{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.971014492753623,
  "eval_steps": 121,
  "global_step": 966,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002070393374741201,
      "grad_norm": 0.09777352213859558,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.6308,
      "step": 1
    },
    {
      "epoch": 0.002070393374741201,
      "eval_loss": 0.5235128402709961,
      "eval_runtime": 93.3308,
      "eval_samples_per_second": 5.914,
      "eval_steps_per_second": 0.739,
      "step": 1
    },
    {
      "epoch": 0.004140786749482402,
      "grad_norm": 0.09036517143249512,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.4787,
      "step": 2
    },
    {
      "epoch": 0.006211180124223602,
      "grad_norm": 0.0952724814414978,
      "learning_rate": 3e-06,
      "loss": 0.5643,
      "step": 3
    },
    {
      "epoch": 0.008281573498964804,
      "grad_norm": 0.10640479624271393,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.6445,
      "step": 4
    },
    {
      "epoch": 0.010351966873706004,
      "grad_norm": 0.0985221341252327,
      "learning_rate": 5e-06,
      "loss": 0.5846,
      "step": 5
    },
    {
      "epoch": 0.012422360248447204,
      "grad_norm": 0.07589095830917358,
      "learning_rate": 6e-06,
      "loss": 0.3481,
      "step": 6
    },
    {
      "epoch": 0.014492753623188406,
      "grad_norm": 0.09868590533733368,
      "learning_rate": 7e-06,
      "loss": 0.5721,
      "step": 7
    },
    {
      "epoch": 0.016563146997929608,
      "grad_norm": 0.0714084655046463,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.3615,
      "step": 8
    },
    {
      "epoch": 0.018633540372670808,
      "grad_norm": 0.074221171438694,
      "learning_rate": 9e-06,
      "loss": 0.4541,
      "step": 9
    },
    {
      "epoch": 0.020703933747412008,
      "grad_norm": 0.0852167010307312,
      "learning_rate": 1e-05,
      "loss": 0.5734,
      "step": 10
    },
    {
      "epoch": 0.022774327122153208,
      "grad_norm": 0.08485738188028336,
      "learning_rate": 9.999973002498209e-06,
      "loss": 0.4918,
      "step": 11
    },
    {
      "epoch": 0.024844720496894408,
      "grad_norm": 0.09552149474620819,
      "learning_rate": 9.999892010284378e-06,
      "loss": 0.5242,
      "step": 12
    },
    {
      "epoch": 0.026915113871635612,
      "grad_norm": 0.09766828268766403,
      "learning_rate": 9.999757024233143e-06,
      "loss": 0.5503,
      "step": 13
    },
    {
      "epoch": 0.028985507246376812,
      "grad_norm": 0.10151398926973343,
      "learning_rate": 9.999568045802216e-06,
      "loss": 0.6607,
      "step": 14
    },
    {
      "epoch": 0.031055900621118012,
      "grad_norm": 0.09601914882659912,
      "learning_rate": 9.999325077032382e-06,
      "loss": 0.6295,
      "step": 15
    },
    {
      "epoch": 0.033126293995859216,
      "grad_norm": 0.09811417013406754,
      "learning_rate": 9.999028120547456e-06,
      "loss": 0.4831,
      "step": 16
    },
    {
      "epoch": 0.035196687370600416,
      "grad_norm": 0.0988076776266098,
      "learning_rate": 9.99867717955427e-06,
      "loss": 0.513,
      "step": 17
    },
    {
      "epoch": 0.037267080745341616,
      "grad_norm": 0.08319983631372452,
      "learning_rate": 9.99827225784264e-06,
      "loss": 0.4423,
      "step": 18
    },
    {
      "epoch": 0.039337474120082816,
      "grad_norm": 0.08122409880161285,
      "learning_rate": 9.997813359785313e-06,
      "loss": 0.5584,
      "step": 19
    },
    {
      "epoch": 0.041407867494824016,
      "grad_norm": 0.09155779331922531,
      "learning_rate": 9.99730049033793e-06,
      "loss": 0.4512,
      "step": 20
    },
    {
      "epoch": 0.043478260869565216,
      "grad_norm": 0.09295297414064407,
      "learning_rate": 9.99673365503897e-06,
      "loss": 0.4382,
      "step": 21
    },
    {
      "epoch": 0.045548654244306416,
      "grad_norm": 0.12233804911375046,
      "learning_rate": 9.996112860009689e-06,
      "loss": 0.6128,
      "step": 22
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 0.1146979108452797,
      "learning_rate": 9.995438111954047e-06,
      "loss": 0.421,
      "step": 23
    },
    {
      "epoch": 0.049689440993788817,
      "grad_norm": 0.11026836186647415,
      "learning_rate": 9.994709418158652e-06,
      "loss": 0.6736,
      "step": 24
    },
    {
      "epoch": 0.051759834368530024,
      "grad_norm": 0.11445167660713196,
      "learning_rate": 9.993926786492672e-06,
      "loss": 0.5125,
      "step": 25
    },
    {
      "epoch": 0.053830227743271224,
      "grad_norm": 0.0917944386601448,
      "learning_rate": 9.993090225407743e-06,
      "loss": 0.436,
      "step": 26
    },
    {
      "epoch": 0.055900621118012424,
      "grad_norm": 0.09246832877397537,
      "learning_rate": 9.992199743937888e-06,
      "loss": 0.4278,
      "step": 27
    },
    {
      "epoch": 0.057971014492753624,
      "grad_norm": 0.07395437359809875,
      "learning_rate": 9.991255351699422e-06,
      "loss": 0.3163,
      "step": 28
    },
    {
      "epoch": 0.060041407867494824,
      "grad_norm": 0.0859774798154831,
      "learning_rate": 9.990257058890834e-06,
      "loss": 0.4385,
      "step": 29
    },
    {
      "epoch": 0.062111801242236024,
      "grad_norm": 0.11907877773046494,
      "learning_rate": 9.98920487629269e-06,
      "loss": 0.5022,
      "step": 30
    },
    {
      "epoch": 0.06418219461697723,
      "grad_norm": 0.08157500624656677,
      "learning_rate": 9.988098815267507e-06,
      "loss": 0.3605,
      "step": 31
    },
    {
      "epoch": 0.06625258799171843,
      "grad_norm": 0.10693008452653885,
      "learning_rate": 9.986938887759643e-06,
      "loss": 0.5329,
      "step": 32
    },
    {
      "epoch": 0.06832298136645963,
      "grad_norm": 0.11197963356971741,
      "learning_rate": 9.985725106295154e-06,
      "loss": 0.6241,
      "step": 33
    },
    {
      "epoch": 0.07039337474120083,
      "grad_norm": 0.07341015338897705,
      "learning_rate": 9.98445748398167e-06,
      "loss": 0.3944,
      "step": 34
    },
    {
      "epoch": 0.07246376811594203,
      "grad_norm": 0.08865534514188766,
      "learning_rate": 9.98313603450824e-06,
      "loss": 0.4709,
      "step": 35
    },
    {
      "epoch": 0.07453416149068323,
      "grad_norm": 0.09436415880918503,
      "learning_rate": 9.981760772145201e-06,
      "loss": 0.4746,
      "step": 36
    },
    {
      "epoch": 0.07660455486542443,
      "grad_norm": 0.07138030976057053,
      "learning_rate": 9.980331711744013e-06,
      "loss": 0.3825,
      "step": 37
    },
    {
      "epoch": 0.07867494824016563,
      "grad_norm": 0.11378210783004761,
      "learning_rate": 9.978848868737099e-06,
      "loss": 0.5328,
      "step": 38
    },
    {
      "epoch": 0.08074534161490683,
      "grad_norm": 0.10511545836925507,
      "learning_rate": 9.97731225913768e-06,
      "loss": 0.6454,
      "step": 39
    },
    {
      "epoch": 0.08281573498964803,
      "grad_norm": 0.08871125429868698,
      "learning_rate": 9.975721899539607e-06,
      "loss": 0.5153,
      "step": 40
    },
    {
      "epoch": 0.08488612836438923,
      "grad_norm": 0.07444795966148376,
      "learning_rate": 9.974077807117176e-06,
      "loss": 0.3639,
      "step": 41
    },
    {
      "epoch": 0.08695652173913043,
      "grad_norm": 0.06881678849458694,
      "learning_rate": 9.972379999624935e-06,
      "loss": 0.3338,
      "step": 42
    },
    {
      "epoch": 0.08902691511387163,
      "grad_norm": 0.07631891965866089,
      "learning_rate": 9.970628495397516e-06,
      "loss": 0.4224,
      "step": 43
    },
    {
      "epoch": 0.09109730848861283,
      "grad_norm": 0.0697372555732727,
      "learning_rate": 9.968823313349412e-06,
      "loss": 0.2783,
      "step": 44
    },
    {
      "epoch": 0.09316770186335403,
      "grad_norm": 0.07862786948680878,
      "learning_rate": 9.966964472974783e-06,
      "loss": 0.3833,
      "step": 45
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 0.09798941016197205,
      "learning_rate": 9.96505199434725e-06,
      "loss": 0.5329,
      "step": 46
    },
    {
      "epoch": 0.09730848861283643,
      "grad_norm": 0.07815828919410706,
      "learning_rate": 9.96308589811967e-06,
      "loss": 0.5146,
      "step": 47
    },
    {
      "epoch": 0.09937888198757763,
      "grad_norm": 0.10919066518545151,
      "learning_rate": 9.961066205523917e-06,
      "loss": 0.5496,
      "step": 48
    },
    {
      "epoch": 0.10144927536231885,
      "grad_norm": 0.07423596829175949,
      "learning_rate": 9.958992938370655e-06,
      "loss": 0.3172,
      "step": 49
    },
    {
      "epoch": 0.10351966873706005,
      "grad_norm": 0.0891297236084938,
      "learning_rate": 9.956866119049095e-06,
      "loss": 0.473,
      "step": 50
    },
    {
      "epoch": 0.10559006211180125,
      "grad_norm": 0.0813724547624588,
      "learning_rate": 9.954685770526762e-06,
      "loss": 0.3897,
      "step": 51
    },
    {
      "epoch": 0.10766045548654245,
      "grad_norm": 0.0775756984949112,
      "learning_rate": 9.952451916349242e-06,
      "loss": 0.3008,
      "step": 52
    },
    {
      "epoch": 0.10973084886128365,
      "grad_norm": 0.07236716151237488,
      "learning_rate": 9.950164580639925e-06,
      "loss": 0.3333,
      "step": 53
    },
    {
      "epoch": 0.11180124223602485,
      "grad_norm": 0.1051044836640358,
      "learning_rate": 9.947823788099754e-06,
      "loss": 0.4174,
      "step": 54
    },
    {
      "epoch": 0.11387163561076605,
      "grad_norm": 0.06772394478321075,
      "learning_rate": 9.945429564006945e-06,
      "loss": 0.4263,
      "step": 55
    },
    {
      "epoch": 0.11594202898550725,
      "grad_norm": 0.08235827088356018,
      "learning_rate": 9.942981934216731e-06,
