068155478e-05,
+      "loss": 3.9543,
+      "step": 1297920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.874080412064496e-05,
+      "loss": 3.9529,
+      "step": 1298432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.873241817313444e-05,
+      "loss": 3.9636,
+      "step": 1298944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.872403222562392e-05,
+      "loss": 3.9557,
+      "step": 1299456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.87156462781134e-05,
+      "loss": 3.9725,
+      "step": 1299968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.870726033060288e-05,
+      "loss": 3.9525,
+      "step": 1300480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.869887438309236e-05,
+      "loss": 3.9503,
+      "step": 1300992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8690488435581838e-05,
+      "loss": 3.9468,
+      "step": 1301504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8682102488071318e-05,
+      "loss": 3.9611,
+      "step": 1302016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8673716540560798e-05,
+      "loss": 3.9575,
+      "step": 1302528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8665330593050277e-05,
+      "loss": 3.9542,
+      "step": 1303040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8656944645539757e-05,
+      "loss": 3.9597,
+      "step": 1303552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8648575076832967e-05,
+      "loss": 3.9434,
+      "step": 1304064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8640189129322446e-05,
+      "loss": 3.9575,
+      "step": 1304576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8631803181811926e-05,
+      "loss": 3.9445,
+      "step": 1305088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.862341723430141e-05,
+      "loss": 3.9479,
+      "step": 1305600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.861503128679089e-05,
+      "loss": 3.9541,
+      "step": 1306112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.86066617180841e-05,
+      "loss": 3.9502,
+      "step": 1306624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.859827577057358e-05,
+      "loss": 3.9533,
+      "step": 1307136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.858988982306306e-05,
+      "loss": 3.9757,
+      "step": 1307648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.858150387555254e-05,
+      "loss": 3.9502,
+      "step": 1308160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.857311792804202e-05,
+      "loss": 3.9591,
+      "step": 1308672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8564731980531502e-05,
+      "loss": 3.9559,
+      "step": 1309184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8556346033020982e-05,
+      "loss": 3.9646,
+      "step": 1309696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8547960085510462e-05,
+      "loss": 3.9523,
+      "step": 1310208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8539574137999942e-05,
+      "loss": 3.9554,
+      "step": 1310720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8531188190489422e-05,
+      "loss": 3.945,
+      "step": 1311232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.852281862178263e-05,
+      "loss": 3.9566,
+      "step": 1311744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.851443267427211e-05,
+      "loss": 3.945,
+      "step": 1312256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8506063105565323e-05,
+      "loss": 3.9536,
+      "step": 1312768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8497677158054803e-05,
+      "loss": 3.9589,
+      "step": 1313280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8489291210544283e-05,
+      "loss": 3.9548,
+      "step": 1313792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8480905263033763e-05,
+      "loss": 3.9622,
+      "step": 1314304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8472519315523243e-05,
+      "loss": 3.9548,
+      "step": 1314816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8464149746816456e-05,
+      "loss": 3.9602,
+      "step": 1315328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8455763799305936e-05,
+      "loss": 3.9577,
+      "step": 1315840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8447377851795416e-05,
+      "loss": 3.9498,
+      "step": 1316352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8438991904284896e-05,
+      "loss": 3.9486,
+      "step": 1316864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8430605956774375e-05,
+      "loss": 3.9358,
+      "step": 1317376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.842222000926385e-05,
+      "loss": 3.9505,
+      "step": 1317888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8413834061753332e-05,
+      "loss": 3.9475,
+      "step": 1318400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8405448114242812e-05,
+      "loss": 3.9572,
+      "step": 1318912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8397062166732292e-05,
+      "loss": 3.9464,
+      "step": 1319424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8388676219221772e-05,
+      "loss": 3.948,
+      "step": 1319936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8380290271711252e-05,
+      "loss": 3.9531,
+      "step": 1320448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8371904324200732e-05,
+      "loss": 3.9522,
+      "step": 1320960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.836353475549394e-05,
+      "loss": 3.948,
+      "step": 1321472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8355148807983424e-05,
+      "loss": 3.9381,
+      "step": 1321984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8346762860472904e-05,
+      "loss": 3.9418,
+      "step": 1322496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8338376912962384e-05,
+      "loss": 3.966,
+      "step": 1323008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8329990965451864e-05,
+      "loss": 3.9607,
+      "step": 1323520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8321621396745073e-05,
+      "loss": 3.9455,
+      "step": 1324032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8313235449234553e-05,
+      "loss": 3.9471,
+      "step": 1324544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8304865880527766e-05,
+      "loss": 3.9576,
+      "step": 1325056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8296479933017246e-05,
+      "loss": 3.9386,
+      "step": 1325568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8288110364310455e-05,
+      "loss": 3.9585,
+      "step": 1326080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8279724416799935e-05,
+      "loss": 3.937,
+      "step": 1326592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8271338469289415e-05,
+      "loss": 3.9647,
+      "step": 1327104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8262952521778895e-05,
+      "loss": 3.9568,
+      "step": 1327616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8254566574268378e-05,
+      "loss": 3.9458,
+      "step": 1328128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8246180626757858e-05,
+      "loss": 3.9379,
+      "step": 1328640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8237794679247338e-05,
+      "loss": 3.9597,
+      "step": 1329152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8229408731736818e-05,
+      "loss": 3.9289,
+      "step": 1329664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8221022784226298e-05,
+      "loss": 3.9431,
+      "step": 1330176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8212636836715778e-05,
+      "loss": 3.9576,
+      "step": 1330688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8204250889205258e-05,
+      "loss": 3.9541,
+      "step": 1331200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8195864941694738e-05,
+      "loss": 3.9452,
+      "step": 1331712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.818749537298795e-05,
+      "loss": 3.9336,
+      "step": 1332224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.817910942547743e-05,
+      "loss": 3.9417,
+      "step": 1332736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.817072347796691e-05,
+      "loss": 3.9414,
+      "step": 1333248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8162337530456383e-05,
+      "loss": 3.9547,
+      "step": 1333760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.81539679617496e-05,
+      "loss": 3.9427,
+      "step": 1334272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.814558201423908e-05,
+      "loss": 3.9525,
+      "step": 1334784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.813719606672856e-05,
+      "loss": 3.9553,
+      "step": 1335296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8128810119218036e-05,
+      "loss": 3.9595,
+      "step": 1335808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.812044055051125e-05,
+      "loss": 3.9437,
+      "step": 1336320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.811205460300073e-05,
+      "loss": 3.9467,
+      "step": 1336832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.810366865549021e-05,
+      "loss": 3.9546,
+      "step": 1337344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8095282707979688e-05,
+      "loss": 3.9508,
+      "step": 1337856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8086913139272904e-05,
+      "loss": 3.9475,
+      "step": 1338368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8078527191762384e-05,
+      "loss": 3.9446,
+      "step": 1338880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8070141244251857e-05,
+      "loss": 3.9562,
+      "step": 1339392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8061755296741337e-05,
+      "loss": 3.9488,
+      "step": 1339904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8053369349230817e-05,
+      "loss": 3.9382,
+      "step": 1340416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8044983401720297e-05,
+      "loss": 3.9429,
+      "step": 1340928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.803659745420978e-05,
+      "loss": 3.9483,
+      "step": 1341440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.802822788550299e-05,
+      "loss": 3.9536,
+      "step": 1341952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.801984193799247e-05,
+      "loss": 3.9491,
+      "step": 1342464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.801145599048195e-05,
+      "loss": 3.9451,
+      "step": 1342976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.800307004297143e-05,
+      "loss": 3.9429,
+      "step": 1343488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.799470047426464e-05,
+      "loss": 3.9373,
+      "step": 1344000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.798631452675412e-05,
+      "loss": 3.9393,
+      "step": 1344512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.79779285792436e-05,
+      "loss": 3.952,
+      "step": 1345024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.796954263173308e-05,
+      "loss": 3.9504,
+      "step": 1345536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.796115668422256e-05,
+      "loss": 3.9544,
+      "step": 1346048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.795278711551577e-05,
+      "loss": 3.953,
+      "step": 1346560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.794440116800525e-05,
+      "loss": 3.939,
+      "step": 1347072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7936015220494734e-05,
+      "loss": 3.9442,
+      "step": 1347584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7927629272984214e-05,
+      "loss": 3.9498,
+      "step": 1348096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7919243325473694e-05,
+      "loss": 3.9542,
+      "step": 1348608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7910857377963174e-05,
+      "loss": 3.9552,
+      "step": 1349120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7902487809256383e-05,
+      "loss": 3.9459,
+      "step": 1349632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7894101861745863e-05,
+      "loss": 3.9382,
+      "step": 1350144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7885715914235343e-05,
+      "loss": 3.9602,
+      "step": 1350656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7877329966724826e-05,
+      "loss": 3.9353,
+      "step": 1351168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7868960398018035e-05,
+      "loss": 3.9454,
+      "step": 1351680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7860574450507515e-05,
+      "loss": 3.945,
+      "step": 1352192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7852188502996995e-05,
+      "loss": 3.948,
+      "step": 1352704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7843802555486475e-05,
+      "loss": 3.9382,
+      "step": 1353216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7835416607975955e-05,
+      "loss": 3.9466,
+      "step": 1353728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7827030660465435e-05,
+      "loss": 3.939,
+      "step": 1354240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7818661091758647e-05,
+      "loss": 3.9489,
+      "step": 1354752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7810275144248127e-05,
+      "loss": 3.943,
+      "step": 1355264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7801889196737607e-05,
+      "loss": 3.9391,
+      "step": 1355776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7793503249227087e-05,
+      "loss": 3.9567,
+      "step": 1356288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7785117301716567e-05,
+      "loss": 3.9637,
+      "step": 1356800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7776731354206047e-05,
+      "loss": 3.9515,
+      "step": 1357312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7768345406695524e-05,
+      "loss": 3.9432,
+      "step": 1357824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7759959459185004e-05,
+      "loss": 3.9422,
+      "step": 1358336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.775158989047822e-05,
+      "loss": 3.9425,
+      "step": 1358848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.774322032177143e-05,
+      "loss": 3.9444,
+      "step": 1359360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.773485075306464e-05,
+      "loss": 3.9528,
+      "step": 1359872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.772646480555412e-05,
+      "loss": 3.9361,
+      "step": 1360384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.77180788580436e-05,
+      "loss": 3.9471,
+      "step": 1360896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.770969291053308e-05,
+      "loss": 3.9495,
+      "step": 1361408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.770130696302256e-05,
+      "loss": 3.95,
+      "step": 1361920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.769292101551204e-05,
+      "loss": 3.9395,
+      "step": 1362432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.768453506800152e-05,
+      "loss": 3.9434,
+      "step": 1362944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7676149120490997e-05,
+      "loss": 3.9465,
+      "step": 1363456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7667779551784213e-05,
+      "loss": 3.9508,
+      "step": 1363968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7659393604273693e-05,
+      "loss": 3.956,
+      "step": 1364480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7651007656763166e-05,
+      "loss": 3.9553,
+      "step": 1364992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7642621709252646e-05,
+      "loss": 3.9365,
+      "step": 1365504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7634235761742126e-05,
+      "loss": 3.9497,
+      "step": 1366016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7625866193035342e-05,
+      "loss": 3.9517,
+      "step": 1366528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.761748024552482e-05,
+      "loss": 3.9471,
+      "step": 1367040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.76090942980143e-05,
+      "loss": 3.9479,
+      "step": 1367552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.760070835050378e-05,
+      "loss": 3.9566,
+      "step": 1368064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.759232240299326e-05,
+      "loss": 3.946,
+      "step": 1368576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.758395283428647e-05,
+      "loss": 3.9639,
+      "step": 1369088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.757556688677595e-05,
+      "loss": 3.9391,
+      "step": 1369600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.756718093926543e-05,
+      "loss": 3.9545,
+      "step": 1370112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.755879499175491e-05,
+      "loss": 3.947,
+      "step": 1370624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.755040904424439e-05,
+      "loss": 3.9419,
+      "step": 1371136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.754202309673387e-05,
+      "loss": 3.9496,
+      "step": 1371648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.753365352802708e-05,
+      "loss": 3.9401,
+      "step": 1372160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7525267580516563e-05,
+      "loss": 3.941,
+      "step": 1372672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7516881633006043e-05,
+      "loss": 3.945,
+      "step": 1373184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7508495685495523e-05,
+      "loss": 3.9419,
+      "step": 1373696
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.014351844787598,
+      "eval_runtime": 296.8234,
+      "eval_samples_per_second": 1285.583,
+      "eval_steps_per_second": 40.175,
+      "step": 1373747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7500109737985003e-05,
+      "loss": 3.9468,
+      "step": 1374208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7491723790474483e-05,
+      "loss": 3.9404,
+      "step": 1374720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7483337842963963e-05,
+      "loss": 3.9573,
+      "step": 1375232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7474951895453443e-05,
+      "loss": 3.9447,
+      "step": 1375744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7466582326746652e-05,
+      "loss": 3.962,
+      "step": 1376256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7458196379236135e-05,
+      "loss": 3.941,
+      "step": 1376768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7449810431725615e-05,
+      "loss": 3.947,
+      "step": 1377280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7441424484215095e-05,
+      "loss": 3.933,
+      "step": 1377792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7433071294312034e-05,
+      "loss": 3.9551,
+      "step": 1378304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7424685346801514e-05,
+      "loss": 3.9446,
+      "step": 1378816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7416299399290997e-05,
+      "loss": 3.9415,
+      "step": 1379328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7407913451780477e-05,
+      "loss": 3.9568,
+      "step": 1379840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7399543883073686e-05,
+      "loss": 3.9353,
+      "step": 1380352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7391157935563166e-05,
+      "loss": 3.9508,
+      "step": 1380864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7382771988052646e-05,
+      "loss": 3.9329,
+      "step": 1381376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7374386040542126e-05,
+      "loss": 3.9424,
+      "step": 1381888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7366000093031606e-05,
+      "loss": 3.9414,
+      "step": 1382400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.735761414552109e-05,
+      "loss": 3.944,
+      "step": 1382912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.734922819801057e-05,
+      "loss": 3.9425,
+      "step": 1383424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.734084225050005e-05,
+      "loss": 3.9665,
+      "step": 1383936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7332472681793258e-05,
+      "loss": 3.942,
+      "step": 1384448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7324086734282738e-05,
+      "loss": 3.9496,
+      "step": 1384960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7315700786772218e-05,
+      "loss": 3.9465,
+      "step": 1385472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7307314839261698e-05,
+      "loss": 3.9571,
+      "step": 1385984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.729892889175118e-05,
+      "loss": 3.9426,
+      "step": 1386496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.729055932304439e-05,
+      "loss": 3.9495,
+      "step": 1387008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.728217337553387e-05,
+      "loss": 3.9343,
+      "step": 1387520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.727378742802335e-05,
+      "loss": 3.9447,
+      "step": 1388032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7265401480512827e-05,
+      "loss": 3.9347,
+      "step": 1388544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7257015533002307e-05,
+      "loss": 3.9481,
+      "step": 1389056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7248645964295523e-05,
+      "loss": 3.9463,
+      "step": 1389568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7240260016785003e-05,
+      "loss": 3.9486,
+      "step": 1390080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7231874069274476e-05,
+      "loss": 3.9556,
+      "step": 1390592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7223488121763956e-05,
+      "loss": 3.9444,
+      "step": 1391104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7215118553057172e-05,
+      "loss": 3.9525,
+      "step": 1391616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.720673260554665e-05,
+      "loss": 3.9501,
+      "step": 1392128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7198346658036128e-05,
+      "loss": 3.9424,
+      "step": 1392640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7189960710525608e-05,
+      "loss": 3.9358,
+      "step": 1393152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7181574763015088e-05,
+      "loss": 3.9323,
+      "step": 1393664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7173188815504568e-05,
+      "loss": 3.938,
+      "step": 1394176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7164802867994048e-05,
+      "loss": 3.9377,
+      "step": 1394688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7156416920483528e-05,
+      "loss": 3.9477,
+      "step": 1395200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.714804735177674e-05,
+      "loss": 3.9408,
+      "step": 1395712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.713967778306995e-05,
+      "loss": 3.9356,
+      "step": 1396224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.713129183555943e-05,
+      "loss": 3.9484,
+      "step": 1396736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.712290588804891e-05,
+      "loss": 3.9411,
+      "step": 1397248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.711451994053839e-05,
+      "loss": 3.9394,
+      "step": 1397760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7106133993027873e-05,
+      "loss": 3.9337,
+      "step": 1398272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7097748045517353e-05,
+      "loss": 3.9279,
+      "step": 1398784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7089362098006833e-05,
+      "loss": 3.9577,
+      "step": 1399296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7080976150496313e-05,
+      "loss": 3.952,
+      "step": 1399808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.707262296059325e-05,
+      "loss": 3.9389,
+      "step": 1400320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7064237013082734e-05,
+      "loss": 3.9392,
+      "step": 1400832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7055851065572214e-05,
+      "loss": 3.9466,
+      "step": 1401344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7047465118061694e-05,
+      "loss": 3.9286,
+      "step": 1401856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7039079170551174e-05,
+      "loss": 3.9499,
+      "step": 1402368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7030693223040654e-05,
+      "loss": 3.9275,
+      "step": 1402880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7022307275530134e-05,
+      "loss": 3.9562,
+      "step": 1403392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7013921328019614e-05,
+      "loss": 3.9512,
+      "step": 1403904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7005535380509094e-05,
+      "loss": 3.936,
+      "step": 1404416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6997149432998574e-05,
+      "loss": 3.9309,
+      "step": 1404928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6988763485488054e-05,
+      "loss": 3.9509,
+      "step": 1405440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6980377537977537e-05,
+      "loss": 3.919,
+      "step": 1405952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.697199159046701e-05,
+      "loss": 3.936,
+      "step": 1406464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6963622021760226e-05,
+      "loss": 3.9481,
+      "step": 1406976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6955236074249706e-05,
+      "loss": 3.9462,
+      "step": 1407488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6946850126739186e-05,
+      "loss": 3.9354,
+      "step": 1408000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.69384805580324e-05,
+      "loss": 3.9263,
+      "step": 1408512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6930110989325608e-05,
+      "loss": 3.9314,
+      "step": 1409024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6921725041815088e-05,
+      "loss": 3.9317,
+      "step": 1409536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6913339094304568e-05,
+      "loss": 3.9444,
+      "step": 1410048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6904953146794048e-05,
+      "loss": 3.933,
+      "step": 1410560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6896567199283528e-05,
+      "loss": 3.9455,
+      "step": 1411072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6888181251773008e-05,
+      "loss": 3.9491,
+      "step": 1411584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6879795304262484e-05,
+      "loss": 3.9477,
+      "step": 1412096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6871409356751964e-05,
+      "loss": 3.9359,
+      "step": 1412608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6863023409241444e-05,
+      "loss": 3.938,
+      "step": 1413120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6854637461730924e-05,
+      "loss": 3.9467,
+      "step": 1413632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6846251514220404e-05,
+      "loss": 3.9393,
+      "step": 1414144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6837865566709884e-05,
+      "loss": 3.9394,
+      "step": 1414656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6829479619199367e-05,
+      "loss": 3.9408,
+      "step": 1415168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6821126429296312e-05,
+      "loss": 3.9415,
+      "step": 1415680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6812740481785785e-05,
+      "loss": 3.9432,
+      "step": 1416192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6804354534275265e-05,
+      "loss": 3.9287,
+      "step": 1416704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6795968586764745e-05,
+      "loss": 3.9354,
+      "step": 1417216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.678758263925423e-05,
+      "loss": 3.9371,
+      "step": 1417728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.677919669174371e-05,
+      "loss": 3.9464,
+      "step": 1418240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.677081074423319e-05,
+      "loss": 3.9405,
+      "step": 1418752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6762441175526398e-05,
+      "loss": 3.9377,
+      "step": 1419264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6754055228015878e-05,
+      "loss": 3.9347,
+      "step": 1419776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6745669280505358e-05,
+      "loss": 3.9271,
+      "step": 1420288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6737283332994838e-05,
+      "loss": 3.9309,
+      "step": 1420800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.672889738548432e-05,
+      "loss": 3.9396,
+      "step": 1421312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.67205114379738e-05,
+      "loss": 3.9484,
+      "step": 1421824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.671212549046328e-05,
+      "loss": 3.9458,
+      "step": 1422336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.670375592175649e-05,
+      "loss": 3.9409,
+      "step": 1422848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.669536997424597e-05,
+      "loss": 3.9344,
+      "step": 1423360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.668698402673545e-05,
+      "loss": 3.9331,
+      "step": 1423872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.667859807922493e-05,
+      "loss": 3.9403,
+      "step": 1424384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6670212131714413e-05,
+      "loss": 3.947,
+      "step": 1424896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6661826184203893e-05,
+      "loss": 3.9438,
+      "step": 1425408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6653440236693373e-05,
+      "loss": 3.9364,
+      "step": 1425920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6645054289182846e-05,
+      "loss": 3.9316,
+      "step": 1426432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6636684720476062e-05,
+      "loss": 3.948,
+      "step": 1426944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6628315151769275e-05,
+      "loss": 3.9283,
+      "step": 1427456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6619929204258755e-05,
+      "loss": 3.939,
+      "step": 1427968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6611543256748234e-05,
+      "loss": 3.9372,
+      "step": 1428480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6603157309237714e-05,
+      "loss": 3.9376,
+      "step": 1428992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6594771361727194e-05,
+      "loss": 3.9326,
+      "step": 1429504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6586385414216674e-05,
+      "loss": 3.939,
+      "step": 1430016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.657799946670615e-05,
+      "loss": 3.9285,
+      "step": 1430528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.656961351919563e-05,
+      "loss": 3.9348,
+      "step": 1431040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6561243950488847e-05,
+      "loss": 3.9411,
+      "step": 1431552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.655285800297832e-05,
+      "loss": 3.9277,
+      "step": 1432064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6544488434271536e-05,
+      "loss": 3.9522,
+      "step": 1432576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6536102486761016e-05,
+      "loss": 3.9525,
+      "step": 1433088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6527716539250496e-05,
+      "loss": 3.9482,
+      "step": 1433600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6519330591739972e-05,
+      "loss": 3.9327,
+      "step": 1434112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6510944644229452e-05,
+      "loss": 3.9356,
+      "step": 1434624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6502558696718932e-05,
+      "loss": 3.9347,
+      "step": 1435136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6494172749208412e-05,
+      "loss": 3.9319,
+      "step": 1435648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6485786801697892e-05,
+      "loss": 3.9472,
+      "step": 1436160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6477417232991105e-05,
+      "loss": 3.9281,
+      "step": 1436672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6469031285480585e-05,
+      "loss": 3.9401,
+      "step": 1437184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6460645337970064e-05,
+      "loss": 3.9431,
+      "step": 1437696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6452259390459544e-05,
+      "loss": 3.9392,
+      "step": 1438208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6443889821752754e-05,
+      "loss": 3.9312,
+      "step": 1438720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.643552025304597e-05,
+      "loss": 3.9377,
+      "step": 1439232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6427134305535446e-05,
+      "loss": 3.9386,
+      "step": 1439744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6418748358024926e-05,
+      "loss": 3.9429,
+      "step": 1440256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6410362410514406e-05,
+      "loss": 3.9427,
+      "step": 1440768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6401976463003886e-05,
+      "loss": 3.9491,
+      "step": 1441280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6393590515493366e-05,
+      "loss": 3.927,
+      "step": 1441792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6385204567982846e-05,
+      "loss": 3.9449,
+      "step": 1442304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6376818620472326e-05,
+      "loss": 3.9453,
+      "step": 1442816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6368432672961806e-05,
+      "loss": 3.9367,
+      "step": 1443328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6360063104255018e-05,
+      "loss": 3.9391,
+      "step": 1443840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6351693535548227e-05,
+      "loss": 3.951,
+      "step": 1444352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6343307588037707e-05,
+      "loss": 3.9401,
+      "step": 1444864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6334921640527187e-05,
+      "loss": 3.9547,
+      "step": 1445376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6326535693016667e-05,
+      "loss": 3.9339,
+      "step": 1445888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.631816612430988e-05,
+      "loss": 3.9443,
+      "step": 1446400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.630978017679936e-05,
+      "loss": 3.9395,
+      "step": 1446912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.630139422928884e-05,
+      "loss": 3.9376,
+      "step": 1447424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.629300828177832e-05,
+      "loss": 3.9393,
+      "step": 1447936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.628463871307153e-05,
+      "loss": 3.9327,
+      "step": 1448448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6276252765561012e-05,
+      "loss": 3.9321,
+      "step": 1448960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6267866818050492e-05,
+      "loss": 3.9369,
+      "step": 1449472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6259480870539972e-05,
+      "loss": 3.9319,
+      "step": 1449984
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.010540962219238,
+      "eval_runtime": 296.8543,
+      "eval_samples_per_second": 1285.449,
+      "eval_steps_per_second": 40.171,
+      "step": 1450067
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6251094923029452e-05,
+      "loss": 3.94,
+      "step": 1450496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6242708975518932e-05,
+      "loss": 3.932,
+      "step": 1451008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.623432302800841e-05,
+      "loss": 3.9481,
+      "step": 1451520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.622593708049789e-05,
+      "loss": 3.9359,
+      "step": 1452032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.621755113298737e-05,
+      "loss": 3.959,
+      "step": 1452544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.620916518547685e-05,
+      "loss": 3.9319,
+      "step": 1453056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.620077923796633e-05,
+      "loss": 3.9371,
+      "step": 1453568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6192393290455808e-05,
+      "loss": 3.9277,
+      "step": 1454080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6184007342945288e-05,
+      "loss": 3.9421,
+      "step": 1454592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6175621395434768e-05,
+      "loss": 3.9407,
+      "step": 1455104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6167235447924248e-05,
+      "loss": 3.9328,
+      "step": 1455616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6158849500413728e-05,
+      "loss": 3.9461,
+      "step": 1456128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.615047993170694e-05,
+      "loss": 3.9266,
+      "step": 1456640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.614209398419642e-05,
+      "loss": 3.9415,
+      "step": 1457152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.61337080366859e-05,
+      "loss": 3.9281,
+      "step": 1457664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.612532208917538e-05,
+      "loss": 3.9304,
+      "step": 1458176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.611693614166486e-05,
+      "loss": 3.9301,
+      "step": 1458688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.610855019415434e-05,
+      "loss": 3.9371,
+      "step": 1459200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.610016424664382e-05,
+      "loss": 3.9382,
+      "step": 1459712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.60917782991333e-05,
+      "loss": 3.954,
+      "step": 1460224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6083392351622783e-05,
+      "loss": 3.9345,
+      "step": 1460736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6075022782915993e-05,
+      "loss": 3.9444,
+      "step": 1461248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6066636835405472e-05,
+      "loss": 3.9381,
+      "step": 1461760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6058250887894952e-05,
+      "loss": 3.9448,
+      "step": 1462272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6049864940384432e-05,
+      "loss": 3.9353,
+      "step": 1462784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6041495371677645e-05,
+      "loss": 3.9382,
+      "step": 1463296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6033109424167125e-05,
+      "loss": 3.9317,
+      "step": 1463808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6024723476656605e-05,
+      "loss": 3.9343,
+      "step": 1464320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6016337529146085e-05,
+      "loss": 3.9274,
+      "step": 1464832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6007967960439294e-05,
+      "loss": 3.9379,
+      "step": 1465344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5999582012928774e-05,
+      "loss": 3.9434,
+      "step": 1465856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5991196065418254e-05,
+      "loss": 3.9368,
+      "step": 1466368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5982810117907737e-05,
+      "loss": 3.9477,
+      "step": 1466880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5974424170397217e-05,
+      "loss": 3.9375,
+      "step": 1467392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5966070980494155e-05,
+      "loss": 3.9376,
+      "step": 1467904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5957685032983635e-05,
+      "loss": 3.9432,
+      "step": 1468416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5949299085473115e-05,
+      "loss": 3.9375,
+      "step": 1468928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.59409131379626e-05,
+      "loss": 3.9276,
+      "step": 1469440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.593252719045208e-05,