      "loss": 0.4846,
      "step": 56
    },
    {
      "epoch": 0.11801242236024845,
      "grad_norm": 0.07849966734647751,
      "learning_rate": 9.940480925161065e-06,
      "loss": 0.4597,
      "step": 57
    },
    {
      "epoch": 0.12008281573498965,
      "grad_norm": 0.08702969551086426,
      "learning_rate": 9.937926563848345e-06,
      "loss": 0.468,
      "step": 58
    },
    {
      "epoch": 0.12215320910973085,
      "grad_norm": 0.08962167054414749,
      "learning_rate": 9.935318877863123e-06,
      "loss": 0.4397,
      "step": 59
    },
    {
      "epoch": 0.12422360248447205,
      "grad_norm": 0.06720270216464996,
      "learning_rate": 9.9326578953658e-06,
      "loss": 0.3646,
      "step": 60
    },
    {
      "epoch": 0.12629399585921325,
      "grad_norm": 0.07080550491809845,
      "learning_rate": 9.929943645092328e-06,
      "loss": 0.4352,
      "step": 61
    },
    {
      "epoch": 0.12836438923395446,
      "grad_norm": 0.10958462953567505,
      "learning_rate": 9.9271761563539e-06,
      "loss": 0.5284,
      "step": 62
    },
    {
      "epoch": 0.13043478260869565,
      "grad_norm": 0.06647852063179016,
      "learning_rate": 9.924355459036624e-06,
      "loss": 0.4193,
      "step": 63
    },
    {
      "epoch": 0.13250517598343686,
      "grad_norm": 0.08806627988815308,
      "learning_rate": 9.921481583601218e-06,
      "loss": 0.3681,
      "step": 64
    },
    {
      "epoch": 0.13457556935817805,
      "grad_norm": 0.08910389989614487,
      "learning_rate": 9.91855456108266e-06,
      "loss": 0.5611,
      "step": 65
    },
    {
      "epoch": 0.13664596273291926,
      "grad_norm": 0.08766467124223709,
      "learning_rate": 9.915574423089872e-06,
      "loss": 0.3665,
      "step": 66
    },
    {
      "epoch": 0.13871635610766045,
      "grad_norm": 0.05857433006167412,
      "learning_rate": 9.912541201805362e-06,
      "loss": 0.3057,
      "step": 67
    },
    {
      "epoch": 0.14078674948240166,
      "grad_norm": 0.07592141628265381,
      "learning_rate": 9.909454929984894e-06,
      "loss": 0.3898,
      "step": 68
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.06521393358707428,
      "learning_rate": 9.906315640957117e-06,
      "loss": 0.424,
      "step": 69
    },
    {
      "epoch": 0.14492753623188406,
      "grad_norm": 0.07010532915592194,
      "learning_rate": 9.903123368623216e-06,
      "loss": 0.2962,
      "step": 70
    },
    {
      "epoch": 0.14699792960662525,
      "grad_norm": 0.08857309818267822,
      "learning_rate": 9.89987814745654e-06,
      "loss": 0.5392,
      "step": 71
    },
    {
      "epoch": 0.14906832298136646,
      "grad_norm": 0.11536813527345657,
      "learning_rate": 9.896580012502238e-06,
      "loss": 0.4399,
      "step": 72
    },
    {
      "epoch": 0.15113871635610765,
      "grad_norm": 0.13220086693763733,
      "learning_rate": 9.89322899937687e-06,
      "loss": 0.36,
      "step": 73
    },
    {
      "epoch": 0.15320910973084886,
      "grad_norm": 0.10927895456552505,
      "learning_rate": 9.889825144268029e-06,
      "loss": 0.6121,
      "step": 74
    },
    {
      "epoch": 0.15527950310559005,
      "grad_norm": 0.07254831492900848,
      "learning_rate": 9.88636848393395e-06,
      "loss": 0.3819,
      "step": 75
    },
    {
      "epoch": 0.15734989648033126,
      "grad_norm": 0.06788697838783264,
      "learning_rate": 9.882859055703109e-06,
      "loss": 0.2731,
      "step": 76
    },
    {
      "epoch": 0.15942028985507245,
      "grad_norm": 0.06452120095491409,
      "learning_rate": 9.879296897473825e-06,
      "loss": 0.261,
      "step": 77
    },
    {
      "epoch": 0.16149068322981366,
      "grad_norm": 0.078448086977005,
      "learning_rate": 9.875682047713847e-06,
      "loss": 0.3601,
      "step": 78
    },
    {
      "epoch": 0.16356107660455488,
      "grad_norm": 0.07225798815488815,
      "learning_rate": 9.87201454545994e-06,
      "loss": 0.4058,
      "step": 79
    },
    {
      "epoch": 0.16563146997929606,
      "grad_norm": 0.07582813501358032,
      "learning_rate": 9.868294430317464e-06,
      "loss": 0.3832,
      "step": 80
    },
    {
      "epoch": 0.16770186335403728,
      "grad_norm": 0.08351872116327286,
      "learning_rate": 9.864521742459943e-06,
      "loss": 0.4399,
      "step": 81
    },
    {
      "epoch": 0.16977225672877846,
      "grad_norm": 0.0709993913769722,
      "learning_rate": 9.860696522628638e-06,
      "loss": 0.474,
      "step": 82
    },
    {
      "epoch": 0.17184265010351968,
      "grad_norm": 0.06954403221607208,
      "learning_rate": 9.856818812132101e-06,
      "loss": 0.3555,
      "step": 83
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": 0.07412121444940567,
      "learning_rate": 9.852888652845729e-06,
      "loss": 0.3702,
      "step": 84
    },
    {
      "epoch": 0.17598343685300208,
      "grad_norm": 0.059929411858320236,
      "learning_rate": 9.848906087211317e-06,
      "loss": 0.335,
      "step": 85
    },
    {
      "epoch": 0.17805383022774326,
      "grad_norm": 0.08295787125825882,
      "learning_rate": 9.84487115823659e-06,
      "loss": 0.4468,
      "step": 86
    },
    {
      "epoch": 0.18012422360248448,
      "grad_norm": 0.10332628339529037,
      "learning_rate": 9.840783909494753e-06,
      "loss": 0.4893,
      "step": 87
    },
    {
      "epoch": 0.18219461697722567,
      "grad_norm": 0.06922546774148941,
      "learning_rate": 9.836644385124006e-06,
      "loss": 0.3423,
      "step": 88
    },
    {
      "epoch": 0.18426501035196688,
      "grad_norm": 0.0938820093870163,
      "learning_rate": 9.832452629827079e-06,
      "loss": 0.5032,
      "step": 89
    },
    {
      "epoch": 0.18633540372670807,
      "grad_norm": 0.08957293629646301,
      "learning_rate": 9.828208688870736e-06,
      "loss": 0.4738,
      "step": 90
    },
    {
      "epoch": 0.18840579710144928,
      "grad_norm": 0.08262480795383453,
      "learning_rate": 9.823912608085299e-06,
      "loss": 0.4135,
      "step": 91
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 0.10143536329269409,
      "learning_rate": 9.81956443386415e-06,
      "loss": 0.4859,
      "step": 92
    },
    {
      "epoch": 0.19254658385093168,
      "grad_norm": 0.08232709765434265,
      "learning_rate": 9.815164213163224e-06,
      "loss": 0.454,
      "step": 93
    },
    {
      "epoch": 0.19461697722567287,
      "grad_norm": 0.08575092256069183,
      "learning_rate": 9.810711993500506e-06,
      "loss": 0.377,
      "step": 94
    },
    {
      "epoch": 0.19668737060041408,
      "grad_norm": 0.08846532553434372,
      "learning_rate": 9.806207822955524e-06,
      "loss": 0.4189,
      "step": 95
    },
    {
      "epoch": 0.19875776397515527,
      "grad_norm": 0.07108867168426514,
      "learning_rate": 9.801651750168815e-06,
      "loss": 0.421,
      "step": 96
    },
    {
      "epoch": 0.20082815734989648,
      "grad_norm": 0.09010709822177887,
      "learning_rate": 9.797043824341413e-06,
      "loss": 0.4374,
      "step": 97
    },
    {
      "epoch": 0.2028985507246377,
      "grad_norm": 0.11712660640478134,
      "learning_rate": 9.792384095234312e-06,
      "loss": 0.4972,
      "step": 98
    },
    {
      "epoch": 0.20496894409937888,
      "grad_norm": 0.10217002779245377,
      "learning_rate": 9.787672613167931e-06,
      "loss": 0.5974,
      "step": 99
    },
    {
      "epoch": 0.2070393374741201,
      "grad_norm": 0.0857044979929924,
      "learning_rate": 9.782909429021568e-06,
      "loss": 0.507,
      "step": 100
    },
    {
      "epoch": 0.20910973084886128,
      "grad_norm": 0.08360479027032852,
      "learning_rate": 9.778094594232853e-06,
      "loss": 0.452,
      "step": 101
    },
    {
      "epoch": 0.2111801242236025,
      "grad_norm": 0.08019517362117767,
      "learning_rate": 9.773228160797187e-06,
      "loss": 0.4235,
      "step": 102
    },
    {
      "epoch": 0.21325051759834368,
      "grad_norm": 0.08056485652923584,
      "learning_rate": 9.76831018126719e-06,
      "loss": 0.453,
      "step": 103
    },
    {
      "epoch": 0.2153209109730849,
      "grad_norm": 0.06662017107009888,
      "learning_rate": 9.76334070875213e-06,
      "loss": 0.3214,
      "step": 104
    },
    {
      "epoch": 0.21739130434782608,
      "grad_norm": 0.1045277863740921,
      "learning_rate": 9.75831979691734e-06,
      "loss": 0.3387,
      "step": 105
    },
    {
      "epoch": 0.2194616977225673,
      "grad_norm": 0.0656440332531929,
      "learning_rate": 9.753247499983649e-06,
      "loss": 0.3272,
      "step": 106
    },
    {
      "epoch": 0.22153209109730848,
      "grad_norm": 0.09750998765230179,
      "learning_rate": 9.748123872726801e-06,
      "loss": 0.4563,
      "step": 107
    },
    {
      "epoch": 0.2236024844720497,
      "grad_norm": 0.09482847899198532,
      "learning_rate": 9.742948970476845e-06,
      "loss": 0.4397,
      "step": 108
    },
    {
      "epoch": 0.22567287784679088,
      "grad_norm": 0.0843338593840599,
      "learning_rate": 9.737722849117556e-06,
      "loss": 0.3945,
      "step": 109
    },
    {
      "epoch": 0.2277432712215321,
      "grad_norm": 0.07675352692604065,
      "learning_rate": 9.732445565085823e-06,
      "loss": 0.3514,
      "step": 110
    },
    {
      "epoch": 0.22981366459627328,
      "grad_norm": 0.09576563537120819,
      "learning_rate": 9.72711717537104e-06,
      "loss": 0.3179,
      "step": 111
    },
    {
      "epoch": 0.2318840579710145,
      "grad_norm": 0.09575846046209335,
      "learning_rate": 9.721737737514492e-06,
      "loss": 0.3947,
      "step": 112
    },
    {
      "epoch": 0.23395445134575568,
      "grad_norm": 0.06680424511432648,
      "learning_rate": 9.71630730960873e-06,
      "loss": 0.2413,
      "step": 113
    },