+      "loss": 3.9211,
+      "step": 1469952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.592414124294156e-05,
+      "loss": 3.9293,
+      "step": 1470464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.591575529543104e-05,
+      "loss": 3.9328,
+      "step": 1470976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5907385726724247e-05,
+      "loss": 3.9387,
+      "step": 1471488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5898999779213727e-05,
+      "loss": 3.931,
+      "step": 1472000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5890613831703207e-05,
+      "loss": 3.9261,
+      "step": 1472512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.588222788419269e-05,
+      "loss": 3.941,
+      "step": 1473024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.587384193668217e-05,
+      "loss": 3.9317,
+      "step": 1473536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5865455989171644e-05,
+      "loss": 3.9328,
+      "step": 1474048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5857070041661124e-05,
+      "loss": 3.9244,
+      "step": 1474560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5848684094150604e-05,
+      "loss": 3.9168,
+      "step": 1475072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5840314525443816e-05,
+      "loss": 3.9535,
+      "step": 1475584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5831928577933296e-05,
+      "loss": 3.9428,
+      "step": 1476096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5823542630422776e-05,
+      "loss": 3.9339,
+      "step": 1476608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5815156682912256e-05,
+      "loss": 3.9331,
+      "step": 1477120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5806770735401736e-05,
+      "loss": 3.9358,
+      "step": 1477632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5798384787891216e-05,
+      "loss": 3.9219,
+      "step": 1478144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.579001521918443e-05,
+      "loss": 3.9397,
+      "step": 1478656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.578162927167391e-05,
+      "loss": 3.9247,
+      "step": 1479168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.577324332416339e-05,
+      "loss": 3.9386,
+      "step": 1479680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.576485737665287e-05,
+      "loss": 3.9474,
+      "step": 1480192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.575647142914235e-05,
+      "loss": 3.9281,
+      "step": 1480704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.574808548163183e-05,
+      "loss": 3.9209,
+      "step": 1481216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5739699534121308e-05,
+      "loss": 3.9458,
+      "step": 1481728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5731313586610788e-05,
+      "loss": 3.9115,
+      "step": 1482240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5722944017904e-05,
+      "loss": 3.9272,
+      "step": 1482752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.571455807039348e-05,
+      "loss": 3.9411,
+      "step": 1483264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.570617212288296e-05,
+      "loss": 3.9398,
+      "step": 1483776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.569778617537244e-05,
+      "loss": 3.9311,
+      "step": 1484288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.568941660666565e-05,
+      "loss": 3.9206,
+      "step": 1484800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5681047037958862e-05,
+      "loss": 3.9203,
+      "step": 1485312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5672661090448342e-05,
+      "loss": 3.9218,
+      "step": 1485824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5664275142937822e-05,
+      "loss": 3.9416,
+      "step": 1486336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5655889195427302e-05,
+      "loss": 3.9241,
+      "step": 1486848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5647503247916782e-05,
+      "loss": 3.9369,
+      "step": 1487360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5639117300406262e-05,
+      "loss": 3.9383,
+      "step": 1487872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5630731352895742e-05,
+      "loss": 3.9458,
+      "step": 1488384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5622345405385222e-05,
+      "loss": 3.9307,
+      "step": 1488896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5613975836678434e-05,
+      "loss": 3.9238,
+      "step": 1489408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5605589889167914e-05,
+      "loss": 3.9406,
+      "step": 1489920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5597203941657394e-05,
+      "loss": 3.9304,
+      "step": 1490432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5588817994146874e-05,
+      "loss": 3.9356,
+      "step": 1490944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5580432046636354e-05,
+      "loss": 3.9322,
+      "step": 1491456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5572062477929563e-05,
+      "loss": 3.9306,
+      "step": 1491968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5563676530419047e-05,
+      "loss": 3.9381,
+      "step": 1492480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5555290582908527e-05,
+      "loss": 3.9178,
+      "step": 1492992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5546904635398e-05,
+      "loss": 3.9307,
+      "step": 1493504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5538535066691216e-05,
+      "loss": 3.931,
+      "step": 1494016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5530165497984425e-05,
+      "loss": 3.9368,
+      "step": 1494528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5521779550473908e-05,
+      "loss": 3.9316,
+      "step": 1495040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5513393602963388e-05,
+      "loss": 3.9295,
+      "step": 1495552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5505007655452868e-05,
+      "loss": 3.9258,
+      "step": 1496064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5496621707942348e-05,
+      "loss": 3.9243,
+      "step": 1496576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5488235760431828e-05,
+      "loss": 3.9211,
+      "step": 1497088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.54798498129213e-05,
+      "loss": 3.9332,
+      "step": 1497600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5471463865410784e-05,
+      "loss": 3.9447,
+      "step": 1498112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5463094296704e-05,
+      "loss": 3.9378,
+      "step": 1498624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5454708349193473e-05,
+      "loss": 3.9332,
+      "step": 1499136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5446322401682953e-05,
+      "loss": 3.922,
+      "step": 1499648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5437936454172433e-05,
+      "loss": 3.9283,
+      "step": 1500160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5429550506661913e-05,
+      "loss": 3.9318,
+      "step": 1500672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5421180937955126e-05,
+      "loss": 3.9372,
+      "step": 1501184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5412794990444606e-05,
+      "loss": 3.9363,
+      "step": 1501696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.540442542173782e-05,
+      "loss": 3.9319,
+      "step": 1502208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.53960394742273e-05,
+      "loss": 3.9216,
+      "step": 1502720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5387653526716775e-05,
+      "loss": 3.944,
+      "step": 1503232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5379267579206255e-05,
+      "loss": 3.9188,
+      "step": 1503744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5370881631695738e-05,
+      "loss": 3.9331,
+      "step": 1504256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5362512062988947e-05,
+      "loss": 3.9222,
+      "step": 1504768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5354126115478427e-05,
+      "loss": 3.9329,
+      "step": 1505280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5345740167967907e-05,
+      "loss": 3.9218,
+      "step": 1505792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5337354220457387e-05,
+      "loss": 3.9315,
+      "step": 1506304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5328968272946867e-05,
+      "loss": 3.9262,
+      "step": 1506816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5320582325436347e-05,
+      "loss": 3.9242,
+      "step": 1507328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.531221275672956e-05,
+      "loss": 3.9306,
+      "step": 1507840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.530382680921904e-05,
+      "loss": 3.9228,
+      "step": 1508352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.529544086170852e-05,
+      "loss": 3.9416,
+      "step": 1508864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5287054914198e-05,
+      "loss": 3.9472,
+      "step": 1509376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.527866896668748e-05,
+      "loss": 3.9393,
+      "step": 1509888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.527028301917696e-05,
+      "loss": 3.9247,
+      "step": 1510400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.526189707166644e-05,
+      "loss": 3.9281,
+      "step": 1510912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.525351112415592e-05,
+      "loss": 3.9258,
+      "step": 1511424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.524514155544913e-05,
+      "loss": 3.9277,
+      "step": 1511936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.523675560793861e-05,
+      "loss": 3.9367,
+      "step": 1512448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.522836966042809e-05,
+      "loss": 3.9184,
+      "step": 1512960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.52200000917213e-05,
+      "loss": 3.9334,
+      "step": 1513472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5211614144210784e-05,
+      "loss": 3.9325,
+      "step": 1513984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5203228196700264e-05,
+      "loss": 3.9328,
+      "step": 1514496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5194842249189744e-05,
+      "loss": 3.926,
+      "step": 1515008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5186456301679224e-05,
+      "loss": 3.9304,
+      "step": 1515520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5178070354168704e-05,
+      "loss": 3.9282,
+      "step": 1516032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5169684406658184e-05,
+      "loss": 3.935,
+      "step": 1516544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.516129845914766e-05,
+      "loss": 3.9357,
+      "step": 1517056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5152928890440873e-05,
+      "loss": 3.939,
+      "step": 1517568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5144542942930356e-05,
+      "loss": 3.9203,
+      "step": 1518080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5136156995419836e-05,
+      "loss": 3.9363,
+      "step": 1518592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.512777104790931e-05,
+      "loss": 3.9361,
+      "step": 1519104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.511938510039879e-05,
+      "loss": 3.9295,
+      "step": 1519616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5111015531692005e-05,
+      "loss": 3.9327,
+      "step": 1520128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5102645962985218e-05,
+      "loss": 3.9404,
+      "step": 1520640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5094260015474698e-05,
+      "loss": 3.9329,
+      "step": 1521152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5085874067964177e-05,
+      "loss": 3.9476,
+      "step": 1521664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5077488120453657e-05,
+      "loss": 3.9235,
+      "step": 1522176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.506910217294313e-05,
+      "loss": 3.9355,
+      "step": 1522688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5060732604236346e-05,
+      "loss": 3.9294,
+      "step": 1523200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5052346656725826e-05,
+      "loss": 3.9307,
+      "step": 1523712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.504396070921531e-05,
+      "loss": 3.9281,
+      "step": 1524224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5035574761704783e-05,
+      "loss": 3.9299,
+      "step": 1524736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5027188814194263e-05,
+      "loss": 3.9243,
+      "step": 1525248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5018802866683743e-05,
+      "loss": 3.9254,
+      "step": 1525760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.501043329797696e-05,
+      "loss": 3.9296,
+      "step": 1526272
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.007101535797119,
+      "eval_runtime": 301.593,
+      "eval_samples_per_second": 1265.251,
+      "eval_steps_per_second": 39.54,
+      "step": 1526387
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5002047350466435e-05,
+      "loss": 3.9348,
+      "step": 1526784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.499366140295592e-05,
+      "loss": 3.9245,
+      "step": 1527296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.49852754554454e-05,
+      "loss": 3.9393,
+      "step": 1527808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4976905886738608e-05,
+      "loss": 3.9294,
+      "step": 1528320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4968519939228088e-05,
+      "loss": 3.9464,
+      "step": 1528832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4960133991717568e-05,
+      "loss": 3.9295,
+      "step": 1529344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4951748044207048e-05,
+      "loss": 3.929,
+      "step": 1529856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4943362096696528e-05,
+      "loss": 3.9177,
+      "step": 1530368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4934976149186007e-05,
+      "loss": 3.9343,
+      "step": 1530880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4926590201675487e-05,
+      "loss": 3.9323,
+      "step": 1531392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4918204254164967e-05,
+      "loss": 3.9296,
+      "step": 1531904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4909834685458176e-05,
+      "loss": 3.932,
+      "step": 1532416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4901465116751392e-05,
+      "loss": 3.9237,
+      "step": 1532928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4893079169240872e-05,
+      "loss": 3.9349,
+      "step": 1533440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.488469322173035e-05,
+      "loss": 3.9187,
+      "step": 1533952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.487630727421983e-05,
+      "loss": 3.9218,
+      "step": 1534464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.486792132670931e-05,
+      "loss": 3.9245,
+      "step": 1534976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.485953537919879e-05,
+      "loss": 3.9293,
+      "step": 1535488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.485114943168827e-05,
+      "loss": 3.9279,
+      "step": 1536000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.484276348417775e-05,
+      "loss": 3.9443,
+      "step": 1536512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.483439391547096e-05,
+      "loss": 3.9284,
+      "step": 1537024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.482600796796044e-05,
+      "loss": 3.9383,
+      "step": 1537536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.481762202044992e-05,
+      "loss": 3.9272,
+      "step": 1538048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.48092360729394e-05,
+      "loss": 3.9375,
+      "step": 1538560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4800850125428878e-05,
+      "loss": 3.9281,
+      "step": 1539072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4792496935525823e-05,
+      "loss": 3.933,
+      "step": 1539584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4784127366819035e-05,
+      "loss": 3.9217,
+      "step": 1540096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4775741419308515e-05,
+      "loss": 3.9254,
+      "step": 1540608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4767355471797995e-05,
+      "loss": 3.9223,
+      "step": 1541120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.475896952428747e-05,
+      "loss": 3.9274,
+      "step": 1541632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4750583576776955e-05,
+      "loss": 3.9362,
+      "step": 1542144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4742197629266435e-05,
+      "loss": 3.9322,
+      "step": 1542656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4733811681755915e-05,
+      "loss": 3.9403,
+      "step": 1543168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4725442113049124e-05,
+      "loss": 3.9294,
+      "step": 1543680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4717072544342336e-05,
+      "loss": 3.9311,
+      "step": 1544192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4708686596831816e-05,
+      "loss": 3.9357,
+      "step": 1544704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4700300649321296e-05,
+      "loss": 3.9333,
+      "step": 1545216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4691914701810776e-05,
+      "loss": 3.9157,
+      "step": 1545728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4683528754300256e-05,
+      "loss": 3.9178,
+      "step": 1546240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4675142806789736e-05,
+      "loss": 3.9188,
+      "step": 1546752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4666773238082945e-05,
+      "loss": 3.9255,
+      "step": 1547264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4658387290572425e-05,
+      "loss": 3.9335,
+      "step": 1547776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.465000134306191e-05,
+      "loss": 3.9255,
+      "step": 1548288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.464161539555139e-05,
+      "loss": 3.9182,
+      "step": 1548800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.463322944804087e-05,
+      "loss": 3.9271,
+      "step": 1549312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.462484350053035e-05,
+      "loss": 3.9264,
+      "step": 1549824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4616457553019825e-05,
+      "loss": 3.9255,
+      "step": 1550336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4608071605509305e-05,
+      "loss": 3.9176,
+      "step": 1550848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4599685657998785e-05,
+      "loss": 3.9052,
+      "step": 1551360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4591299710488265e-05,
+      "loss": 3.946,
+      "step": 1551872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4582913762977748e-05,
+      "loss": 3.9323,
+      "step": 1552384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4574527815467228e-05,
+      "loss": 3.9301,
+      "step": 1552896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4566158246760437e-05,
+      "loss": 3.9273,
+      "step": 1553408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4557772299249917e-05,
+      "loss": 3.9234,
+      "step": 1553920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4549386351739397e-05,
+      "loss": 3.9183,
+      "step": 1554432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4541000404228877e-05,
+      "loss": 3.9286,
+      "step": 1554944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4532614456718357e-05,
+      "loss": 3.9198,
+      "step": 1555456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4524228509207837e-05,