    {
      "epoch": 0.2360248447204969,
      "grad_norm": 0.09097225219011307,
      "learning_rate": 9.71082595029695e-06,
      "loss": 0.3346,
      "step": 114
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 0.08269556611776352,
      "learning_rate": 9.705293718772355e-06,
      "loss": 0.3557,
      "step": 115
    },
    {
      "epoch": 0.2401656314699793,
      "grad_norm": 0.06669075042009354,
      "learning_rate": 9.699710674777519e-06,
      "loss": 0.2507,
      "step": 116
    },
    {
      "epoch": 0.2422360248447205,
      "grad_norm": 0.0904952734708786,
      "learning_rate": 9.694076878603736e-06,
      "loss": 0.4732,
      "step": 117
    },
    {
      "epoch": 0.2443064182194617,
      "grad_norm": 0.08195852488279343,
      "learning_rate": 9.688392391090374e-06,
      "loss": 0.3463,
      "step": 118
    },
    {
      "epoch": 0.2463768115942029,
      "grad_norm": 0.0797749012708664,
      "learning_rate": 9.682657273624221e-06,
      "loss": 0.417,
      "step": 119
    },
    {
      "epoch": 0.2484472049689441,
      "grad_norm": 0.07914768904447556,
      "learning_rate": 9.676871588138812e-06,
      "loss": 0.3735,
      "step": 120
    },
    {
      "epoch": 0.2505175983436853,
      "grad_norm": 0.08743813633918762,
      "learning_rate": 9.671035397113769e-06,
      "loss": 0.4272,
      "step": 121
    },
    {
      "epoch": 0.2505175983436853,
      "eval_loss": 0.44131994247436523,
      "eval_runtime": 93.4201,
      "eval_samples_per_second": 5.909,
      "eval_steps_per_second": 0.739,
      "step": 121
    },
    {
      "epoch": 0.2525879917184265,
      "grad_norm": 0.07742875814437866,
      "learning_rate": 9.665148763574123e-06,
      "loss": 0.3742,
      "step": 122
    },
    {
      "epoch": 0.2546583850931677,
      "grad_norm": 0.07875270396471024,
      "learning_rate": 9.659211751089636e-06,
      "loss": 0.2702,
      "step": 123
    },
    {
      "epoch": 0.2567287784679089,
      "grad_norm": 0.08928168565034866,
      "learning_rate": 9.653224423774107e-06,
      "loss": 0.399,
      "step": 124
    },
    {
      "epoch": 0.2587991718426501,
      "grad_norm": 0.10149750858545303,
      "learning_rate": 9.647186846284692e-06,
      "loss": 0.5301,
      "step": 125
    },
    {
      "epoch": 0.2608695652173913,
      "grad_norm": 0.12193959951400757,
      "learning_rate": 9.64109908382119e-06,
      "loss": 0.6185,
      "step": 126
    },
    {
      "epoch": 0.2629399585921325,
      "grad_norm": 0.0960773155093193,
      "learning_rate": 9.634961202125355e-06,
      "loss": 0.4239,
      "step": 127
    },
    {
      "epoch": 0.2650103519668737,
      "grad_norm": 0.10699175298213959,
      "learning_rate": 9.628773267480177e-06,
      "loss": 0.3386,
      "step": 128
    },
    {
      "epoch": 0.2670807453416149,
      "grad_norm": 0.08573795109987259,
      "learning_rate": 9.622535346709161e-06,
      "loss": 0.3367,
      "step": 129
    },
    {
      "epoch": 0.2691511387163561,
      "grad_norm": 0.09970315545797348,
      "learning_rate": 9.616247507175624e-06,
      "loss": 0.4698,
      "step": 130
    },
    {
      "epoch": 0.2712215320910973,
      "grad_norm": 0.07994365692138672,
      "learning_rate": 9.609909816781945e-06,
      "loss": 0.3454,
      "step": 131
    },
    {
      "epoch": 0.2732919254658385,
      "grad_norm": 0.07029321044683456,
      "learning_rate": 9.603522343968852e-06,
      "loss": 0.3227,
      "step": 132
    },
    {
      "epoch": 0.2753623188405797,
      "grad_norm": 0.10029646009206772,
      "learning_rate": 9.597085157714666e-06,
      "loss": 0.4583,
      "step": 133
    },
    {
      "epoch": 0.2774327122153209,
      "grad_norm": 0.10204445570707321,
      "learning_rate": 9.590598327534563e-06,
      "loss": 0.5137,
      "step": 134
    },
    {
      "epoch": 0.2795031055900621,
      "grad_norm": 0.09162542223930359,
      "learning_rate": 9.584061923479833e-06,
      "loss": 0.413,
      "step": 135
    },
    {
      "epoch": 0.2815734989648033,
      "grad_norm": 0.08620335161685944,
      "learning_rate": 9.577476016137105e-06,
      "loss": 0.4518,
      "step": 136
    },
    {
      "epoch": 0.2836438923395445,
      "grad_norm": 0.07258706539869308,
      "learning_rate": 9.570840676627593e-06,
      "loss": 0.3211,
      "step": 137
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.10509861260652542,
      "learning_rate": 9.56415597660634e-06,
      "loss": 0.4713,
      "step": 138
    },
    {
      "epoch": 0.28778467908902694,
      "grad_norm": 0.09548232704401016,
      "learning_rate": 9.557421988261422e-06,
      "loss": 0.4447,
      "step": 139
    },
    {
      "epoch": 0.2898550724637681,
      "grad_norm": 0.08879134804010391,
      "learning_rate": 9.550638784313187e-06,
      "loss": 0.4448,
      "step": 140
    },
    {
      "epoch": 0.2919254658385093,
      "grad_norm": 0.07666925340890884,
      "learning_rate": 9.543806438013454e-06,
      "loss": 0.3178,
      "step": 141
    },
    {
      "epoch": 0.2939958592132505,
      "grad_norm": 0.15115399658679962,
      "learning_rate": 9.536925023144742e-06,
      "loss": 0.4872,
      "step": 142
    },
    {
      "epoch": 0.29606625258799174,
      "grad_norm": 0.08174444735050201,
      "learning_rate": 9.52999461401945e-06,
      "loss": 0.2779,
      "step": 143
    },
    {
      "epoch": 0.2981366459627329,
      "grad_norm": 0.0791732594370842,
      "learning_rate": 9.523015285479076e-06,
      "loss": 0.3226,
      "step": 144
    },
    {
      "epoch": 0.3002070393374741,
      "grad_norm": 0.10522507131099701,
      "learning_rate": 9.51598711289339e-06,
      "loss": 0.4394,
      "step": 145
    },
    {
      "epoch": 0.3022774327122153,
      "grad_norm": 0.08529925346374512,
      "learning_rate": 9.508910172159635e-06,
      "loss": 0.3471,
      "step": 146
    },
    {
      "epoch": 0.30434782608695654,
      "grad_norm": 0.11986136436462402,
      "learning_rate": 9.501784539701697e-06,
      "loss": 0.359,
      "step": 147
    },
    {
      "epoch": 0.3064182194616977,
      "grad_norm": 0.08316604793071747,
      "learning_rate": 9.494610292469287e-06,
      "loss": 0.3872,
      "step": 148
    },
    {
      "epoch": 0.3084886128364389,
      "grad_norm": 0.07519234716892242,
      "learning_rate": 9.487387507937106e-06,
      "loss": 0.3103,
      "step": 149
    },
    {
      "epoch": 0.3105590062111801,
      "grad_norm": 0.11129677295684814,
      "learning_rate": 9.48011626410401e-06,
      "loss": 0.5156,
      "step": 150
    },
    {
      "epoch": 0.31262939958592134,
      "grad_norm": 0.11108915507793427,
      "learning_rate": 9.472796639492166e-06,
      "loss": 0.365,
      "step": 151
    },
    {
      "epoch": 0.3146997929606625,
      "grad_norm": 0.0829719603061676,
      "learning_rate": 9.465428713146206e-06,
      "loss": 0.4079,
      "step": 152
    },
    {
      "epoch": 0.3167701863354037,
      "grad_norm": 0.10608082264661789,
      "learning_rate": 9.45801256463237e-06,
      "loss": 0.4702,
      "step": 153
    },
    {
      "epoch": 0.3188405797101449,
      "grad_norm": 0.08056455105543137,
      "learning_rate": 9.450548274037652e-06,
      "loss": 0.2567,
      "step": 154
    },
    {
      "epoch": 0.32091097308488614,
      "grad_norm": 0.12216547131538391,
      "learning_rate": 9.443035921968932e-06,
      "loss": 0.4258,
      "step": 155
    },
    {
      "epoch": 0.32298136645962733,
      "grad_norm": 0.10286413133144379,
      "learning_rate": 9.435475589552107e-06,
      "loss": 0.3515,
      "step": 156
    },
    {
      "epoch": 0.3250517598343685,
      "grad_norm": 0.10109732300043106,
      "learning_rate": 9.427867358431209e-06,
      "loss": 0.4022,
      "step": 157
    },
    {
      "epoch": 0.32712215320910976,
      "grad_norm": 0.09029930084943771,
      "learning_rate": 9.420211310767534e-06,
      "loss": 0.2505,
      "step": 158
    },
    {
      "epoch": 0.32919254658385094,
      "grad_norm": 0.13366563618183136,
      "learning_rate": 9.412507529238741e-06,
      "loss": 0.4932,
      "step": 159
    },
    {
      "epoch": 0.33126293995859213,
      "grad_norm": 0.08471870422363281,
      "learning_rate": 9.40475609703798e-06,
      "loss": 0.3423,
      "step": 160
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.09047792106866837,
      "learning_rate": 9.396957097872967e-06,
      "loss": 0.2468,
      "step": 161
    },
    {
      "epoch": 0.33540372670807456,
      "grad_norm": 0.09556887298822403,
      "learning_rate": 9.389110615965102e-06,
      "loss": 0.3965,
      "step": 162
    },
    {
      "epoch": 0.33747412008281574,
      "grad_norm": 0.10276677459478378,
      "learning_rate": 9.38121673604855e-06,
      "loss": 0.4059,
      "step": 163
    },
    {
      "epoch": 0.33954451345755693,
      "grad_norm": 0.11112170666456223,
      "learning_rate": 9.37327554336932e-06,
      "loss": 0.3625,
      "step": 164
    },
    {
      "epoch": 0.3416149068322981,
      "grad_norm": 0.12287746369838715,
      "learning_rate": 9.365287123684365e-06,
      "loss": 0.3374,
      "step": 165
    },
    {
      "epoch": 0.34368530020703936,
      "grad_norm": 0.12305274605751038,
      "learning_rate": 9.35725156326063e-06,
      "loss": 0.3823,
      "step": 166
    },
    {
      "epoch": 0.34575569358178054,
      "grad_norm": 0.10408051311969757,
      "learning_rate": 9.34916894887414e-06,
      "loss": 0.4478,
      "step": 167
    },
    {
      "epoch": 0.34782608695652173,
      "grad_norm": 0.09180740267038345,
      "learning_rate": 9.341039367809056e-06,
      "loss": 0.3882,
      "step": 168
    },
    {
      "epoch": 0.3498964803312629,
      "grad_norm": 0.09142342954874039,
      "learning_rate": 9.332862907856722e-06,
      "loss": 0.312,
      "step": 169
    },
    {
      "epoch": 0.35196687370600416,
      "grad_norm": 0.14173369109630585,
      "learning_rate": 9.324639657314742e-06,
      "loss": 0.4436,
      "step": 170
    },
    {
      "epoch": 0.35403726708074534,
      "grad_norm": 0.12495898455381393,
      "learning_rate": 9.316369704986001e-06,
      "loss": 0.532,
      "step": 171
    },
    {
      "epoch": 0.35610766045548653,
      "grad_norm": 0.08666636794805527,
      "learning_rate": 9.308053140177722e-06,
      "loss": 0.3174,
      "step": 172
    },
    {
      "epoch": 0.3581780538302277,
      "grad_norm": 0.1289636641740799,
      "learning_rate": 9.299690052700492e-06,
      "loss": 0.3586,
      "step": 173
    },
    {
      "epoch": 0.36024844720496896,
      "grad_norm": 0.11257122457027435,
      "learning_rate": 9.291280532867301e-06,
      "loss": 0.48,
      "step": 174
    },
    {
      "epoch": 0.36231884057971014,
      "grad_norm": 0.12760891020298004,
      "learning_rate": 9.28282467149256e-06,
      "loss": 0.5194,
      "step": 175
    },
    {
      "epoch": 0.36438923395445133,
      "grad_norm": 0.1176360547542572,
      "learning_rate": 9.27432255989112e-06,
      "loss": 0.5659,
      "step": 176
    },
    {
      "epoch": 0.36645962732919257,
      "grad_norm": 0.10507313907146454,
      "learning_rate": 9.265774289877291e-06,
      "loss": 0.3186,
      "step": 177
    },
    {
      "epoch": 0.36853002070393376,
      "grad_norm": 0.08758605271577835,
      "learning_rate": 9.257179953763846e-06,
      "loss": 0.2769,
      "step": 178
    },
    {
      "epoch": 0.37060041407867494,
      "grad_norm": 0.10309769958257675,
      "learning_rate": 9.248539644361029e-06,
      "loss": 0.3603,
      "step": 179
    },
    {
      "epoch": 0.37267080745341613,
      "grad_norm": 0.08716636151075363,
      "learning_rate": 9.239853454975548e-06,
      "loss": 0.3524,
      "step": 180
    },
    {
      "epoch": 0.3747412008281574,
      "grad_norm": 0.09289219230413437,
      "learning_rate": 9.231121479409567e-06,
      "loss": 0.2974,
      "step": 181
    },
    {
      "epoch": 0.37681159420289856,
      "grad_norm": 0.12185970693826675,
      "learning_rate": 9.222343811959694e-06,
      "loss": 0.5011,
      "step": 182
    },
    {
      "epoch": 0.37888198757763975,
      "grad_norm": 0.16153395175933838,
      "learning_rate": 9.213520547415968e-06,
      "loss": 0.3844,
      "step": 183
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 0.09068119525909424,
      "learning_rate": 9.204651781060832e-06,
      "loss": 0.2212,
      "step": 184
    },
    {
      "epoch": 0.3830227743271222,
      "grad_norm": 0.10143032670021057,
      "learning_rate": 9.195737608668096e-06,
      "loss": 0.3724,
      "step": 185
    },
    {
      "epoch": 0.38509316770186336,
      "grad_norm": 0.14434686303138733,
      "learning_rate": 9.186778126501916e-06,
      "loss": 0.5017,
      "step": 186
    },
    {
      "epoch": 0.38716356107660455,
      "grad_norm": 0.0935361459851265,
      "learning_rate": 9.177773431315748e-06,
      "loss": 0.2123,
      "step": 187
    },
    {
      "epoch": 0.38923395445134573,
      "grad_norm": 0.09748780727386475,
      "learning_rate": 9.168723620351298e-06,
      "loss": 0.373,
      "step": 188
    },
    {
      "epoch": 0.391304347826087,
      "grad_norm": 0.10403398424386978,
      "learning_rate": 9.159628791337483e-06,
      "loss": 0.2962,
      "step": 189
    },
    {
      "epoch": 0.39337474120082816,
      "grad_norm": 0.11641765385866165,
      "learning_rate": 9.150489042489368e-06,
      "loss": 0.4867,
      "step": 190
    },
    {
      "epoch": 0.39544513457556935,
      "grad_norm": 0.09807530045509338,
      "learning_rate": 9.141304472507109e-06,
      "loss": 0.3982,
      "step": 191
    },
    {
      "epoch": 0.39751552795031053,
      "grad_norm": 0.1166764423251152,
      "learning_rate": 9.13207518057488e-06,
      "loss": 0.384,
      "step": 192
    },
    {
      "epoch": 0.3995859213250518,
      "grad_norm": 0.11490147560834885,
      "learning_rate": 9.122801266359815e-06,
      "loss": 0.4438,
      "step": 193
    },
    {
      "epoch": 0.40165631469979296,
      "grad_norm": 0.1409396082162857,
      "learning_rate": 9.113482830010918e-06,
      "loss": 0.5902,
      "step": 194
    },
    {
      "epoch": 0.40372670807453415,
      "grad_norm": 0.1281116008758545,
      "learning_rate": 9.10411997215799e-06,
      "loss": 0.329,
      "step": 195
    },
    {
      "epoch": 0.4057971014492754,
      "grad_norm": 0.1288306564092636,
      "learning_rate": 9.094712793910541e-06,
      "loss": 0.4421,
      "step": 196
    },
    {
      "epoch": 0.4078674948240166,
      "grad_norm": 0.1229570135474205,
      "learning_rate": 9.085261396856694e-06,
      "loss": 0.5416,
      "step": 197
    },
    {
      "epoch": 0.40993788819875776,
      "grad_norm": 0.09709765762090683,
      "learning_rate": 9.075765883062093e-06,
      "loss": 0.359,
      "step": 198
    },
    {
      "epoch": 0.41200828157349895,
      "grad_norm": 0.12015578895807266,
      "learning_rate": 9.0662263550688e-06,
      "loss": 0.3323,
      "step": 199
    },
    {
      "epoch": 0.4140786749482402,
      "grad_norm": 0.09695754200220108,
      "learning_rate": 9.056642915894182e-06,
      "loss": 0.3129,
      "step": 200
    },
    {
      "epoch": 0.4161490683229814,
      "grad_norm": 0.09204108268022537,
      "learning_rate": 9.047015669029808e-06,
      "loss": 0.2289,
      "step": 201
    },
    {
      "epoch": 0.41821946169772256,
      "grad_norm": 0.11979050934314728,
      "learning_rate": 9.037344718440321e-06,
      "loss": 0.3477,
      "step": 202
    },
    {
      "epoch": 0.42028985507246375,
      "grad_norm": 0.14150062203407288,
      "learning_rate": 9.027630168562326e-06,
      "loss": 0.5484,
      "step": 203
    },
    {
      "epoch": 0.422360248447205,
      "grad_norm": 0.12129364162683487,
      "learning_rate": 9.017872124303255e-06,
      "loss": 0.4362,
      "step": 204
    },
    {
      "epoch": 0.4244306418219462,
      "grad_norm": 0.09622828662395477,
      "learning_rate": 9.00807069104023e-06,
      "loss": 0.2868,
      "step": 205
    },
    {
      "epoch": 0.42650103519668736,
      "grad_norm": 0.11513911187648773,
      "learning_rate": 8.99822597461894e-06,
      "loss": 0.4002,
      "step": 206
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.11842213571071625,
      "learning_rate": 8.988338081352484e-06,
      "loss": 0.3668,
      "step": 207
    },
    {
      "epoch": 0.4306418219461698,
      "grad_norm": 0.10251864790916443,
      "learning_rate": 8.978407118020226e-06,
      "loss": 0.3032,
      "step": 208
    },
    {
      "epoch": 0.432712215320911,
      "grad_norm": 0.12288280576467514,
      "learning_rate": 8.96843319186665e-06,
      "loss": 0.4732,
      "step": 209
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": 0.1105843186378479,
      "learning_rate": 8.958416410600188e-06,
      "loss": 0.3651,
      "step": 210
    },
    {
      "epoch": 0.43685300207039335,
      "grad_norm": 0.09893371909856796,
      "learning_rate": 8.948356882392072e-06,
      "loss": 0.4154,
      "step": 211
    },
    {
      "epoch": 0.4389233954451346,
      "grad_norm": 0.12337377667427063,
      "learning_rate": 8.938254715875152e-06,
      "loss": 0.4027,
      "step": 212
    },
    {
      "epoch": 0.4409937888198758,
      "grad_norm": 0.12832386791706085,
      "learning_rate": 8.928110020142729e-06,
      "loss": 0.4879,
      "step": 213
    },
    {
      "epoch": 0.44306418219461696,
      "grad_norm": 0.13107864558696747,
      "learning_rate": 8.917922904747385e-06,
      "loss": 0.3745,
      "step": 214
    },
    {
      "epoch": 0.4451345755693582,
      "grad_norm": 0.11851947754621506,
      "learning_rate": 8.907693479699783e-06,
      "loss": 0.2237,
      "step": 215
    },
    {
      "epoch": 0.4472049689440994,
      "grad_norm": 0.10010041296482086,
      "learning_rate": 8.897421855467491e-06,
      "loss": 0.3745,
      "step": 216
    },
    {
      "epoch": 0.4492753623188406,
      "grad_norm": 0.1251843422651291,
      "learning_rate": 8.887108142973788e-06,
      "loss": 0.4469,
      "step": 217
    },
    {
      "epoch": 0.45134575569358176,
      "grad_norm": 0.08511991798877716,
      "learning_rate": 8.876752453596462e-06,
      "loss": 0.2385,
      "step": 218
    },
    {
      "epoch": 0.453416149068323,
      "grad_norm": 0.14126259088516235,
      "learning_rate": 8.86635489916661e-06,
      "loss": 0.5021,
      "step": 219
    },
    {
      "epoch": 0.4554865424430642,
      "grad_norm": 0.12813101708889008,
      "learning_rate": 8.85591559196743e-06,
      "loss": 0.2839,
      "step": 220
    },
    {
      "epoch": 0.4575569358178054,
      "grad_norm": 0.09878244251012802,
      "learning_rate": 8.845434644733009e-06,
      "loss": 0.2601,
      "step": 221
    },
    {
      "epoch": 0.45962732919254656,
      "grad_norm": 0.12693674862384796,
      "learning_rate": 8.834912170647102e-06,
      "loss": 0.3788,
      "step": 222
    },
    {
      "epoch": 0.4616977225672878,
      "grad_norm": 0.14547109603881836,
      "learning_rate": 8.824348283341912e-06,
      "loss": 0.5055,
      "step": 223
    },
    {
      "epoch": 0.463768115942029,
      "grad_norm": 0.1215929314494133,
      "learning_rate": 8.813743096896872e-06,
      "loss": 0.3685,
      "step": 224
    },
    {
      "epoch": 0.4658385093167702,
      "grad_norm": 0.14049437642097473,
      "learning_rate": 8.803096725837392e-06,
      "loss": 0.4663,
      "step": 225
    },
    {
      "epoch": 0.46790890269151136,
      "grad_norm": 0.12880191206932068,
      "learning_rate": 8.792409285133644e-06,
      "loss": 0.3377,
      "step": 226