+      "loss": 3.9269,
+      "step": 1555968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4515842561697317e-05,
+      "loss": 3.9437,
+      "step": 1556480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4507456614186797e-05,
+      "loss": 3.919,
+      "step": 1556992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4499070666676277e-05,
+      "loss": 3.9123,
+      "step": 1557504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4490701097969486e-05,
+      "loss": 3.9367,
+      "step": 1558016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4482331529262702e-05,
+      "loss": 3.9075,
+      "step": 1558528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4473945581752182e-05,
+      "loss": 3.9178,
+      "step": 1559040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.446555963424166e-05,
+      "loss": 3.9307,
+      "step": 1559552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.445717368673114e-05,
+      "loss": 3.9287,
+      "step": 1560064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.444878773922062e-05,
+      "loss": 3.9241,
+      "step": 1560576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4440401791710098e-05,
+      "loss": 3.9174,
+      "step": 1561088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.443203222300331e-05,
+      "loss": 3.9072,
+      "step": 1561600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.442364627549279e-05,
+      "loss": 3.9149,
+      "step": 1562112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.441526032798227e-05,
+      "loss": 3.9344,
+      "step": 1562624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.440687438047175e-05,
+      "loss": 3.9175,
+      "step": 1563136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.439848843296123e-05,
+      "loss": 3.928,
+      "step": 1563648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.439010248545071e-05,
+      "loss": 3.9326,
+      "step": 1564160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4381716537940187e-05,
+      "loss": 3.9413,
+      "step": 1564672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.437333059042967e-05,
+      "loss": 3.9193,
+      "step": 1565184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4364961021722883e-05,
+      "loss": 3.9176,
+      "step": 1565696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4356575074212363e-05,
+      "loss": 3.9332,
+      "step": 1566208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.434818912670184e-05,
+      "loss": 3.9183,
+      "step": 1566720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.433980317919132e-05,
+      "loss": 3.9334,
+      "step": 1567232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.43314172316808e-05,
+      "loss": 3.9225,
+      "step": 1567744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4323064041777744e-05,
+      "loss": 3.9236,
+      "step": 1568256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4314678094267224e-05,
+      "loss": 3.9298,
+      "step": 1568768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4306292146756704e-05,
+      "loss": 3.9131,
+      "step": 1569280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4297906199246184e-05,
+      "loss": 3.9184,
+      "step": 1569792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.428952025173566e-05,
+      "loss": 3.924,
+      "step": 1570304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4281150683028873e-05,
+      "loss": 3.9306,
+      "step": 1570816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4272764735518357e-05,
+      "loss": 3.9203,
+      "step": 1571328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.4264378788007837e-05,
+      "loss": 3.9248,
+      "step": 1571840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4255992840497313e-05,
+      "loss": 3.917,
+      "step": 1572352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4247606892986793e-05,
+      "loss": 3.9187,
+      "step": 1572864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4239220945476273e-05,
+      "loss": 3.9131,
+      "step": 1573376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4230851376769486e-05,
+      "loss": 3.9227,
+      "step": 1573888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4222465429258966e-05,
+      "loss": 3.9384,
+      "step": 1574400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4214079481748445e-05,
+      "loss": 3.9312,
+      "step": 1574912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4205709913041658e-05,
+      "loss": 3.925,
+      "step": 1575424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4197323965531135e-05,
+      "loss": 3.9143,
+      "step": 1575936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4188938018020614e-05,
+      "loss": 3.9198,
+      "step": 1576448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4180552070510094e-05,
+      "loss": 3.9222,
+      "step": 1576960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4172166122999574e-05,
+      "loss": 3.9314,
+      "step": 1577472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4163780175489058e-05,
+      "loss": 3.9283,
+      "step": 1577984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4155394227978538e-05,
+      "loss": 3.9233,
+      "step": 1578496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4147008280468018e-05,
+      "loss": 3.9159,
+      "step": 1579008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4138622332957494e-05,
+      "loss": 3.9356,
+      "step": 1579520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4130236385446974e-05,
+      "loss": 3.9127,
+      "step": 1580032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4121850437936454e-05,
+      "loss": 3.9251,
+      "step": 1580544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4113480869229667e-05,
+      "loss": 3.9184,
+      "step": 1581056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4105094921719147e-05,
+      "loss": 3.9228,
+      "step": 1581568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4096708974208627e-05,
+      "loss": 3.9147,
+      "step": 1582080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4088323026698106e-05,
+      "loss": 3.9272,
+      "step": 1582592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4079937079187586e-05,
+      "loss": 3.9168,
+      "step": 1583104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4071551131677066e-05,
+      "loss": 3.9209,
+      "step": 1583616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4063165184166546e-05,
+      "loss": 3.9207,
+      "step": 1584128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4054779236656026e-05,
+      "loss": 3.92,
+      "step": 1584640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4046393289145506e-05,
+      "loss": 3.9308,
+      "step": 1585152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4038040099242448e-05,
+      "loss": 3.9408,
+      "step": 1585664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4029654151731928e-05,
+      "loss": 3.9271,
+      "step": 1586176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4021268204221408e-05,
+      "loss": 3.9202,
+      "step": 1586688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4012882256710888e-05,
+      "loss": 3.92,
+      "step": 1587200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.400449630920037e-05,
+      "loss": 3.918,
+      "step": 1587712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3996110361689848e-05,
+      "loss": 3.918,
+      "step": 1588224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.398774079298306e-05,
+      "loss": 3.9318,
+      "step": 1588736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.397935484547254e-05,
+      "loss": 3.9089,
+      "step": 1589248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.397096889796202e-05,
+      "loss": 3.9261,
+      "step": 1589760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3962582950451497e-05,
+      "loss": 3.9218,
+      "step": 1590272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.395419700294098e-05,
+      "loss": 3.9288,
+      "step": 1590784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3945827434234192e-05,
+      "loss": 3.9179,
+      "step": 1591296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.393744148672367e-05,
+      "loss": 3.925,
+      "step": 1591808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.392905553921315e-05,
+      "loss": 3.9167,
+      "step": 1592320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.392066959170263e-05,
+      "loss": 3.9317,
+      "step": 1592832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.391228364419211e-05,
+      "loss": 3.9266,
+      "step": 1593344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.390389769668159e-05,
+      "loss": 3.9341,
+      "step": 1593856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3895511749171072e-05,
+      "loss": 3.9119,
+      "step": 1594368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.388714218046428e-05,
+      "loss": 3.9277,
+      "step": 1594880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.387875623295376e-05,
+      "loss": 3.9317,
+      "step": 1595392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.387037028544324e-05,
+      "loss": 3.92,
+      "step": 1595904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.386198433793272e-05,
+      "loss": 3.9248,
+      "step": 1596416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.38535983904222e-05,
+      "loss": 3.9348,
+      "step": 1596928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3845228821715414e-05,
+      "loss": 3.9268,
+      "step": 1597440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3836842874204894e-05,
+      "loss": 3.942,
+      "step": 1597952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3828456926694374e-05,
+      "loss": 3.9146,
+      "step": 1598464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.382007097918385e-05,
+      "loss": 3.9286,
+      "step": 1598976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.381168503167333e-05,
+      "loss": 3.9212,
+      "step": 1599488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3803315462966543e-05,
+      "loss": 3.924,
+      "step": 1600000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3794929515456026e-05,
+      "loss": 3.9266,
+      "step": 1600512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3786559946749235e-05,
+      "loss": 3.9179,
+      "step": 1601024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3778173999238715e-05,
+      "loss": 3.9173,
+      "step": 1601536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3769788051728195e-05,
+      "loss": 3.9134,
+      "step": 1602048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3761402104217675e-05,
+      "loss": 3.9249,
+      "step": 1602560
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.004420280456543,
+      "eval_runtime": 301.4626,
+      "eval_samples_per_second": 1265.799,
+      "eval_steps_per_second": 39.557,
+      "step": 1602707
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.375301615670715e-05,
+      "loss": 3.9263,
+      "step": 1603072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3744630209196635e-05,
+      "loss": 3.918,
+      "step": 1603584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3736244261686115e-05,
+      "loss": 3.9287,
+      "step": 1604096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3727858314175595e-05,
+      "loss": 3.9272,
+      "step": 1604608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3719472366665075e-05,
+      "loss": 3.9334,
+      "step": 1605120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3711086419154555e-05,
+      "loss": 3.9244,
+      "step": 1605632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.370270047164403e-05,
+      "loss": 3.9198,
+      "step": 1606144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.369431452413351e-05,
+      "loss": 3.914,
+      "step": 1606656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.368592857662299e-05,
+      "loss": 3.9256,
+      "step": 1607168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3677542629112474e-05,
+      "loss": 3.9236,
+      "step": 1607680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3669156681601954e-05,
+      "loss": 3.9251,
+      "step": 1608192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3660770734091434e-05,
+      "loss": 3.9238,
+      "step": 1608704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3652401165384643e-05,
+      "loss": 3.9217,
+      "step": 1609216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3644015217874123e-05,
+      "loss": 3.9229,
+      "step": 1609728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3635629270363603e-05,
+      "loss": 3.9153,
+      "step": 1610240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3627243322853083e-05,
+      "loss": 3.9092,
+      "step": 1610752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3618857375342567e-05,
+      "loss": 3.9171,
+      "step": 1611264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3610471427832043e-05,
+      "loss": 3.9222,
+      "step": 1611776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3602085480321523e-05,
+      "loss": 3.9221,
+      "step": 1612288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3593699532811003e-05,
+      "loss": 3.9373,
+      "step": 1612800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3585329964104212e-05,
+      "loss": 3.924,
+      "step": 1613312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3576944016593696e-05,
+      "loss": 3.9303,
+      "step": 1613824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3568558069083175e-05,
+      "loss": 3.9181,
+      "step": 1614336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3560172121572655e-05,
+      "loss": 3.9286,
+      "step": 1614848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3551786174062135e-05,
+      "loss": 3.9193,
+      "step": 1615360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3543416605355344e-05,
+      "loss": 3.9277,
+      "step": 1615872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3535047036648557e-05,
+      "loss": 3.9187,
+      "step": 1616384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3526661089138037e-05,
+      "loss": 3.9167,
+      "step": 1616896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3518275141627517e-05,
+      "loss": 3.9158,
+      "step": 1617408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3509889194116997e-05,
+      "loss": 3.9229,
+      "step": 1617920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3501503246606477e-05,
+      "loss": 3.9243,
+      "step": 1618432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3493117299095957e-05,
+      "loss": 3.9269,
+      "step": 1618944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3484731351585437e-05,
+      "loss": 3.9301,
+      "step": 1619456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3476345404074917e-05,
+      "loss": 3.9231,
+      "step": 1619968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3467959456564397e-05,
+      "loss": 3.9206,
+      "step": 1620480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.345958988785761e-05,
+      "loss": 3.9328,
+      "step": 1620992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.345120394034709e-05,
+      "loss": 3.9245,
+      "step": 1621504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.344281799283657e-05,
+      "loss": 3.9102,
+      "step": 1622016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3434432045326046e-05,
+      "loss": 3.9109,
+      "step": 1622528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3426046097815526e-05,
+      "loss": 3.9118,
+      "step": 1623040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3417676529108738e-05,
+      "loss": 3.914,
+      "step": 1623552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3409290581598218e-05,
+      "loss": 3.9294,
+      "step": 1624064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3400904634087698e-05,
+      "loss": 3.9188,
+      "step": 1624576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3392518686577178e-05,
+      "loss": 3.9126,
+      "step": 1625088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.338414911787039e-05,
+      "loss": 3.9233,
+      "step": 1625600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3375763170359867e-05,
+      "loss": 3.9164,
+      "step": 1626112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3367393601653083e-05,
+      "loss": 3.9218,
+      "step": 1626624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3359007654142563e-05,
+      "loss": 3.9105,
+      "step": 1627136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3350621706632043e-05,
+      "loss": 3.8948,
+      "step": 1627648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.334223575912152e-05,
+      "loss": 3.9437,
+      "step": 1628160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3333849811611e-05,
+      "loss": 3.9227,
+      "step": 1628672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.332546386410048e-05,
+      "loss": 3.9265,
+      "step": 1629184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.331707791658996e-05,
+      "loss": 3.916,
+      "step": 1629696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3308691969079442e-05,
+      "loss": 3.919,
+      "step": 1630208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3300306021568922e-05,
+      "loss": 3.9123,
+      "step": 1630720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.32919200740584e-05,
+      "loss": 3.9182,
+      "step": 1631232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.328355050535161e-05,
+      "loss": 3.9145,
+      "step": 1631744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.327516455784109e-05,
+      "loss": 3.9198,
+      "step": 1632256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.326677861033057e-05,
+      "loss": 3.9351,
+      "step": 1632768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.325839266282005e-05,
+      "loss": 3.9133,
+      "step": 1633280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.325000671530953e-05,
+      "loss": 3.9035,
+      "step": 1633792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.324162076779901e-05,
+      "loss": 3.933,
+      "step": 1634304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3233251199092224e-05,
+      "loss": 3.9006,
+      "step": 1634816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.32248652515817e-05,
+      "loss": 3.9072,
+      "step": 1635328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.321647930407118e-05,
+      "loss": 3.9253,
+      "step": 1635840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.320809335656066e-05,
+      "loss": 3.927,
+      "step": 1636352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3199707409050144e-05,
+      "loss": 3.9138,
+      "step": 1636864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3191321461539624e-05,
+      "loss": 3.9145,
+      "step": 1637376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3182935514029103e-05,
+      "loss": 3.8964,
+      "step": 1637888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.317454956651858e-05,
+      "loss": 3.9124,
+      "step": 1638400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.316619637661552e-05,
+      "loss": 3.9264,
+      "step": 1638912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3157810429105005e-05,
+      "loss": 3.9096,
+      "step": 1639424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3149424481594485e-05,
+      "loss": 3.919,
+      "step": 1639936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3141038534083965e-05,
+      "loss": 3.9236,
+      "step": 1640448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3132652586573445e-05,
+      "loss": 3.9345,
+      "step": 1640960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3124266639062925e-05,
+      "loss": 3.9119,
+      "step": 1641472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3115897070356134e-05,