    },
    {
      "epoch": 0.4699792960662526,
      "grad_norm": 0.1347607672214508,
      "learning_rate": 8.781680890199306e-06,
      "loss": 0.4076,
      "step": 227
    },
    {
      "epoch": 0.4720496894409938,
      "grad_norm": 0.12177550792694092,
      "learning_rate": 8.770911656890325e-06,
      "loss": 0.3779,
      "step": 228
    },
    {
      "epoch": 0.474120082815735,
      "grad_norm": 0.15446029603481293,
      "learning_rate": 8.760101701503656e-06,
      "loss": 0.3923,
      "step": 229
    },
    {
      "epoch": 0.47619047619047616,
      "grad_norm": 0.11769753694534302,
      "learning_rate": 8.749251140776016e-06,
      "loss": 0.3247,
      "step": 230
    },
    {
      "epoch": 0.4782608695652174,
      "grad_norm": 0.14402779936790466,
      "learning_rate": 8.73836009188262e-06,
      "loss": 0.3048,
      "step": 231
    },
    {
      "epoch": 0.4803312629399586,
      "grad_norm": 0.14113663136959076,
      "learning_rate": 8.727428672435911e-06,
      "loss": 0.2981,
      "step": 232
    },
    {
      "epoch": 0.4824016563146998,
      "grad_norm": 0.15857934951782227,
      "learning_rate": 8.716457000484296e-06,
      "loss": 0.6381,
      "step": 233
    },
    {
      "epoch": 0.484472049689441,
      "grad_norm": 0.12054027616977692,
      "learning_rate": 8.705445194510868e-06,
      "loss": 0.3409,
      "step": 234
    },
    {
      "epoch": 0.4865424430641822,
      "grad_norm": 0.104091577231884,
      "learning_rate": 8.694393373432129e-06,
      "loss": 0.3034,
      "step": 235
    },
    {
      "epoch": 0.4886128364389234,
      "grad_norm": 0.10336735844612122,
      "learning_rate": 8.6833016565967e-06,
      "loss": 0.1919,
      "step": 236
    },
    {
      "epoch": 0.4906832298136646,
      "grad_norm": 0.13413453102111816,
      "learning_rate": 8.672170163784042e-06,
      "loss": 0.4726,
      "step": 237
    },
    {
      "epoch": 0.4927536231884058,
      "grad_norm": 0.15956872701644897,
      "learning_rate": 8.660999015203152e-06,
      "loss": 0.4549,
      "step": 238
    },
    {
      "epoch": 0.494824016563147,
      "grad_norm": 0.10782767832279205,
      "learning_rate": 8.64978833149127e-06,
      "loss": 0.2556,
      "step": 239
    },
    {
      "epoch": 0.4968944099378882,
      "grad_norm": 0.11683110892772675,
      "learning_rate": 8.638538233712581e-06,
      "loss": 0.3507,
      "step": 240
    },
    {
      "epoch": 0.4989648033126294,
      "grad_norm": 0.1023135557770729,
      "learning_rate": 8.6272488433569e-06,
      "loss": 0.2694,
      "step": 241
    },
    {
      "epoch": 0.5010351966873706,
      "grad_norm": 0.11874102801084518,
      "learning_rate": 8.615920282338355e-06,
      "loss": 0.3197,
      "step": 242
    },
    {
      "epoch": 0.5010351966873706,
      "eval_loss": 0.4105847179889679,
      "eval_runtime": 93.4074,
      "eval_samples_per_second": 5.91,
      "eval_steps_per_second": 0.739,
      "step": 242
    },
    {
      "epoch": 0.5031055900621118,
      "grad_norm": 0.1680690497159958,
      "learning_rate": 8.60455267299409e-06,
      "loss": 0.4984,
      "step": 243
    },
    {
      "epoch": 0.505175983436853,
      "grad_norm": 0.1034797728061676,
      "learning_rate": 8.593146138082925e-06,
      "loss": 0.258,
      "step": 244
    },
    {
      "epoch": 0.5072463768115942,
      "grad_norm": 0.09805410355329514,
      "learning_rate": 8.581700800784038e-06,
      "loss": 0.2416,
      "step": 245
    },
    {
      "epoch": 0.5093167701863354,
      "grad_norm": 0.11391060799360275,
      "learning_rate": 8.570216784695637e-06,
      "loss": 0.3188,
      "step": 246
    },
    {
      "epoch": 0.5113871635610766,
      "grad_norm": 0.12456899881362915,
      "learning_rate": 8.558694213833618e-06,
      "loss": 0.3211,
      "step": 247
    },
    {
      "epoch": 0.5134575569358178,
      "grad_norm": 0.10828563570976257,
      "learning_rate": 8.54713321263023e-06,
      "loss": 0.221,
      "step": 248
    },
    {
      "epoch": 0.515527950310559,
      "grad_norm": 0.09911594539880753,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.2517,
      "step": 249
    },
    {
      "epoch": 0.5175983436853002,
      "grad_norm": 0.14771407842636108,
      "learning_rate": 8.52389641900206e-06,
      "loss": 0.3483,
      "step": 250
    },
    {
      "epoch": 0.5196687370600414,
      "grad_norm": 0.13070322573184967,
      "learning_rate": 8.512220877511428e-06,
      "loss": 0.3345,
      "step": 251
    },
    {
      "epoch": 0.5217391304347826,
      "grad_norm": 0.18328367173671722,
      "learning_rate": 8.50050740754502e-06,
      "loss": 0.4107,
      "step": 252
    },
    {
      "epoch": 0.5238095238095238,
      "grad_norm": 0.11168783158063889,
      "learning_rate": 8.488756135596609e-06,
      "loss": 0.2455,
      "step": 253
    },
    {
      "epoch": 0.525879917184265,
      "grad_norm": 0.16458994150161743,
      "learning_rate": 8.476967188568187e-06,
      "loss": 0.4157,
      "step": 254
    },
    {
      "epoch": 0.5279503105590062,
      "grad_norm": 0.14378412067890167,
      "learning_rate": 8.465140693768606e-06,
      "loss": 0.4946,
      "step": 255
    },
    {
      "epoch": 0.5300207039337475,
      "grad_norm": 0.14328721165657043,
      "learning_rate": 8.453276778912186e-06,
      "loss": 0.3925,
      "step": 256
    },
    {
      "epoch": 0.5320910973084886,
      "grad_norm": 0.12056344002485275,
      "learning_rate": 8.441375572117356e-06,
      "loss": 0.2878,
      "step": 257
    },
    {
      "epoch": 0.5341614906832298,
      "grad_norm": 0.115929014980793,
      "learning_rate": 8.429437201905254e-06,
      "loss": 0.2342,
      "step": 258
    },
    {
      "epoch": 0.5362318840579711,
      "grad_norm": 0.17692901194095612,
      "learning_rate": 8.41746179719835e-06,
      "loss": 0.3761,
      "step": 259
    },
    {
      "epoch": 0.5383022774327122,
      "grad_norm": 0.14847911894321442,
      "learning_rate": 8.405449487319049e-06,
      "loss": 0.4452,
      "step": 260
    },
    {
      "epoch": 0.5403726708074534,
      "grad_norm": 0.18752224743366241,
      "learning_rate": 8.393400401988293e-06,
      "loss": 0.313,
      "step": 261
    },
    {
      "epoch": 0.5424430641821946,
      "grad_norm": 0.18232041597366333,
      "learning_rate": 8.38131467132416e-06,
      "loss": 0.3092,
      "step": 262
    },
    {
      "epoch": 0.5445134575569358,
      "grad_norm": 0.12929639220237732,
      "learning_rate": 8.369192425840469e-06,
      "loss": 0.344,
      "step": 263
    },
    {
      "epoch": 0.546583850931677,
      "grad_norm": 0.1309351623058319,
      "learning_rate": 8.357033796445356e-06,
      "loss": 0.2635,
      "step": 264
    },
    {
      "epoch": 0.5486542443064182,
      "grad_norm": 0.14819861948490143,
      "learning_rate": 8.344838914439869e-06,
      "loss": 0.4324,
      "step": 265
    },
    {
      "epoch": 0.5507246376811594,
      "grad_norm": 0.10555935651063919,
      "learning_rate": 8.332607911516545e-06,
      "loss": 0.3072,
      "step": 266
    },
    {
      "epoch": 0.5527950310559007,
      "grad_norm": 0.12563839554786682,
      "learning_rate": 8.320340919757997e-06,
      "loss": 0.3399,
      "step": 267
    },
    {
      "epoch": 0.5548654244306418,
      "grad_norm": 0.14571739733219147,
      "learning_rate": 8.308038071635475e-06,
      "loss": 0.3716,
      "step": 268
    },
    {
      "epoch": 0.556935817805383,
      "grad_norm": 0.14644472301006317,
      "learning_rate": 8.295699500007447e-06,
      "loss": 0.3165,
      "step": 269
    },
    {
      "epoch": 0.5590062111801242,
      "grad_norm": 0.33376821875572205,
      "learning_rate": 8.283325338118154e-06,
      "loss": 0.4557,
      "step": 270
    },
    {
      "epoch": 0.5610766045548654,
      "grad_norm": 0.1606837511062622,
      "learning_rate": 8.27091571959618e-06,
      "loss": 0.4879,
      "step": 271
    },
    {
      "epoch": 0.5631469979296067,
      "grad_norm": 0.13942554593086243,
      "learning_rate": 8.258470778453005e-06,
      "loss": 0.4332,
      "step": 272
    },
    {
      "epoch": 0.5652173913043478,
      "grad_norm": 0.14843975007534027,
      "learning_rate": 8.245990649081559e-06,
      "loss": 0.317,
      "step": 273
    },
    {
      "epoch": 0.567287784679089,
      "grad_norm": 0.1546916365623474,
      "learning_rate": 8.233475466254766e-06,
      "loss": 0.4053,
      "step": 274
    },
    {
      "epoch": 0.5693581780538303,
      "grad_norm": 0.15177839994430542,
      "learning_rate": 8.220925365124092e-06,
      "loss": 0.3222,
      "step": 275
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.15528550744056702,
      "learning_rate": 8.208340481218094e-06,
      "loss": 0.4215,
      "step": 276
    },
    {
      "epoch": 0.5734989648033126,
      "grad_norm": 0.11311841756105423,
      "learning_rate": 8.195720950440937e-06,
      "loss": 0.295,
      "step": 277
    },
    {
      "epoch": 0.5755693581780539,
      "grad_norm": 0.22601701319217682,
      "learning_rate": 8.183066909070946e-06,
      "loss": 0.4662,
      "step": 278
    },
    {
      "epoch": 0.577639751552795,
      "grad_norm": 0.10438232123851776,
      "learning_rate": 8.170378493759122e-06,
      "loss": 0.2396,
      "step": 279
    },
    {
      "epoch": 0.5797101449275363,
      "grad_norm": 0.09224749356508255,
      "learning_rate": 8.15765584152767e-06,
      "loss": 0.1875,
      "step": 280
    },
    {
      "epoch": 0.5817805383022774,
      "grad_norm": 0.13479924201965332,
      "learning_rate": 8.144899089768522e-06,
      "loss": 0.2768,
      "step": 281
    },
    {
      "epoch": 0.5838509316770186,
      "grad_norm": 0.17148271203041077,
      "learning_rate": 8.132108376241849e-06,
      "loss": 0.38,
      "step": 282