+      "loss": 3.9143,
+      "step": 1641984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3107511122845614e-05,
+      "loss": 3.9251,
+      "step": 1642496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3099125175335097e-05,
+      "loss": 3.9089,
+      "step": 1643008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3090739227824577e-05,
+      "loss": 3.9244,
+      "step": 1643520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3082353280314054e-05,
+      "loss": 3.9154,
+      "step": 1644032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3073967332803534e-05,
+      "loss": 3.9187,
+      "step": 1644544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3065597764096746e-05,
+      "loss": 3.9236,
+      "step": 1645056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3057211816586226e-05,
+      "loss": 3.9087,
+      "step": 1645568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3048825869075706e-05,
+      "loss": 3.9081,
+      "step": 1646080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3040439921565186e-05,
+      "loss": 3.917,
+      "step": 1646592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3032053974054666e-05,
+      "loss": 3.9235,
+      "step": 1647104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3023668026544146e-05,
+      "loss": 3.9132,
+      "step": 1647616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3015282079033626e-05,
+      "loss": 3.919,
+      "step": 1648128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3006896131523106e-05,
+      "loss": 3.9081,
+      "step": 1648640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2998526562816315e-05,
+      "loss": 3.9103,
+      "step": 1649152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2990156994109527e-05,
+      "loss": 3.9068,
+      "step": 1649664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2981771046599007e-05,
+      "loss": 3.9159,
+      "step": 1650176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2973385099088487e-05,
+      "loss": 3.9305,
+      "step": 1650688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.29650155303817e-05,
+      "loss": 3.9219,
+      "step": 1651200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2956629582871176e-05,
+      "loss": 3.92,
+      "step": 1651712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.294824363536066e-05,
+      "loss": 3.908,
+      "step": 1652224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.293985768785014e-05,
+      "loss": 3.9128,
+      "step": 1652736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.293147174033962e-05,
+      "loss": 3.9157,
+      "step": 1653248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.29230857928291e-05,
+      "loss": 3.9257,
+      "step": 1653760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.291469984531858e-05,
+      "loss": 3.9186,
+      "step": 1654272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.290631389780806e-05,
+      "loss": 3.9161,
+      "step": 1654784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.289794432910127e-05,
+      "loss": 3.9157,
+      "step": 1655296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2889558381590752e-05,
+      "loss": 3.9285,
+      "step": 1655808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2881172434080232e-05,
+      "loss": 3.9023,
+      "step": 1656320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.287278648656971e-05,
+      "loss": 3.9173,
+      "step": 1656832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.286440053905919e-05,
+      "loss": 3.9068,
+      "step": 1657344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.285601459154867e-05,
+      "loss": 3.9198,
+      "step": 1657856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.284762864403815e-05,
+      "loss": 3.9102,
+      "step": 1658368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.283924269652763e-05,
+      "loss": 3.9152,
+      "step": 1658880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2830856749017108e-05,
+      "loss": 3.9094,
+      "step": 1659392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.282248718031032e-05,
+      "loss": 3.9136,
+      "step": 1659904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2814117611603533e-05,
+      "loss": 3.9135,
+      "step": 1660416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.280573166409301e-05,
+      "loss": 3.9117,
+      "step": 1660928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2797362095386222e-05,
+      "loss": 3.9222,
+      "step": 1661440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2788976147875706e-05,
+      "loss": 3.9326,
+      "step": 1661952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2780590200365182e-05,
+      "loss": 3.922,
+      "step": 1662464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2772204252854662e-05,
+      "loss": 3.9161,
+      "step": 1662976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2763818305344142e-05,
+      "loss": 3.9103,
+      "step": 1663488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2755432357833622e-05,
+      "loss": 3.9098,
+      "step": 1664000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2747046410323102e-05,
+      "loss": 3.9177,
+      "step": 1664512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2738660462812582e-05,
+      "loss": 3.9243,
+      "step": 1665024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2730274515302062e-05,
+      "loss": 3.8985,
+      "step": 1665536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2721904946595274e-05,
+      "loss": 3.9168,
+      "step": 1666048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2713518999084754e-05,
+      "loss": 3.9163,
+      "step": 1666560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2705133051574234e-05,
+      "loss": 3.9177,
+      "step": 1667072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.269674710406371e-05,
+      "loss": 3.9132,
+      "step": 1667584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.268836115655319e-05,
+      "loss": 3.9137,
+      "step": 1668096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2679975209042674e-05,
+      "loss": 3.9144,
+      "step": 1668608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2671589261532154e-05,
+      "loss": 3.9214,
+      "step": 1669120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2663219692825363e-05,
+      "loss": 3.9199,
+      "step": 1669632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2654833745314843e-05,
+      "loss": 3.927,
+      "step": 1670144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2646447797804323e-05,
+      "loss": 3.9059,
+      "step": 1670656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2638061850293803e-05,
+      "loss": 3.922,
+      "step": 1671168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2629675902783283e-05,
+      "loss": 3.9234,
+      "step": 1671680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2621289955272763e-05,
+      "loss": 3.914,
+      "step": 1672192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2612904007762246e-05,
+      "loss": 3.9195,
+      "step": 1672704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2604518060251723e-05,
+      "loss": 3.9259,
+      "step": 1673216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2596148491544935e-05,
+      "loss": 3.9156,
+      "step": 1673728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2587778922838145e-05,
+      "loss": 3.9371,
+      "step": 1674240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2579392975327628e-05,
+      "loss": 3.9123,
+      "step": 1674752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2571007027817108e-05,
+      "loss": 3.9163,
+      "step": 1675264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2562621080306588e-05,
+      "loss": 3.9112,
+      "step": 1675776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2554251511599797e-05,
+      "loss": 3.9187,
+      "step": 1676288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2545865564089277e-05,
+      "loss": 3.914,
+      "step": 1676800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2537479616578757e-05,
+      "loss": 3.9118,
+      "step": 1677312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2529093669068237e-05,
+      "loss": 3.9157,
+      "step": 1677824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2520707721557717e-05,
+      "loss": 3.9015,
+      "step": 1678336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.251233815285093e-05,
+      "loss": 3.9193,
+      "step": 1678848
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.001370906829834,
+      "eval_runtime": 314.685,
+      "eval_samples_per_second": 1212.612,
+      "eval_steps_per_second": 37.895,
+      "step": 1679027
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.250395220534041e-05,
+      "loss": 3.9223,
+      "step": 1679360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.249556625782989e-05,
+      "loss": 3.9116,
+      "step": 1679872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2487180310319366e-05,
+      "loss": 3.9177,
+      "step": 1680384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2478810741612578e-05,
+      "loss": 3.9183,
+      "step": 1680896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.247042479410206e-05,
+      "loss": 3.9289,
+      "step": 1681408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.246205522539527e-05,
+      "loss": 3.9129,
+      "step": 1681920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.245366927788475e-05,
+      "loss": 3.9184,
+      "step": 1682432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.244528333037423e-05,
+      "loss": 3.902,
+      "step": 1682944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.243689738286371e-05,
+      "loss": 3.9175,
+      "step": 1683456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.242851143535319e-05,
+      "loss": 3.9188,
+      "step": 1683968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.242012548784267e-05,
+      "loss": 3.9163,
+      "step": 1684480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.241173954033215e-05,
+      "loss": 3.9179,
+      "step": 1684992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2403369971625363e-05,
+      "loss": 3.9115,
+      "step": 1685504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.239498402411484e-05,
+      "loss": 3.9193,
+      "step": 1686016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.238659807660432e-05,
+      "loss": 3.9087,
+      "step": 1686528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.23782121290938e-05,
+      "loss": 3.9048,
+      "step": 1687040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2369826181583283e-05,
+      "loss": 3.9114,
+      "step": 1687552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2361440234072763e-05,
+      "loss": 3.9129,
+      "step": 1688064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2353054286562243e-05,
+      "loss": 3.9156,
+      "step": 1688576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2344668339051723e-05,
+      "loss": 3.9245,
+      "step": 1689088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.23362823915412e-05,
+      "loss": 3.9193,
+      "step": 1689600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.232789644403068e-05,
+      "loss": 3.9268,
+      "step": 1690112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.231951049652016e-05,
+      "loss": 3.9087,
+      "step": 1690624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.231112454900964e-05,
+      "loss": 3.9186,
+      "step": 1691136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2302738601499122e-05,
+      "loss": 3.9167,
+      "step": 1691648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.229436903279233e-05,
+      "loss": 3.9211,
+      "step": 1692160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.228598308528181e-05,
+      "loss": 3.9074,
+      "step": 1692672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.227759713777129e-05,
+      "loss": 3.9132,
+      "step": 1693184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.226921119026077e-05,
+      "loss": 3.9099,
+      "step": 1693696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.226082524275025e-05,
+      "loss": 3.9122,
+      "step": 1694208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2252455674043464e-05,
+      "loss": 3.9168,
+      "step": 1694720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2244069726532944e-05,
+      "loss": 3.9208,
+      "step": 1695232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2235683779022424e-05,
+      "loss": 3.9221,
+      "step": 1695744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2227314210315633e-05,
+      "loss": 3.9187,
+      "step": 1696256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2218928262805113e-05,
+      "loss": 3.9164,
+      "step": 1696768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2210558694098325e-05,
+      "loss": 3.9207,
+      "step": 1697280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2202172746587805e-05,
+      "loss": 3.9177,
+      "step": 1697792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2193786799077285e-05,
+      "loss": 3.9053,
+      "step": 1698304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2185400851566765e-05,
+      "loss": 3.9034,
+      "step": 1698816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2177014904056245e-05,
+      "loss": 3.9027,
+      "step": 1699328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2168628956545725e-05,
+      "loss": 3.909,
+      "step": 1699840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.21602430090352e-05,
+      "loss": 3.9184,
+      "step": 1700352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2151857061524685e-05,
+      "loss": 3.9109,
+      "step": 1700864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2143471114014165e-05,
+      "loss": 3.9078,
+      "step": 1701376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2135085166503645e-05,
+      "loss": 3.9119,
+      "step": 1701888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2126699218993125e-05,
+      "loss": 3.9134,
+      "step": 1702400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2118329650286334e-05,
+      "loss": 3.9112,
+      "step": 1702912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2109943702775814e-05,
+      "loss": 3.9049,
+      "step": 1703424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2101557755265294e-05,
+      "loss": 3.8875,
+      "step": 1703936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2093171807754777e-05,
+      "loss": 3.9339,
+      "step": 1704448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2084785860244257e-05,
+      "loss": 3.9154,
+      "step": 1704960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2076399912733734e-05,
+      "loss": 3.9215,
+      "step": 1705472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2068013965223214e-05,
+      "loss": 3.9088,
+      "step": 1705984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2059628017712693e-05,
+      "loss": 3.9125,
+      "step": 1706496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2051242070202173e-05,
+      "loss": 3.9069,
+      "step": 1707008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.204288888029912e-05,
+      "loss": 3.911,
+      "step": 1707520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.20345029327886e-05,
+      "loss": 3.9127,
+      "step": 1708032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.202611698527808e-05,
+      "loss": 3.9083,
+      "step": 1708544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.201773103776756e-05,
+      "loss": 3.9309,
+      "step": 1709056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2009345090257035e-05,
+      "loss": 3.9051,
+      "step": 1709568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2000959142746515e-05,
+      "loss": 3.8987,
+      "step": 1710080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.199258957403973e-05,
+      "loss": 3.9241,
+      "step": 1710592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1984203626529207e-05,
+      "loss": 3.8947,
+      "step": 1711104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1975817679018687e-05,
+      "loss": 3.8985,
+      "step": 1711616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1967431731508167e-05,
+      "loss": 3.9206,
+      "step": 1712128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1959045783997647e-05,
+      "loss": 3.9196,
+      "step": 1712640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1950659836487127e-05,
+      "loss": 3.9008,
+      "step": 1713152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.194229026778034e-05,
+      "loss": 3.9109,
+      "step": 1713664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.193390432026982e-05,
+      "loss": 3.8891,
+      "step": 1714176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.19255183727593e-05,
+      "loss": 3.9071,
+      "step": 1714688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.191713242524878e-05,
+      "loss": 3.9132,
+      "step": 1715200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.190874647773826e-05,
+      "loss": 3.9054,
+      "step": 1715712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.190036053022774e-05,
+      "loss": 3.9112,
+      "step": 1716224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1891974582717216e-05,
+      "loss": 3.9158,
+      "step": 1716736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.18835886352067e-05,
+      "loss": 3.9273,
+      "step": 1717248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1875219066499912e-05,
+      "loss": 3.9067,
+      "step": 1717760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.186683311898939e-05,
+      "loss": 3.9053,
+      "step": 1718272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.18584635502826e-05,
+      "loss": 3.9223,
+      "step": 1718784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.185007760277208e-05,
+      "loss": 3.9042,
+      "step": 1719296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.184169165526156e-05,
+      "loss": 3.9145,
+      "step": 1719808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.183330570775104e-05,
+      "loss": 3.912,
+      "step": 1720320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.182491976024052e-05,
+      "loss": 3.9076,
+      "step": 1720832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.181653381273e-05,
+      "loss": 3.9215,
+      "step": 1721344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.180814786521948e-05,
+      "loss": 3.8992,
+      "step": 1721856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.179977829651269e-05,
+      "loss": 3.9008,
+      "step": 1722368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.179139234900217e-05,
+      "loss": 3.913,
+      "step": 1722880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1783006401491653e-05,
+      "loss": 3.9151,
+      "step": 1723392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1774620453981133e-05,
+      "loss": 3.9071,
+      "step": 1723904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.1766250885274342e-05,
+      "loss": 3.9115,
+      "step": 1724416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1757864937763822e-05,
+      "loss": 3.8967,
+      "step": 1724928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1749478990253302e-05,
+      "loss": 3.909,
+      "step": 1725440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1741093042742782e-05,
+      "loss": 3.8982,
+      "step": 1725952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1732723474035994e-05,
+      "loss": 3.9074,
+      "step": 1726464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1724337526525474e-05,
+      "loss": 3.927,
+      "step": 1726976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1715951579014954e-05,
+      "loss": 3.9115,
+      "step": 1727488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1707565631504434e-05,
+      "loss": 3.9124,
+      "step": 1728000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1699179683993914e-05,
+      "loss": 3.904,
+      "step": 1728512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1690793736483394e-05,
+      "loss": 3.9033,
+      "step": 1729024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.168240778897287e-05,
+      "loss": 3.9067,
+      "step": 1729536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1674021841462354e-05,
+      "loss": 3.9183,
+      "step": 1730048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1665635893951834e-05,
+      "loss": 3.9132,
+      "step": 1730560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1657266325245043e-05,
+      "loss": 3.9156,
+      "step": 1731072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1648880377734523e-05,
+      "loss": 3.904,
+      "step": 1731584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1640494430224003e-05,
+      "loss": 3.9166,
+      "step": 1732096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1632108482713483e-05,
+      "loss": 3.9021,
+      "step": 1732608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1623722535202963e-05,
+      "loss": 3.9127,
+      "step": 1733120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1615336587692446e-05,
+      "loss": 3.8958,
+      "step": 1733632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1606950640181926e-05,
+      "loss": 3.9145,
+      "step": 1734144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1598581071475135e-05,
+      "loss": 3.9029,
+      "step": 1734656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1590195123964615e-05,
+      "loss": 3.9099,
+      "step": 1735168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1581809176454095e-05,
+      "loss": 3.9021,
+      "step": 1735680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1573423228943572e-05,
+      "loss": 3.9101,
+      "step": 1736192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1565053660236788e-05,
+      "loss": 3.9046,
+      "step": 1736704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1556684091529997e-05,
+      "loss": 3.9045,
+      "step": 1737216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1548298144019477e-05,
+      "loss": 3.9174,
+      "step": 1737728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1539912196508957e-05,
+      "loss": 3.9246,
+      "step": 1738240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1531526248998437e-05,
+      "loss": 3.9163,
+      "step": 1738752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1523140301487917e-05,
+      "loss": 3.9102,
+      "step": 1739264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.15147543539774e-05,
+      "loss": 3.9042,
+      "step": 1739776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1506368406466877e-05,
+      "loss": 3.9045,
+      "step": 1740288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1497982458956356e-05,
+      "loss": 3.9107,
+      "step": 1740800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1489596511445836e-05,
+      "loss": 3.9154,
+      "step": 1741312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1481210563935316e-05,
+      "loss": 3.8942,
+      "step": 1741824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1472824616424796e-05,
+      "loss": 3.912,
+      "step": 1742336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1464438668914276e-05,
+      "loss": 3.9117,
+      "step": 1742848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.145606910020749e-05,
+      "loss": 3.9105,
+      "step": 1743360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.144768315269697e-05,
+      "loss": 3.9039,
+      "step": 1743872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.143929720518645e-05,
+      "loss": 3.9119,
+      "step": 1744384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.143091125767593e-05,
+      "loss": 3.9034,
+      "step": 1744896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1422541688969138e-05,
+      "loss": 3.9194,
+      "step": 1745408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1414155741458618e-05,
+      "loss": 3.9129,
+      "step": 1745920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.14057697939481e-05,
+      "loss": 3.9144,
+      "step": 1746432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.139738384643758e-05,
+      "loss": 3.9012,
+      "step": 1746944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1388997898927058e-05,
+      "loss": 3.9189,
+      "step": 1747456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1380611951416538e-05,
+      "loss": 3.9165,
+      "step": 1747968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1372226003906017e-05,
+      "loss": 3.9035,
+      "step": 1748480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1363856435199227e-05,
+      "loss": 3.9135,
+      "step": 1748992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.135547048768871e-05,
+      "loss": 3.9167,
+      "step": 1749504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.134708454017819e-05,
+      "loss": 3.9129,
+      "step": 1750016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.133869859266767e-05,
+      "loss": 3.9291,
+      "step": 1750528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.133031264515715e-05,
+      "loss": 3.9026,
+      "step": 1751040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.132192669764663e-05,
+      "loss": 3.9145,
+      "step": 1751552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.131354075013611e-05,
+      "loss": 3.9069,
+      "step": 1752064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.130517118142932e-05,
+      "loss": 3.9132,
+      "step": 1752576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1296785233918802e-05,
+      "loss": 3.9062,
+      "step": 1753088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1288399286408282e-05,
+      "loss": 3.9052,
+      "step": 1753600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.1280013338897762e-05,
+      "loss": 3.9105,
+      "step": 1754112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.127162739138724e-05,
+      "loss": 3.8958,
+      "step": 1754624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.126324144387672e-05,
+      "loss": 3.9168,
+      "step": 1755136
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9990670680999756,
+      "eval_runtime": 305.6579,
+      "eval_samples_per_second": 1248.425,
+      "eval_steps_per_second": 39.014,
+      "step": 1755347
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.12548554963662e-05,
+      "loss": 3.9158,
+      "step": 1755648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.124646954885568e-05,
+      "loss": 3.9046,
+      "step": 1756160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.123808360134516e-05,
+      "loss": 3.9142,
+      "step": 1756672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1229697653834642e-05,
+      "loss": 3.9115,
+      "step": 1757184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.122131170632412e-05,
+      "loss": 3.9187,
+      "step": 1757696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1212925758813598e-05,
+      "loss": 3.9123,
+      "step": 1758208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1204539811303078e-05,
+      "loss": 3.9127,
+      "step": 1758720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1196153863792558e-05,
+      "loss": 3.8992,
+      "step": 1759232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1187767916282038e-05,
+      "loss": 3.9055,
+      "step": 1759744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1179381968771518e-05,
+      "loss": 3.9153,
+      "step": 1760256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1170996021261e-05,
+      "loss": 3.9077,
+      "step": 1760768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1162610073750478e-05,
+      "loss": 3.9072,
+      "step": 1761280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.115424050504369e-05,
+      "loss": 3.9112,
+      "step": 1761792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.114585455753317e-05,
+      "loss": 3.9075,
+      "step": 1762304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.113746861002265e-05,
+      "loss": 3.9019,
+      "step": 1762816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1129082662512127e-05,
+      "loss": 3.8989,
+      "step": 1763328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.112069671500161e-05,
+      "loss": 3.9032,
+      "step": 1763840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.111231076749109e-05,
+      "loss": 3.9113,
+      "step": 1764352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.110392481998057e-05,
+      "loss": 3.9026,
+      "step": 1764864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.109553887247005e-05,
+      "loss": 3.9217,
+      "step": 1765376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.108715292495953e-05,
+      "loss": 3.9165,
+      "step": 1765888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1078766977449007e-05,
+      "loss": 3.9198,
+      "step": 1766400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1070381029938487e-05,
+      "loss": 3.9045,
+      "step": 1766912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1062011461231703e-05,
+      "loss": 3.9108,
+      "step": 1767424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1053625513721183e-05,
+      "loss": 3.9126,
+      "step": 1767936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.104523956621066e-05,
+      "loss": 3.9122,
+      "step": 1768448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.103685361870014e-05,
+      "loss": 3.9043,
+      "step": 1768960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.102846767118962e-05,
+      "loss": 3.9068,
+      "step": 1769472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.10200817236791e-05,
+      "loss": 3.901,
+      "step": 1769984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.101169577616858e-05,
+      "loss": 3.9065,
+      "step": 1770496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.100330982865806e-05,
+      "loss": 3.9101,
+      "step": 1771008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.099494025995127e-05,
+      "loss": 3.9174,
+      "step": 1771520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.098655431244075e-05,
+      "loss": 3.912,
+      "step": 1772032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.097818474373396e-05,
+      "loss": 3.916,
+      "step": 1772544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.096979879622344e-05,
+      "loss": 3.9102,
+      "step": 1773056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0961429227516656e-05,
+      "loss": 3.9127,
+      "step": 1773568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0953043280006133e-05,
+      "loss": 3.9143,
+      "step": 1774080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0944657332495613e-05,
+      "loss": 3.8995,
+      "step": 1774592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0936271384985093e-05,
+      "loss": 3.8971,
+      "step": 1775104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0927885437474573e-05,
+      "loss": 3.8934,
+      "step": 1775616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0919499489964053e-05,
+      "loss": 3.9046,
+      "step": 1776128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0911113542453533e-05,
+      "loss": 3.9068,
+      "step": 1776640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0902727594943013e-05,
+      "loss": 3.9067,
+      "step": 1777152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0894341647432492e-05,
+      "loss": 3.9047,
+      "step": 1777664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0885972078725705e-05,
+      "loss": 3.904,
+      "step": 1778176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0877586131215185e-05,
+      "loss": 3.9063,
+      "step": 1778688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.086920018370466e-05,
+      "loss": 3.911,
+      "step": 1779200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.086081423619414e-05,
+      "loss": 3.8971,
+      "step": 1779712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.085242828868362e-05,
+      "loss": 3.8826,
+      "step": 1780224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0844042341173105e-05,
+      "loss": 3.922,
+      "step": 1780736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0835656393662585e-05,
+      "loss": 3.9118,
+      "step": 1781248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0827270446152065e-05,
+      "loss": 3.9173,
+      "step": 1781760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0818900877445274e-05,
+      "loss": 3.899,
+      "step": 1782272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0810514929934754e-05,
+      "loss": 3.9072,
+      "step": 1782784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0802145361227966e-05,
+      "loss": 3.8985,
+      "step": 1783296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.079377579252118e-05,
+      "loss": 3.9022,
+      "step": 1783808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.078538984501066e-05,
+      "loss": 3.908,
+      "step": 1784320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0777003897500135e-05,
+      "loss": 3.9008,
+      "step": 1784832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0768617949989615e-05,
+      "loss": 3.9258,
+      "step": 1785344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0760232002479095e-05,
+      "loss": 3.9041,
+      "step": 1785856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0751846054968575e-05,
+      "loss": 3.8887,
+      "step": 1786368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.074346010745806e-05,
+      "loss": 3.9184,
+      "step": 1786880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0735090538751267e-05,
+      "loss": 3.8913,
+      "step": 1787392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0726704591240747e-05,
+      "loss": 3.8916,
+      "step": 1787904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0718318643730227e-05,
+      "loss": 3.9159,
+      "step": 1788416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0709932696219707e-05,
+      "loss": 3.9119,
+      "step": 1788928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0701546748709187e-05,
+      "loss": 3.8961,
+      "step": 1789440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0693160801198667e-05,
+      "loss": 3.9042,
+      "step": 1789952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0684774853688147e-05,
+      "loss": 3.8881,
+      "step": 1790464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0676388906177627e-05,
+      "loss": 3.8981,
+      "step": 1790976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0668002958667107e-05,
+      "loss": 3.9045,
+      "step": 1791488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0659617011156587e-05,
+      "loss": 3.9004,
+      "step": 1792000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0651231063646067e-05,
+      "loss": 3.9075,
+      "step": 1792512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.064286149493928e-05,
+      "loss": 3.9086,
+      "step": 1793024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.063447554742876e-05,
+      "loss": 3.9207,
+      "step": 1793536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.062608959991824e-05,
+      "loss": 3.9008,
+      "step": 1794048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.061770365240772e-05,
+      "loss": 3.9048,
+      "step": 1794560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.06093177048972e-05,
+      "loss": 3.909,
+      "step": 1795072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0600931757386676e-05,
+      "loss": 3.8995,
+      "step": 1795584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0592545809876156e-05,
+      "loss": 3.9079,
+      "step": 1796096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0584159862365636e-05,
+      "loss": 3.9048,
+      "step": 1796608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.057579029365885e-05,
+      "loss": 3.9023,
+      "step": 1797120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0567404346148328e-05,
+      "loss": 3.9146,
+      "step": 1797632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0559018398637808e-05,
+      "loss": 3.8897,
+      "step": 1798144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0550632451127288e-05,
+      "loss": 3.897,
+      "step": 1798656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0542262882420497e-05,
+      "loss": 3.9075,
+      "step": 1799168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.053387693490998e-05,
+      "loss": 3.9103,
+      "step": 1799680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.052549098739946e-05,
+      "loss": 3.9,
+      "step": 1800192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0517121418692673e-05,
+      "loss": 3.9094,
+      "step": 1800704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.050873547118215e-05,
+      "loss": 3.8883,
+      "step": 1801216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.050034952367163e-05,
+      "loss": 3.9014,
+      "step": 1801728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.049196357616111e-05,
+      "loss": 3.8959,
+      "step": 1802240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.048357762865059e-05,
+      "loss": 3.901,
+      "step": 1802752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0475191681140073e-05,
+      "loss": 3.9152,
+      "step": 1803264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0466822112433282e-05,
+      "loss": 3.9048,
+      "step": 1803776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0458436164922762e-05,
+      "loss": 3.9097,
+      "step": 1804288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0450050217412242e-05,
+      "loss": 3.8978,
+      "step": 1804800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0441664269901722e-05,
+      "loss": 3.8992,
+      "step": 1805312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0433278322391202e-05,
+      "loss": 3.9005,
+      "step": 1805824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0424892374880682e-05,
+      "loss": 3.9108,
+      "step": 1806336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.041650642737016e-05,
+      "loss": 3.9071,
+      "step": 1806848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.040812047985964e-05,
+      "loss": 3.9121,
+      "step": 1807360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.039975091115285e-05,
+      "loss": 3.898,
+      "step": 1807872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.039136496364233e-05,
+      "loss": 3.9068,
+      "step": 1808384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.038297901613181e-05,
+      "loss": 3.8995,
+      "step": 1808896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0374609447425027e-05,
+      "loss": 3.904,
+      "step": 1809408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0366223499914503e-05,
+      "loss": 3.8932,
+      "step": 1809920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0357837552403983e-05,
+      "loss": 3.9101,
+      "step": 1810432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0349451604893463e-05,
+      "loss": 3.8947,
+      "step": 1810944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0341065657382943e-05,
+      "loss": 3.9011,
+      "step": 1811456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0332679709872423e-05,
+      "loss": 3.8961,
+      "step": 1811968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0324293762361903e-05,
+      "loss": 3.9054,
+      "step": 1812480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0315924193655115e-05,
+      "loss": 3.8974,
+      "step": 1812992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0307538246144595e-05,
+      "loss": 3.8947,
+      "step": 1813504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0299152298634075e-05,