    },
    {
      "epoch": 0.5859213250517599,
      "grad_norm": 0.18072481453418732,
      "learning_rate": 8.119283839074573e-06,
      "loss": 0.4015,
      "step": 283
    },
    {
      "epoch": 0.587991718426501,
      "grad_norm": 0.14164257049560547,
      "learning_rate": 8.106425616758886e-06,
      "loss": 0.4656,
      "step": 284
    },
    {
      "epoch": 0.5900621118012422,
      "grad_norm": 0.19027261435985565,
      "learning_rate": 8.093533848150736e-06,
      "loss": 0.3513,
      "step": 285
    },
    {
      "epoch": 0.5921325051759835,
      "grad_norm": 0.12463188916444778,
      "learning_rate": 8.08060867246834e-06,
      "loss": 0.2498,
      "step": 286
    },
    {
      "epoch": 0.5942028985507246,
      "grad_norm": 0.14109396934509277,
      "learning_rate": 8.067650229290683e-06,
      "loss": 0.3645,
      "step": 287
    },
    {
      "epoch": 0.5962732919254659,
      "grad_norm": 0.1650514006614685,
      "learning_rate": 8.054658658555998e-06,
      "loss": 0.4303,
      "step": 288
    },
    {
      "epoch": 0.598343685300207,
      "grad_norm": 0.18131284415721893,
      "learning_rate": 8.04163410056027e-06,
      "loss": 0.2617,
      "step": 289
    },
    {
      "epoch": 0.6004140786749482,
      "grad_norm": 0.15095612406730652,
      "learning_rate": 8.028576695955711e-06,
      "loss": 0.4539,
      "step": 290
    },
    {
      "epoch": 0.6024844720496895,
      "grad_norm": 0.15879206359386444,
      "learning_rate": 8.01548658574924e-06,
      "loss": 0.3165,
      "step": 291
    },
    {
      "epoch": 0.6045548654244306,
      "grad_norm": 0.12368228286504745,
      "learning_rate": 8.002363911300966e-06,
      "loss": 0.2427,
      "step": 292
    },
    {
      "epoch": 0.6066252587991718,
      "grad_norm": 0.13043078780174255,
      "learning_rate": 7.989208814322662e-06,
      "loss": 0.2737,
      "step": 293
    },
    {
      "epoch": 0.6086956521739131,
      "grad_norm": 0.13512444496154785,
      "learning_rate": 7.976021436876232e-06,
      "loss": 0.3112,
      "step": 294
    },
    {
      "epoch": 0.6107660455486542,
      "grad_norm": 0.1622573733329773,
      "learning_rate": 7.962801921372168e-06,
      "loss": 0.4542,
      "step": 295
    },
    {
      "epoch": 0.6128364389233955,
      "grad_norm": 0.14550381898880005,
      "learning_rate": 7.949550410568033e-06,
      "loss": 0.2678,
      "step": 296
    },
    {
      "epoch": 0.6149068322981367,
      "grad_norm": 0.1805560141801834,
      "learning_rate": 7.936267047566897e-06,
      "loss": 0.5813,
      "step": 297
    },
    {
      "epoch": 0.6169772256728778,
      "grad_norm": 0.14014331996440887,
      "learning_rate": 7.92295197581581e-06,
      "loss": 0.3421,
      "step": 298
    },
    {
      "epoch": 0.6190476190476191,
      "grad_norm": 0.13757413625717163,
      "learning_rate": 7.909605339104243e-06,
      "loss": 0.342,
      "step": 299
    },
    {
      "epoch": 0.6211180124223602,
      "grad_norm": 0.09992540627717972,
      "learning_rate": 7.89622728156253e-06,
      "loss": 0.1582,
      "step": 300
    },
    {
      "epoch": 0.6231884057971014,
      "grad_norm": 0.15090250968933105,
      "learning_rate": 7.882817947660328e-06,
      "loss": 0.4279,
      "step": 301
    },
    {
      "epoch": 0.6252587991718427,
      "grad_norm": 0.22069942951202393,
      "learning_rate": 7.869377482205042e-06,
      "loss": 0.403,
      "step": 302
    },
    {
      "epoch": 0.6273291925465838,
      "grad_norm": 0.21121914684772491,
      "learning_rate": 7.855906030340268e-06,
      "loss": 0.3499,
      "step": 303
    },
    {
      "epoch": 0.629399585921325,
      "grad_norm": 0.16488689184188843,
      "learning_rate": 7.842403737544226e-06,
      "loss": 0.3166,
      "step": 304
    },
    {
      "epoch": 0.6314699792960663,
      "grad_norm": 0.14506235718727112,
      "learning_rate": 7.828870749628184e-06,
      "loss": 0.2906,
      "step": 305
    },
    {
      "epoch": 0.6335403726708074,
      "grad_norm": 0.15652890503406525,
      "learning_rate": 7.815307212734888e-06,
      "loss": 0.2071,
      "step": 306
    },
    {
      "epoch": 0.6356107660455487,
      "grad_norm": 0.17045624554157257,
      "learning_rate": 7.801713273336984e-06,
      "loss": 0.3238,
      "step": 307
    },
    {
      "epoch": 0.6376811594202898,
      "grad_norm": 0.12453598529100418,
      "learning_rate": 7.788089078235432e-06,
      "loss": 0.2382,
      "step": 308
    },
    {
      "epoch": 0.639751552795031,
      "grad_norm": 0.15407682955265045,
      "learning_rate": 7.774434774557926e-06,
      "loss": 0.3982,
      "step": 309
    },
    {
      "epoch": 0.6418219461697723,
      "grad_norm": 0.16916336119174957,
      "learning_rate": 7.7607505097573e-06,
      "loss": 0.3729,
      "step": 310
    },
    {
      "epoch": 0.6438923395445134,
      "grad_norm": 0.16933326423168182,
      "learning_rate": 7.747036431609938e-06,
      "loss": 0.3201,
      "step": 311
    },
    {
      "epoch": 0.6459627329192547,
      "grad_norm": 0.18653111159801483,
      "learning_rate": 7.733292688214182e-06,
      "loss": 0.4113,
      "step": 312
    },
    {
      "epoch": 0.6480331262939959,
      "grad_norm": 0.1700931340456009,
      "learning_rate": 7.719519427988728e-06,
      "loss": 0.2422,
      "step": 313
    },
    {
      "epoch": 0.650103519668737,
      "grad_norm": 0.2505623996257782,
      "learning_rate": 7.705716799671019e-06,
      "loss": 0.4215,
      "step": 314
    },
    {
      "epoch": 0.6521739130434783,
      "grad_norm": 0.1554984301328659,
      "learning_rate": 7.691884952315654e-06,
      "loss": 0.2661,
      "step": 315
    },
    {
      "epoch": 0.6542443064182195,
      "grad_norm": 0.1653880923986435,
      "learning_rate": 7.678024035292757e-06,
      "loss": 0.5224,
      "step": 316
    },
    {
      "epoch": 0.6563146997929606,
      "grad_norm": 0.19526076316833496,
      "learning_rate": 7.66413419828638e-06,
      "loss": 0.4749,
      "step": 317
    },
    {
      "epoch": 0.6583850931677019,
      "grad_norm": 0.11639489978551865,
      "learning_rate": 7.650215591292888e-06,
      "loss": 0.2833,
      "step": 318
    },
    {
      "epoch": 0.660455486542443,
      "grad_norm": 0.15517526865005493,
      "learning_rate": 7.636268364619328e-06,
      "loss": 0.2725,
      "step": 319
    },
    {
      "epoch": 0.6625258799171843,
      "grad_norm": 0.14740225672721863,
      "learning_rate": 7.622292668881805e-06,
      "loss": 0.2737,
      "step": 320
    },
    {
      "epoch": 0.6645962732919255,
      "grad_norm": 0.14966771006584167,
      "learning_rate": 7.608288655003872e-06,
      "loss": 0.3643,
      "step": 321
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.19022630155086517,
      "learning_rate": 7.594256474214883e-06,
      "loss": 0.4768,
      "step": 322
    },
    {
      "epoch": 0.6687370600414079,
      "grad_norm": 0.1434745341539383,
      "learning_rate": 7.580196278048368e-06,
      "loss": 0.443,
      "step": 323
    },
    {
      "epoch": 0.6708074534161491,
      "grad_norm": 0.16878576576709747,
      "learning_rate": 7.566108218340399e-06,
      "loss": 0.3245,
      "step": 324
    },
    {
      "epoch": 0.6728778467908902,
      "grad_norm": 0.12783066928386688,
      "learning_rate": 7.551992447227939e-06,
      "loss": 0.2576,
      "step": 325
    },
    {
      "epoch": 0.6749482401656315,
      "grad_norm": 0.131923645734787,
      "learning_rate": 7.537849117147212e-06,
      "loss": 0.3536,
      "step": 326
    },
    {
      "epoch": 0.6770186335403726,
      "grad_norm": 0.16058656573295593,
      "learning_rate": 7.523678380832049e-06,
      "loss": 0.3952,
      "step": 327
    },
    {
      "epoch": 0.6790890269151139,
      "grad_norm": 0.1566784381866455,
      "learning_rate": 7.509480391312243e-06,
      "loss": 0.4466,
      "step": 328
    },
    {
      "epoch": 0.6811594202898551,
      "grad_norm": 0.13159289956092834,
      "learning_rate": 7.4952553019118915e-06,
      "loss": 0.3165,
      "step": 329
    },
    {
      "epoch": 0.6832298136645962,
      "grad_norm": 0.17786771059036255,
      "learning_rate": 7.481003266247745e-06,
      "loss": 0.4053,
      "step": 330
    },
    {
      "epoch": 0.6853002070393375,
      "grad_norm": 0.09457248449325562,
      "learning_rate": 7.4667244382275486e-06,
      "loss": 0.1508,
      "step": 331
    },
    {
      "epoch": 0.6873706004140787,
      "grad_norm": 0.17693917453289032,
      "learning_rate": 7.452418972048372e-06,
      "loss": 0.3678,
      "step": 332
    },
    {
      "epoch": 0.6894409937888198,
      "grad_norm": 0.15471003949642181,
      "learning_rate": 7.4380870221949604e-06,
      "loss": 0.2698,
      "step": 333
    },
    {
      "epoch": 0.6915113871635611,
      "grad_norm": 0.14803005754947662,
      "learning_rate": 7.4237287434380485e-06,
      "loss": 0.2693,
      "step": 334
    },
    {
      "epoch": 0.6935817805383023,
      "grad_norm": 0.1645655333995819,
      "learning_rate": 7.4093442908326985e-06,
      "loss": 0.3928,
      "step": 335
    },
    {
      "epoch": 0.6956521739130435,
      "grad_norm": 0.17979875206947327,
      "learning_rate": 7.394933819716625e-06,
      "loss": 0.3584,
      "step": 336
    },
    {
      "epoch": 0.6977225672877847,
      "grad_norm": 0.1641678512096405,
      "learning_rate": 7.380497485708512e-06,
      "loss": 0.3704,
      "step": 337
    },
    {
      "epoch": 0.6997929606625258,
      "grad_norm": 0.1472330093383789,
      "learning_rate": 7.366035444706346e-06,
      "loss": 0.3375,
      "step": 338
    },
    {
      "epoch": 0.7018633540372671,
      "grad_norm": 0.13386555016040802,
      "learning_rate": 7.351547852885716e-06,
      "loss": 0.2683,
      "step": 339
    },
    {