+      "loss": 3.9121,
+      "step": 1814016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0290766351123555e-05,
+      "loss": 3.9181,
+      "step": 1814528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0282396782416764e-05,
+      "loss": 3.914,
+      "step": 1815040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0274027213709977e-05,
+      "loss": 3.904,
+      "step": 1815552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0265641266199457e-05,
+      "loss": 3.8969,
+      "step": 1816064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0257255318688937e-05,
+      "loss": 3.8983,
+      "step": 1816576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0248869371178417e-05,
+      "loss": 3.9052,
+      "step": 1817088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0240483423667897e-05,
+      "loss": 3.9046,
+      "step": 1817600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0232097476157377e-05,
+      "loss": 3.8897,
+      "step": 1818112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0223711528646857e-05,
+      "loss": 3.9074,
+      "step": 1818624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0215325581136336e-05,
+      "loss": 3.9047,
+      "step": 1819136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0206939633625816e-05,
+      "loss": 3.9042,
+      "step": 1819648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0198553686115296e-05,
+      "loss": 3.9008,
+      "step": 1820160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0190184117408505e-05,
+      "loss": 3.9018,
+      "step": 1820672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0181798169897985e-05,
+      "loss": 3.8972,
+      "step": 1821184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0173412222387465e-05,
+      "loss": 3.9111,
+      "step": 1821696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0165026274876945e-05,
+      "loss": 3.9071,
+      "step": 1822208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.015664032736643e-05,
+      "loss": 3.9119,
+      "step": 1822720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.014825437985591e-05,
+      "loss": 3.8907,
+      "step": 1823232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.013986843234539e-05,
+      "loss": 3.9141,
+      "step": 1823744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0131482484834865e-05,
+      "loss": 3.911,
+      "step": 1824256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0123112916128078e-05,
+      "loss": 3.899,
+      "step": 1824768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0114726968617558e-05,
+      "loss": 3.9091,
+      "step": 1825280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0106341021107038e-05,
+      "loss": 3.905,
+      "step": 1825792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0097955073596518e-05,
+      "loss": 3.9097,
+      "step": 1826304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.008960188369346e-05,
+      "loss": 3.9216,
+      "step": 1826816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.008121593618294e-05,
+      "loss": 3.8983,
+      "step": 1827328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.007282998867242e-05,
+      "loss": 3.9055,
+      "step": 1827840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.00644440411619e-05,
+      "loss": 3.91,
+      "step": 1828352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0056058093651382e-05,
+      "loss": 3.9054,
+      "step": 1828864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0047672146140862e-05,
+      "loss": 3.8975,
+      "step": 1829376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.003928619863034e-05,
+      "loss": 3.8978,
+      "step": 1829888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.003090025111982e-05,
+      "loss": 3.9028,
+      "step": 1830400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.002253068241303e-05,
+      "loss": 3.8908,
+      "step": 1830912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.001414473490251e-05,
+      "loss": 3.9074,
+      "step": 1831424
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.996229887008667,
+      "eval_runtime": 278.5941,
+      "eval_samples_per_second": 1369.702,
+      "eval_steps_per_second": 42.804,
+      "step": 1831667
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000575878739199e-05,
+      "loss": 3.908,
+      "step": 1831936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.999737283988147e-05,
+      "loss": 3.8992,
+      "step": 1832448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9989003271174684e-05,
+      "loss": 3.9054,
+      "step": 1832960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.998061732366416e-05,
+      "loss": 3.9063,
+      "step": 1833472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997223137615364e-05,
+      "loss": 3.9135,
+      "step": 1833984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.996384542864312e-05,
+      "loss": 3.902,
+      "step": 1834496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.99554594811326e-05,
+      "loss": 3.9105,
+      "step": 1835008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9947089912425813e-05,
+      "loss": 3.8959,
+      "step": 1835520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9938703964915293e-05,
+      "loss": 3.8979,
+      "step": 1836032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9930318017404773e-05,
+      "loss": 3.9111,
+      "step": 1836544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9921932069894252e-05,
+      "loss": 3.9037,
+      "step": 1837056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9913546122383732e-05,
+      "loss": 3.901,
+      "step": 1837568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9905176553676945e-05,
+      "loss": 3.9075,
+      "step": 1838080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9896790606166425e-05,
+      "loss": 3.8967,
+      "step": 1838592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9888404658655905e-05,
+      "loss": 3.8969,
+      "step": 1839104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9880018711145385e-05,
+      "loss": 3.8906,
+      "step": 1839616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9871632763634865e-05,
+      "loss": 3.9009,
+      "step": 1840128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.986324681612434e-05,
+      "loss": 3.9033,
+      "step": 1840640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9854877247417554e-05,
+      "loss": 3.898,
+      "step": 1841152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9846491299907037e-05,
+      "loss": 3.9113,
+      "step": 1841664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9838105352396517e-05,
+      "loss": 3.9121,
+      "step": 1842176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9829719404885994e-05,
+      "loss": 3.9089,
+      "step": 1842688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9821333457375474e-05,
+      "loss": 3.9022,
+      "step": 1843200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9812963888668686e-05,
+      "loss": 3.9021,
+      "step": 1843712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9804577941158166e-05,
+      "loss": 3.9073,
+      "step": 1844224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9796191993647646e-05,
+      "loss": 3.9069,
+      "step": 1844736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9787806046137126e-05,
+      "loss": 3.8956,
+      "step": 1845248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.977943647743034e-05,
+      "loss": 3.9039,
+      "step": 1845760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9771050529919815e-05,
+      "loss": 3.8925,
+      "step": 1846272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9762664582409295e-05,
+      "loss": 3.9026,
+      "step": 1846784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9754278634898775e-05,
+      "loss": 3.8985,
+      "step": 1847296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9745892687388258e-05,
+      "loss": 3.9088,
+      "step": 1847808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9737506739877738e-05,
+      "loss": 3.9108,
+      "step": 1848320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9729120792367218e-05,
+      "loss": 3.9121,
+      "step": 1848832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9720734844856698e-05,
+      "loss": 3.8997,
+      "step": 1849344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9712365276149907e-05,
+      "loss": 3.9074,
+      "step": 1849856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9703979328639387e-05,
+      "loss": 3.9096,
+      "step": 1850368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9695593381128867e-05,
+      "loss": 3.8926,
+      "step": 1850880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9687207433618347e-05,
+      "loss": 3.8929,
+      "step": 1851392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.967883786491156e-05,
+      "loss": 3.891,
+      "step": 1851904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.967045191740104e-05,
+      "loss": 3.8942,
+      "step": 1852416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.966206596989052e-05,
+      "loss": 3.8981,
+      "step": 1852928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9653680022379996e-05,
+      "loss": 3.9036,
+      "step": 1853440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9645294074869476e-05,
+      "loss": 3.8997,
+      "step": 1853952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.963690812735896e-05,
+      "loss": 3.8956,
+      "step": 1854464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.962852217984844e-05,
+      "loss": 3.9018,
+      "step": 1854976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.962013623233792e-05,
+      "loss": 3.9046,
+      "step": 1855488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.961176666363113e-05,
+      "loss": 3.8909,
+      "step": 1856000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.960338071612061e-05,
+      "loss": 3.8762,
+      "step": 1856512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9594994768610088e-05,
+      "loss": 3.9136,
+      "step": 1857024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9586608821099568e-05,
+      "loss": 3.9074,
+      "step": 1857536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.957823925239278e-05,
+      "loss": 3.9099,
+      "step": 1858048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.956985330488226e-05,
+      "loss": 3.8962,
+      "step": 1858560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.956146735737174e-05,
+      "loss": 3.8972,
+      "step": 1859072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.955308140986122e-05,
+      "loss": 3.8967,
+      "step": 1859584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.954471184115443e-05,
+      "loss": 3.8975,
+      "step": 1860096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9536325893643913e-05,
+      "loss": 3.9044,
+      "step": 1860608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9527939946133393e-05,
+      "loss": 3.8901,
+      "step": 1861120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9519553998622873e-05,
+      "loss": 3.9201,
+      "step": 1861632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9511168051112353e-05,
+      "loss": 3.8988,
+      "step": 1862144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9502798482405562e-05,
+      "loss": 3.8796,
+      "step": 1862656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9494412534895042e-05,
+      "loss": 3.9103,
+      "step": 1863168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9486026587384522e-05,
+      "loss": 3.8924,
+      "step": 1863680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9477640639874002e-05,
+      "loss": 3.8837,
+      "step": 1864192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9469254692363482e-05,
+      "loss": 3.9077,
+      "step": 1864704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9460885123656694e-05,
+      "loss": 3.9068,
+      "step": 1865216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9452499176146174e-05,
+      "loss": 3.8921,
+      "step": 1865728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.944411322863565e-05,
+      "loss": 3.8966,
+      "step": 1866240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.943572728112513e-05,
+      "loss": 3.8812,
+      "step": 1866752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9427341333614614e-05,
+      "loss": 3.8924,
+      "step": 1867264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9418955386104094e-05,
+      "loss": 3.9004,
+      "step": 1867776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9410569438593574e-05,
+      "loss": 3.8954,
+      "step": 1868288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9402183491083054e-05,
+      "loss": 3.8998,
+      "step": 1868800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9393813922376263e-05,
+      "loss": 3.899,
+      "step": 1869312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9385427974865743e-05,
+      "loss": 3.9133,
+      "step": 1869824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9377042027355223e-05,
+      "loss": 3.8983,
+      "step": 1870336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9368656079844706e-05,
+      "loss": 3.9007,
+      "step": 1870848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9360286511137915e-05,
+      "loss": 3.9002,
+      "step": 1871360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9351900563627395e-05,
+      "loss": 3.8957,
+      "step": 1871872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9343514616116875e-05,
+      "loss": 3.9045,
+      "step": 1872384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9335128668606355e-05,
+      "loss": 3.898,
+      "step": 1872896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9326742721095832e-05,
+      "loss": 3.8977,
+      "step": 1873408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9318356773585315e-05,
+      "loss": 3.9112,
+      "step": 1873920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9309970826074795e-05,
+      "loss": 3.8858,
+      "step": 1874432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9301584878564275e-05,
+      "loss": 3.8889,
+      "step": 1874944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9293215309857484e-05,
+      "loss": 3.9019,
+      "step": 1875456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9284829362346964e-05,
+      "loss": 3.902,
+      "step": 1875968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9276443414836444e-05,
+      "loss": 3.8921,
+      "step": 1876480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9268073846129657e-05,
+      "loss": 3.9064,
+      "step": 1876992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9259687898619137e-05,
+      "loss": 3.8871,
+      "step": 1877504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9251301951108617e-05,
+      "loss": 3.8937,
+      "step": 1878016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9242916003598096e-05,
+      "loss": 3.8894,
+      "step": 1878528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9234530056087576e-05,
+      "loss": 3.8914,
+      "step": 1879040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9226144108577056e-05,
+      "loss": 3.9114,
+      "step": 1879552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.921777453987027e-05,
+      "loss": 3.8995,
+      "step": 1880064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9209404971163478e-05,
+      "loss": 3.9033,
+      "step": 1880576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9201019023652958e-05,
+      "loss": 3.8937,
+      "step": 1881088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9192633076142438e-05,
+      "loss": 3.8919,
+      "step": 1881600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9184247128631918e-05,
+      "loss": 3.895,
+      "step": 1882112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9175861181121398e-05,
+      "loss": 3.9045,
+      "step": 1882624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9167475233610878e-05,
+      "loss": 3.901,
+      "step": 1883136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.915910566490409e-05,
+      "loss": 3.9057,
+      "step": 1883648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.915071971739357e-05,
+      "loss": 3.8984,
+      "step": 1884160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.914233376988305e-05,
+      "loss": 3.8984,
+      "step": 1884672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.913394782237253e-05,
+      "loss": 3.8979,
+      "step": 1885184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.912556187486201e-05,
+      "loss": 3.8953,
+      "step": 1885696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.911717592735149e-05,
+      "loss": 3.8853,
+      "step": 1886208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.910878997984097e-05,
+      "loss": 3.9058,
+      "step": 1886720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.910040403233045e-05,
+      "loss": 3.8883,
+      "step": 1887232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.909201808481993e-05,
+      "loss": 3.8954,
+      "step": 1887744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.908363213730941e-05,
+      "loss": 3.8902,
+      "step": 1888256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.907524618979889e-05,
+      "loss": 3.8942,
+      "step": 1888768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9066860242288366e-05,
+      "loss": 3.8947,
+      "step": 1889280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.905849067358158e-05,
+      "loss": 3.8916,
+      "step": 1889792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9050104726071062e-05,
+      "loss": 3.9047,
+      "step": 1890304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9041718778560542e-05,
+      "loss": 3.9131,
+      "step": 1890816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.903333283105002e-05,
+      "loss": 3.9079,
+      "step": 1891328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.902496326234323e-05,
+      "loss": 3.8979,
+      "step": 1891840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.901657731483271e-05,
+      "loss": 3.8908,
+      "step": 1892352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.900819136732219e-05,
+      "loss": 3.8926,
+      "step": 1892864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.899980541981167e-05,
+      "loss": 3.9006,
+      "step": 1893376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8991452229908613e-05,
+      "loss": 3.8987,
+      "step": 1893888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8983066282398093e-05,
+      "loss": 3.8893,
+      "step": 1894400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8974680334887573e-05,
+      "loss": 3.8954,
+      "step": 1894912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8966294387377053e-05,
+      "loss": 3.8995,
+      "step": 1895424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8957908439866533e-05,
+      "loss": 3.8962,
+      "step": 1895936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8949522492356016e-05,
+      "loss": 3.894,
+      "step": 1896448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8941136544845492e-05,
+      "loss": 3.8984,
+      "step": 1896960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8932750597334972e-05,
+      "loss": 3.891,
+      "step": 1897472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8924364649824452e-05,
+      "loss": 3.9079,
+      "step": 1897984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8915995081117665e-05,
+      "loss": 3.8998,
+      "step": 1898496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8907625512410877e-05,
+      "loss": 3.9066,
+      "step": 1899008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8899239564900357e-05,
+      "loss": 3.8885,
+      "step": 1899520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8890853617389837e-05,