      "epoch": 0.7039337474120083,
      "grad_norm": 0.17538785934448242,
      "learning_rate": 7.337034866698138e-06,
      "loss": 0.3281,
      "step": 340
    },
    {
      "epoch": 0.7060041407867494,
      "grad_norm": 0.23761358857154846,
      "learning_rate": 7.322496642869359e-06,
      "loss": 0.5206,
      "step": 341
    },
    {
      "epoch": 0.7080745341614907,
      "grad_norm": 0.14608746767044067,
      "learning_rate": 7.307933338397667e-06,
      "loss": 0.3623,
      "step": 342
    },
    {
      "epoch": 0.7101449275362319,
      "grad_norm": 0.15300914645195007,
      "learning_rate": 7.293345110552199e-06,
      "loss": 0.3471,
      "step": 343
    },
    {
      "epoch": 0.7122153209109731,
      "grad_norm": 0.5103980898857117,
      "learning_rate": 7.278732116871239e-06,
      "loss": 0.4771,
      "step": 344
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.11499185115098953,
      "learning_rate": 7.264094515160515e-06,
      "loss": 0.195,
      "step": 345
    },
    {
      "epoch": 0.7163561076604554,
      "grad_norm": 0.17889314889907837,
      "learning_rate": 7.249432463491498e-06,
      "loss": 0.4114,
      "step": 346
    },
    {
      "epoch": 0.7184265010351967,
      "grad_norm": 0.20027822256088257,
      "learning_rate": 7.234746120199695e-06,
      "loss": 0.3861,
      "step": 347
    },
    {
      "epoch": 0.7204968944099379,
      "grad_norm": 0.2676618695259094,
      "learning_rate": 7.220035643882938e-06,
      "loss": 0.3303,
      "step": 348
    },
    {
      "epoch": 0.722567287784679,
      "grad_norm": 0.14831817150115967,
      "learning_rate": 7.205301193399671e-06,
      "loss": 0.3329,
      "step": 349
    },
    {
      "epoch": 0.7246376811594203,
      "grad_norm": 0.15850083529949188,
      "learning_rate": 7.190542927867234e-06,
      "loss": 0.2805,
      "step": 350
    },
    {
      "epoch": 0.7267080745341615,
      "grad_norm": 0.17525608837604523,
      "learning_rate": 7.175761006660151e-06,
      "loss": 0.4265,
      "step": 351
    },
    {
      "epoch": 0.7287784679089027,
      "grad_norm": 0.22206392884254456,
      "learning_rate": 7.160955589408395e-06,
      "loss": 0.3882,
      "step": 352
    },
    {
      "epoch": 0.7308488612836439,
      "grad_norm": 0.27820146083831787,
      "learning_rate": 7.1461268359956806e-06,
      "loss": 0.4193,
      "step": 353
    },
    {
      "epoch": 0.7329192546583851,
      "grad_norm": 0.15169090032577515,
      "learning_rate": 7.131274906557725e-06,
      "loss": 0.2663,
      "step": 354
    },
    {
      "epoch": 0.7349896480331263,
      "grad_norm": 0.1922980397939682,
      "learning_rate": 7.1163999614805255e-06,
      "loss": 0.2922,
      "step": 355
    },
    {
      "epoch": 0.7370600414078675,
      "grad_norm": 0.14194855093955994,
      "learning_rate": 7.101502161398626e-06,
      "loss": 0.2633,
      "step": 356
    },
    {
      "epoch": 0.7391304347826086,
      "grad_norm": 0.12898112833499908,
      "learning_rate": 7.0865816671933765e-06,
      "loss": 0.1775,
      "step": 357
    },
    {
      "epoch": 0.7412008281573499,
      "grad_norm": 0.24564723670482635,
      "learning_rate": 7.0716386399912075e-06,
      "loss": 0.3405,
      "step": 358
    },
    {
      "epoch": 0.7432712215320911,
      "grad_norm": 0.1857512891292572,
      "learning_rate": 7.056673241161881e-06,
      "loss": 0.3334,
      "step": 359
    },
    {
      "epoch": 0.7453416149068323,
      "grad_norm": 0.12627875804901123,
      "learning_rate": 7.041685632316748e-06,
      "loss": 0.2766,
      "step": 360
    },
    {
      "epoch": 0.7474120082815735,
      "grad_norm": 0.22340427339076996,
      "learning_rate": 7.026675975307009e-06,
      "loss": 0.5199,
      "step": 361
    },
    {
      "epoch": 0.7494824016563147,
      "grad_norm": 0.17509430646896362,
      "learning_rate": 7.0116444322219575e-06,
      "loss": 0.3756,
      "step": 362
    },
    {
      "epoch": 0.7515527950310559,
      "grad_norm": 0.19094456732273102,
      "learning_rate": 6.996591165387241e-06,
      "loss": 0.4757,
      "step": 363
    },
    {
      "epoch": 0.7515527950310559,
      "eval_loss": 0.3904164433479309,
      "eval_runtime": 93.4751,
      "eval_samples_per_second": 5.905,
      "eval_steps_per_second": 0.738,
      "step": 363
    },
    {
      "epoch": 0.7536231884057971,
      "grad_norm": 0.21799920499324799,
      "learning_rate": 6.981516337363099e-06,
      "loss": 0.2783,
      "step": 364
    },
    {
      "epoch": 0.7556935817805382,
      "grad_norm": 0.17750999331474304,
      "learning_rate": 6.966420110942609e-06,
      "loss": 0.2854,
      "step": 365
    },
    {
      "epoch": 0.7577639751552795,
      "grad_norm": 0.16166891157627106,
      "learning_rate": 6.95130264914993e-06,
      "loss": 0.341,
      "step": 366
    },
    {
      "epoch": 0.7598343685300207,
      "grad_norm": 0.17185021936893463,
      "learning_rate": 6.936164115238543e-06,
      "loss": 0.353,
      "step": 367
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 0.16961365938186646,
      "learning_rate": 6.9210046726894885e-06,
      "loss": 0.3104,
      "step": 368
    },
    {
      "epoch": 0.7639751552795031,
      "grad_norm": 0.160708487033844,
      "learning_rate": 6.905824485209598e-06,
      "loss": 0.3879,
      "step": 369
    },
    {
      "epoch": 0.7660455486542443,
      "grad_norm": 0.12559418380260468,
      "learning_rate": 6.890623716729724e-06,
      "loss": 0.1893,
      "step": 370
    },
    {
      "epoch": 0.7681159420289855,
      "grad_norm": 0.1993858963251114,
      "learning_rate": 6.875402531402977e-06,
      "loss": 0.3177,
      "step": 371
    },
    {
      "epoch": 0.7701863354037267,
      "grad_norm": 0.1785745769739151,
      "learning_rate": 6.860161093602949e-06,
      "loss": 0.3831,
      "step": 372
    },
    {
      "epoch": 0.772256728778468,
      "grad_norm": 0.15908163785934448,
      "learning_rate": 6.844899567921938e-06,
      "loss": 0.2946,
      "step": 373
    },
    {
      "epoch": 0.7743271221532091,
      "grad_norm": 0.1833140105009079,
      "learning_rate": 6.829618119169169e-06,
      "loss": 0.3715,
      "step": 374
    },
    {
      "epoch": 0.7763975155279503,
      "grad_norm": 0.1361018717288971,
      "learning_rate": 6.814316912369021e-06,
      "loss": 0.2445,
      "step": 375
    },
    {
      "epoch": 0.7784679089026915,
      "grad_norm": 0.1777130365371704,
      "learning_rate": 6.798996112759233e-06,
      "loss": 0.3649,
      "step": 376
    },
    {
      "epoch": 0.7805383022774327,
      "grad_norm": 0.16926829516887665,
      "learning_rate": 6.783655885789136e-06,
      "loss": 0.2612,
      "step": 377
    },
    {
      "epoch": 0.782608695652174,
      "grad_norm": 0.1698054075241089,
      "learning_rate": 6.768296397117848e-06,
      "loss": 0.397,
      "step": 378
    },
    {
      "epoch": 0.7846790890269151,
      "grad_norm": 0.21158775687217712,
      "learning_rate": 6.7529178126125005e-06,
      "loss": 0.3234,
      "step": 379
    },
    {
      "epoch": 0.7867494824016563,
      "grad_norm": 0.16220830380916595,
      "learning_rate": 6.737520298346438e-06,
      "loss": 0.4353,
      "step": 380
    },
    {
      "epoch": 0.7888198757763976,
      "grad_norm": 0.16705025732517242,
      "learning_rate": 6.722104020597428e-06,
      "loss": 0.4326,
      "step": 381
    },
    {
      "epoch": 0.7908902691511387,
      "grad_norm": 0.18142729997634888,
      "learning_rate": 6.706669145845863e-06,
      "loss": 0.3197,
      "step": 382
    },
    {
      "epoch": 0.7929606625258799,
      "grad_norm": 0.17288783192634583,
      "learning_rate": 6.691215840772971e-06,
      "loss": 0.316,
      "step": 383
    },
    {
      "epoch": 0.7950310559006211,
      "grad_norm": 0.13447780907154083,
      "learning_rate": 6.6757442722590015e-06,
      "loss": 0.2254,
      "step": 384
    },
    {
      "epoch": 0.7971014492753623,
      "grad_norm": 0.16198629140853882,
      "learning_rate": 6.660254607381433e-06,
      "loss": 0.3463,
      "step": 385
    },
    {
      "epoch": 0.7991718426501035,
      "grad_norm": 0.16485463082790375,
      "learning_rate": 6.6447470134131685e-06,
      "loss": 0.2498,
      "step": 386
    },
    {
      "epoch": 0.8012422360248447,
      "grad_norm": 0.17164930701255798,
      "learning_rate": 6.629221657820729e-06,
      "loss": 0.2875,
      "step": 387
    },
    {
      "epoch": 0.8033126293995859,
      "grad_norm": 0.16868580877780914,
      "learning_rate": 6.613678708262439e-06,
      "loss": 0.2725,
      "step": 388
    },
    {
      "epoch": 0.8053830227743272,
      "grad_norm": 0.1734611988067627,
      "learning_rate": 6.598118332586619e-06,
      "loss": 0.3235,
      "step": 389
    },
    {
      "epoch": 0.8074534161490683,
      "grad_norm": 0.1865498423576355,
      "learning_rate": 6.5825406988297815e-06,
      "loss": 0.3326,
      "step": 390
    },
    {
      "epoch": 0.8095238095238095,
      "grad_norm": 0.18980611860752106,
      "learning_rate": 6.566945975214803e-06,
      "loss": 0.3541,
      "step": 391
    },
    {
      "epoch": 0.8115942028985508,
      "grad_norm": 0.15975767374038696,
      "learning_rate": 6.551334330149114e-06,
      "loss": 0.3307,
      "step": 392
    },
    {
      "epoch": 0.8136645962732919,
      "grad_norm": 0.2354326993227005,
      "learning_rate": 6.535705932222881e-06,
      "loss": 0.3146,
      "step": 393
    },
    {
      "epoch": 0.8157349896480331,
      "grad_norm": 0.2226390838623047,
      "learning_rate": 6.520060950207186e-06,
      "loss": 0.3535,
      "step": 394
    },
    {
      "epoch": 0.8178053830227743,
      "grad_norm": 0.20899449288845062,
      "learning_rate": 6.504399553052199e-06,
      "loss": 0.4009,