+      "loss": 3.9043,
+      "step": 1900032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8882467669879314e-05,
+      "loss": 3.9068,
+      "step": 1900544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8874081722368794e-05,
+      "loss": 3.8965,
+      "step": 1901056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8865695774858274e-05,
+      "loss": 3.9007,
+      "step": 1901568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8857309827347754e-05,
+      "loss": 3.9001,
+      "step": 1902080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8848940258640966e-05,
+      "loss": 3.9014,
+      "step": 1902592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.884057068993418e-05,
+      "loss": 3.9168,
+      "step": 1903104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.883218474242366e-05,
+      "loss": 3.8924,
+      "step": 1903616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.882379879491314e-05,
+      "loss": 3.9024,
+      "step": 1904128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8815412847402615e-05,
+      "loss": 3.9032,
+      "step": 1904640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.880704327869583e-05,
+      "loss": 3.8979,
+      "step": 1905152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.879865733118531e-05,
+      "loss": 3.8925,
+      "step": 1905664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8790271383674787e-05,
+      "loss": 3.8891,
+      "step": 1906176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8781901814968e-05,
+      "loss": 3.8963,
+      "step": 1906688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.877351586745748e-05,
+      "loss": 3.8919,
+      "step": 1907200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.876512991994696e-05,
+      "loss": 3.898,
+      "step": 1907712
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9940690994262695,
+      "eval_runtime": 278.8017,
+      "eval_samples_per_second": 1368.683,
+      "eval_steps_per_second": 42.772,
+      "step": 1907987
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.875674397243644e-05,
+      "loss": 3.9003,
+      "step": 1908224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.874835802492592e-05,
+      "loss": 3.8956,
+      "step": 1908736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.87399720774154e-05,
+      "loss": 3.8969,
+      "step": 1909248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.873158612990488e-05,
+      "loss": 3.9014,
+      "step": 1909760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.872320018239436e-05,
+      "loss": 3.9097,
+      "step": 1910272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.871481423488384e-05,
+      "loss": 3.9013,
+      "step": 1910784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.870642828737332e-05,
+      "loss": 3.9005,
+      "step": 1911296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.86980423398628e-05,
+      "loss": 3.8906,
+      "step": 1911808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.868965639235228e-05,
+      "loss": 3.8906,
+      "step": 1912320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.868127044484176e-05,
+      "loss": 3.9059,
+      "step": 1912832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.867288449733124e-05,
+      "loss": 3.8982,
+      "step": 1913344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.866449854982072e-05,
+      "loss": 3.8956,
+      "step": 1913856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.865612898111393e-05,
+      "loss": 3.9045,
+      "step": 1914368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.864774303360341e-05,
+      "loss": 3.8909,
+      "step": 1914880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8639357086092892e-05,
+      "loss": 3.8933,
+      "step": 1915392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8630971138582372e-05,
+      "loss": 3.8811,
+      "step": 1915904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.862258519107185e-05,
+      "loss": 3.897,
+      "step": 1916416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.861421562236506e-05,
+      "loss": 3.9018,
+      "step": 1916928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.860582967485454e-05,
+      "loss": 3.8885,
+      "step": 1917440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.859744372734402e-05,
+      "loss": 3.9054,
+      "step": 1917952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.85890577798335e-05,
+      "loss": 3.9075,
+      "step": 1918464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.858067183232298e-05,
+      "loss": 3.9059,
+      "step": 1918976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.857228588481246e-05,
+      "loss": 3.8969,
+      "step": 1919488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.856389993730194e-05,
+      "loss": 3.9017,
+      "step": 1920000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.855551398979142e-05,
+      "loss": 3.9017,
+      "step": 1920512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.85471280422809e-05,
+      "loss": 3.8992,
+      "step": 1921024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.853874209477038e-05,
+      "loss": 3.8881,
+      "step": 1921536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8530356147259857e-05,
+      "loss": 3.8982,
+      "step": 1922048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.852197019974934e-05,
+      "loss": 3.8889,
+      "step": 1922560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8513600631042553e-05,
+      "loss": 3.8965,
+      "step": 1923072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8505214683532033e-05,
+      "loss": 3.8998,
+      "step": 1923584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.849682873602151e-05,
+      "loss": 3.8982,
+      "step": 1924096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8488459167314722e-05,
+      "loss": 3.9013,
+      "step": 1924608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8480073219804202e-05,
+      "loss": 3.9096,
+      "step": 1925120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.847168727229368e-05,
+      "loss": 3.8948,
+      "step": 1925632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8463317703586894e-05,
+      "loss": 3.9,
+      "step": 1926144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8454931756076374e-05,
+      "loss": 3.9034,
+      "step": 1926656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8446545808565854e-05,
+      "loss": 3.8862,
+      "step": 1927168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.843815986105533e-05,
+      "loss": 3.8875,
+      "step": 1927680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.842977391354481e-05,
+      "loss": 3.8874,
+      "step": 1928192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8421387966034294e-05,
+      "loss": 3.8874,
+      "step": 1928704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8413018397327506e-05,
+      "loss": 3.8931,
+      "step": 1929216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8404632449816983e-05,
+      "loss": 3.9031,
+      "step": 1929728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8396246502306463e-05,
+      "loss": 3.8891,
+      "step": 1930240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8387860554795943e-05,
+      "loss": 3.8941,
+      "step": 1930752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8379474607285423e-05,
+      "loss": 3.8907,
+      "step": 1931264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8371088659774903e-05,
+      "loss": 3.8995,
+      "step": 1931776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8362702712264386e-05,
+      "loss": 3.8843,
+      "step": 1932288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8354316764753863e-05,
+      "loss": 3.8738,
+      "step": 1932800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8345930817243343e-05,
+      "loss": 3.9054,
+      "step": 1933312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8337544869732823e-05,
+      "loss": 3.9017,
+      "step": 1933824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8329158922222303e-05,
+      "loss": 3.9089,
+      "step": 1934336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8320772974711783e-05,
+      "loss": 3.8878,
+      "step": 1934848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8312403406004995e-05,
+      "loss": 3.8922,
+      "step": 1935360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8304017458494475e-05,
+      "loss": 3.8928,
+      "step": 1935872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8295647889787687e-05,
+      "loss": 3.888,
+      "step": 1936384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8287261942277164e-05,
+      "loss": 3.9002,
+      "step": 1936896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8278875994766644e-05,
+      "loss": 3.882,
+      "step": 1937408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8270490047256124e-05,
+      "loss": 3.9155,
+      "step": 1937920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8262104099745604e-05,
+      "loss": 3.8965,
+      "step": 1938432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8253718152235087e-05,
+      "loss": 3.8701,
+      "step": 1938944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8245348583528296e-05,
+      "loss": 3.906,
+      "step": 1939456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.823697901482151e-05,
+      "loss": 3.8904,
+      "step": 1939968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8228593067310985e-05,
+      "loss": 3.8744,
+      "step": 1940480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8220207119800465e-05,
+      "loss": 3.9007,
+      "step": 1940992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.821182117228995e-05,
+      "loss": 3.9014,
+      "step": 1941504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.820343522477943e-05,
+      "loss": 3.8861,
+      "step": 1942016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.819504927726891e-05,
+      "loss": 3.8911,
+      "step": 1942528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.818666332975839e-05,
+      "loss": 3.875,
+      "step": 1943040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.817827738224787e-05,
+      "loss": 3.8862,
+      "step": 1943552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8169891434737345e-05,
+      "loss": 3.8935,
+      "step": 1944064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8161505487226825e-05,
+      "loss": 3.8906,
+      "step": 1944576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.815311953971631e-05,
+      "loss": 3.8965,
+      "step": 1945088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.814473359220579e-05,
+      "loss": 3.8887,
+      "step": 1945600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8136364023498997e-05,
+      "loss": 3.9105,
+      "step": 1946112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.812799445479221e-05,
+      "loss": 3.8952,
+      "step": 1946624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.811960850728169e-05,
+      "loss": 3.8922,
+      "step": 1947136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.811122255977117e-05,
+      "loss": 3.8942,
+      "step": 1947648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.810283661226065e-05,
+      "loss": 3.8919,
+      "step": 1948160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.809445066475013e-05,
+      "loss": 3.8966,
+      "step": 1948672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.808606471723961e-05,
+      "loss": 3.8909,
+      "step": 1949184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.807767876972909e-05,
+      "loss": 3.8954,
+      "step": 1949696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.806929282221857e-05,
+      "loss": 3.9025,
+      "step": 1950208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.806092325351178e-05,
+      "loss": 3.8847,
+      "step": 1950720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8052537306001262e-05,
+      "loss": 3.8817,
+      "step": 1951232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8044151358490742e-05,
+      "loss": 3.895,
+      "step": 1951744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8035765410980222e-05,
+      "loss": 3.9007,
+      "step": 1952256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.802739584227343e-05,
+      "loss": 3.887,
+      "step": 1952768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.801900989476291e-05,
+      "loss": 3.8961,
+      "step": 1953280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.801062394725239e-05,
+      "loss": 3.8834,
+      "step": 1953792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.800223799974187e-05,
+      "loss": 3.8888,
+      "step": 1954304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7993868431035083e-05,
+      "loss": 3.8839,
+      "step": 1954816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7985482483524563e-05,
+      "loss": 3.8885,
+      "step": 1955328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7977096536014043e-05,
+      "loss": 3.9066,
+      "step": 1955840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7968726967307252e-05,
+      "loss": 3.8935,
+      "step": 1956352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7960341019796732e-05,
+      "loss": 3.8976,
+      "step": 1956864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7951955072286212e-05,
+      "loss": 3.8865,
+      "step": 1957376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7943569124775696e-05,
+      "loss": 3.8858,
+      "step": 1957888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7935183177265172e-05,
+      "loss": 3.8884,
+      "step": 1958400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7926813608558385e-05,
+      "loss": 3.899,
+      "step": 1958912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7918427661047865e-05,
+      "loss": 3.8979,
+      "step": 1959424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7910041713537345e-05,
+      "loss": 3.8951,
+      "step": 1959936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7901655766026825e-05,
+      "loss": 3.8952,
+      "step": 1960448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7893286197320037e-05,
+      "loss": 3.8889,
+      "step": 1960960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7884900249809517e-05,
+      "loss": 3.8936,
+      "step": 1961472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7876514302298994e-05,
+      "loss": 3.89,
+      "step": 1961984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7868128354788474e-05,
+      "loss": 3.8827,
+      "step": 1962496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7859742407277954e-05,
+      "loss": 3.8935,
+      "step": 1963008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7851372838571166e-05,
+      "loss": 3.8862,
+      "step": 1963520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7842986891060646e-05,
+      "loss": 3.8865,
+      "step": 1964032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7834600943550126e-05,
+      "loss": 3.8884,
+      "step": 1964544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7826214996039606e-05,
+      "loss": 3.8856,
+      "step": 1965056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7817829048529086e-05,
+      "loss": 3.8911,
+      "step": 1965568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7809443101018566e-05,
+      "loss": 3.8857,
+      "step": 1966080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7801057153508046e-05,
+      "loss": 3.901,
+      "step": 1966592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7792687584801258e-05,
+      "loss": 3.9032,
+      "step": 1967104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7784301637290738e-05,
+      "loss": 3.9032,
+      "step": 1967616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7775915689780218e-05,
+      "loss": 3.8949,
+      "step": 1968128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7767546121073427e-05,
+      "loss": 3.8876,
+      "step": 1968640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7759160173562907e-05,
+      "loss": 3.8883,
+      "step": 1969152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7750774226052387e-05,
+      "loss": 3.895,
+      "step": 1969664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.774238827854187e-05,
+      "loss": 3.8989,
+      "step": 1970176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.773400233103135e-05,
+      "loss": 3.8775,
+      "step": 1970688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7725649141128292e-05,
+      "loss": 3.8926,
+      "step": 1971200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.771726319361777e-05,
+      "loss": 3.8925,
+      "step": 1971712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.770887724610725e-05,
+      "loss": 3.8969,
+      "step": 1972224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7700491298596732e-05,
+      "loss": 3.8881,
+      "step": 1972736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7692105351086212e-05,
+      "loss": 3.8882,
+      "step": 1973248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7683719403575692e-05,
+      "loss": 3.8869,
+      "step": 1973760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7675333456065172e-05,
+      "loss": 3.9038,
+      "step": 1974272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.766694750855465e-05,
+      "loss": 3.8937,
+      "step": 1974784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.765856156104413e-05,
+      "loss": 3.902,
+      "step": 1975296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7650175613533608e-05,
+      "loss": 3.8822,
+      "step": 1975808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7641789666023088e-05,
+      "loss": 3.8963,
+      "step": 1976320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.763340371851257e-05,
+      "loss": 3.9026,
+      "step": 1976832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.762501777100205e-05,
+      "loss": 3.8903,
+      "step": 1977344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.761664820229526e-05,
+      "loss": 3.8964,
+      "step": 1977856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.760826225478474e-05,
+      "loss": 3.8949,
+      "step": 1978368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.759989268607795e-05,
+      "loss": 3.8981,
+      "step": 1978880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7591506738567433e-05,
+      "loss": 3.9067,
+      "step": 1979392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7583120791056913e-05,
+      "loss": 3.892,
+      "step": 1979904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7574734843546393e-05,
+      "loss": 3.8959,
+      "step": 1980416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7566348896035873e-05,
+      "loss": 3.8962,
+      "step": 1980928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7557979327329082e-05,
+      "loss": 3.8929,
+      "step": 1981440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7549593379818562e-05,
+      "loss": 3.8881,
+      "step": 1981952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7541207432308042e-05,
+      "loss": 3.885,
+      "step": 1982464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7532821484797525e-05,
+      "loss": 3.8886,
+      "step": 1982976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7524435537287005e-05,
+      "loss": 3.8887,
+      "step": 1983488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7516065968580214e-05,
+      "loss": 3.8886,
+      "step": 1984000
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.99186372756958,
+      "eval_runtime": 302.9551,
+      "eval_samples_per_second": 1259.563,
+      "eval_steps_per_second": 39.362,
+      "step": 1984307
+    }
+  ],
+  "logging_steps": 512,
+  "max_steps": 3052726,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 10,
+  "total_flos": 8.039574427694412e+17,
+  "trial_name": null,
+  "trial_params": null
+}