      "step": 395
    },
    {
      "epoch": 0.8198757763975155,
      "grad_norm": 0.17171704769134521,
      "learning_rate": 6.488721909885359e-06,
      "loss": 0.3999,
      "step": 396
    },
    {
      "epoch": 0.8219461697722568,
      "grad_norm": 0.1482180655002594,
      "learning_rate": 6.4730281900095474e-06,
      "loss": 0.2778,
      "step": 397
    },
    {
      "epoch": 0.8240165631469979,
      "grad_norm": 0.1984974592924118,
      "learning_rate": 6.457318562901257e-06,
      "loss": 0.3702,
      "step": 398
    },
    {
      "epoch": 0.8260869565217391,
      "grad_norm": 0.1667233407497406,
      "learning_rate": 6.44159319820876e-06,
      "loss": 0.3629,
      "step": 399
    },
    {
      "epoch": 0.8281573498964804,
      "grad_norm": 0.18199460208415985,
      "learning_rate": 6.425852265750282e-06,
      "loss": 0.3925,
      "step": 400
    },
    {
      "epoch": 0.8302277432712215,
      "grad_norm": 0.18740075826644897,
      "learning_rate": 6.4100959355121636e-06,
      "loss": 0.3852,
      "step": 401
    },
    {
      "epoch": 0.8322981366459627,
      "grad_norm": 0.15155306458473206,
      "learning_rate": 6.394324377647028e-06,
      "loss": 0.198,
      "step": 402
    },
    {
      "epoch": 0.8343685300207039,
      "grad_norm": 0.18929250538349152,
      "learning_rate": 6.378537762471937e-06,
      "loss": 0.3827,
      "step": 403
    },
    {
      "epoch": 0.8364389233954451,
      "grad_norm": 0.15943989157676697,
      "learning_rate": 6.362736260466561e-06,
      "loss": 0.3253,
      "step": 404
    },
    {
      "epoch": 0.8385093167701864,
      "grad_norm": 0.2277781367301941,
      "learning_rate": 6.3469200422713316e-06,
      "loss": 0.4614,
      "step": 405
    },
    {
      "epoch": 0.8405797101449275,
      "grad_norm": 0.13026756048202515,
      "learning_rate": 6.331089278685599e-06,
      "loss": 0.2236,
      "step": 406
    },
    {
      "epoch": 0.8426501035196687,
      "grad_norm": 0.2032948136329651,
      "learning_rate": 6.315244140665793e-06,
      "loss": 0.4658,
      "step": 407
    },
    {
      "epoch": 0.84472049689441,
      "grad_norm": 0.16974839568138123,
      "learning_rate": 6.299384799323568e-06,
      "loss": 0.409,
      "step": 408
    },
    {
      "epoch": 0.8467908902691511,
      "grad_norm": 0.14744219183921814,
      "learning_rate": 6.283511425923964e-06,
      "loss": 0.263,
      "step": 409
    },
    {
      "epoch": 0.8488612836438924,
      "grad_norm": 0.24342601001262665,
      "learning_rate": 6.267624191883551e-06,
      "loss": 0.2567,
      "step": 410
    },
    {
      "epoch": 0.8509316770186336,
      "grad_norm": 0.22432643175125122,
      "learning_rate": 6.2517232687685815e-06,
      "loss": 0.3231,
      "step": 411
    },
    {
      "epoch": 0.8530020703933747,
      "grad_norm": 0.25641682744026184,
      "learning_rate": 6.235808828293135e-06,
      "loss": 0.4221,
      "step": 412
    },
    {
      "epoch": 0.855072463768116,
      "grad_norm": 0.22766365110874176,
      "learning_rate": 6.2198810423172655e-06,
      "loss": 0.2003,
      "step": 413
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.15609072148799896,
      "learning_rate": 6.203940082845144e-06,
      "loss": 0.4432,
      "step": 414
    },
    {
      "epoch": 0.8592132505175983,
      "grad_norm": 0.269815593957901,
      "learning_rate": 6.187986122023207e-06,
      "loss": 0.4135,
      "step": 415
    },
    {
      "epoch": 0.8612836438923396,
      "grad_norm": 0.11367934942245483,
      "learning_rate": 6.172019332138285e-06,
      "loss": 0.1468,
      "step": 416
    },
    {
      "epoch": 0.8633540372670807,
      "grad_norm": 0.17702323198318481,
      "learning_rate": 6.1560398856157554e-06,
      "loss": 0.3226,
      "step": 417
    },
    {
      "epoch": 0.865424430641822,
      "grad_norm": 0.18008998036384583,
      "learning_rate": 6.140047955017672e-06,
      "loss": 0.4057,
      "step": 418
    },
    {
      "epoch": 0.8674948240165632,
      "grad_norm": 0.13514308631420135,
      "learning_rate": 6.1240437130409056e-06,
      "loss": 0.2623,
      "step": 419
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": 0.1675521731376648,
      "learning_rate": 6.108027332515276e-06,
      "loss": 0.2896,
      "step": 420
    },
    {
      "epoch": 0.8716356107660456,
      "grad_norm": 0.2000652253627777,
      "learning_rate": 6.091998986401687e-06,
      "loss": 0.3716,
      "step": 421
    },
    {
      "epoch": 0.8737060041407867,
      "grad_norm": 0.13207314908504486,
      "learning_rate": 6.075958847790262e-06,
      "loss": 0.2428,
      "step": 422
    },
    {
      "epoch": 0.8757763975155279,
      "grad_norm": 0.1675049066543579,
      "learning_rate": 6.059907089898468e-06,
      "loss": 0.2582,
      "step": 423
    },
    {
      "epoch": 0.8778467908902692,
      "grad_norm": 0.19665411114692688,
      "learning_rate": 6.043843886069251e-06,
      "loss": 0.3385,
      "step": 424
    },
    {
      "epoch": 0.8799171842650103,
      "grad_norm": 0.2043440341949463,
      "learning_rate": 6.02776940976916e-06,
      "loss": 0.3633,
      "step": 425
    },
    {
      "epoch": 0.8819875776397516,
      "grad_norm": 0.19804300367832184,
      "learning_rate": 6.011683834586474e-06,
      "loss": 0.3014,
      "step": 426
    },
    {
      "epoch": 0.8840579710144928,
      "grad_norm": 0.15968140959739685,
      "learning_rate": 5.995587334229334e-06,
      "loss": 0.2579,
      "step": 427
    },
    {
      "epoch": 0.8861283643892339,
      "grad_norm": 0.213617742061615,
      "learning_rate": 5.979480082523858e-06,
      "loss": 0.2747,
      "step": 428
    },
    {
      "epoch": 0.8881987577639752,
      "grad_norm": 0.1532445102930069,
      "learning_rate": 5.963362253412269e-06,
      "loss": 0.2513,
      "step": 429
    },
    {
      "epoch": 0.8902691511387164,
      "grad_norm": 0.15875503420829773,
      "learning_rate": 5.947234020951015e-06,
      "loss": 0.2581,
      "step": 430
    },
    {
      "epoch": 0.8923395445134575,
      "grad_norm": 0.16298717260360718,
      "learning_rate": 5.931095559308888e-06,
      "loss": 0.3126,
      "step": 431
    },
    {
      "epoch": 0.8944099378881988,
      "grad_norm": 0.18198642134666443,
      "learning_rate": 5.914947042765149e-06,
      "loss": 0.3726,
      "step": 432
    },
    {
      "epoch": 0.8964803312629399,
      "grad_norm": 0.1657399982213974,
      "learning_rate": 5.8987886457076405e-06,
      "loss": 0.2984,
      "step": 433
    },
    {
      "epoch": 0.8985507246376812,
      "grad_norm": 0.13853290677070618,
      "learning_rate": 5.882620542630901e-06,
      "loss": 0.1906,
      "step": 434
    },
    {
      "epoch": 0.9006211180124224,
      "grad_norm": 0.20282739400863647,
      "learning_rate": 5.866442908134291e-06,
      "loss": 0.3678,
      "step": 435
    },
    {
      "epoch": 0.9026915113871635,
      "grad_norm": 0.18411196768283844,
      "learning_rate": 5.850255916920093e-06,
      "loss": 0.3661,
      "step": 436
    },
    {
      "epoch": 0.9047619047619048,
      "grad_norm": 0.16809453070163727,
      "learning_rate": 5.8340597437916395e-06,
      "loss": 0.4337,
      "step": 437
    },
    {
      "epoch": 0.906832298136646,
      "grad_norm": 0.24382607638835907,
      "learning_rate": 5.817854563651415e-06,
      "loss": 0.2962,
      "step": 438
    },
    {
      "epoch": 0.9089026915113871,
      "grad_norm": 0.21871711313724518,
      "learning_rate": 5.80164055149917e-06,
      "loss": 0.413,
      "step": 439
    },
    {
      "epoch": 0.9109730848861284,
      "grad_norm": 0.14632663130760193,
      "learning_rate": 5.785417882430035e-06,
      "loss": 0.3107,
      "step": 440
    },
    {
      "epoch": 0.9130434782608695,
      "grad_norm": 0.19092142581939697,
      "learning_rate": 5.769186731632624e-06,
      "loss": 0.3483,
      "step": 441
    },
    {
      "epoch": 0.9151138716356108,
      "grad_norm": 0.2034963071346283,
      "learning_rate": 5.752947274387147e-06,
      "loss": 0.4898,
      "step": 442
    },
    {
      "epoch": 0.917184265010352,
      "grad_norm": 0.15895184874534607,
      "learning_rate": 5.736699686063515e-06,
      "loss": 0.2704,
      "step": 443
    },
    {
      "epoch": 0.9192546583850931,
      "grad_norm": 0.21739298105239868,
      "learning_rate": 5.720444142119445e-06,
      "loss": 0.2768,
      "step": 444
    },
    {
      "epoch": 0.9213250517598344,
      "grad_norm": 0.17576731741428375,
      "learning_rate": 5.704180818098567e-06,
      "loss": 0.335,
      "step": 445
    },
    {
      "epoch": 0.9233954451345756,
      "grad_norm": 0.15474459528923035,
      "learning_rate": 5.687909889628529e-06,
      "loss": 0.3155,
      "step": 446
    },
    {
      "epoch": 0.9254658385093167,
      "grad_norm": 0.2610938847064972,
      "learning_rate": 5.6716315324191e-06,
      "loss": 0.4243,
      "step": 447
    },
    {
      "epoch": 0.927536231884058,
      "grad_norm": 0.18764007091522217,
      "learning_rate": 5.6553459222602714e-06,
      "loss": 0.3547,
      "step": 448
    },
    {
      "epoch": 0.9296066252587992,
      "grad_norm": 0.19586984813213348,
      "learning_rate": 5.639053235020358e-06,
      "loss": 0.4074,
      "step": 449
    },
    {
      "epoch": 0.9316770186335404,
      "grad_norm": 0.16940924525260925,
      "learning_rate": 5.622753646644102e-06,
      "loss": 0.2609,
      "step": 450
    },
    {
      "epoch": 0.9337474120082816,
      "grad_norm": 0.1849866807460785,
      "learning_rate": 5.606447333150768e-06,
      "loss": 0.3816,
      "step": 451
    },
    {
      "epoch": 0.9358178053830227,
      "grad_norm": 0.23204651474952698,
      "learning_rate": 5.59013447063225e-06,
      "loss": 0.2354,
      "step": 452
    },
    {
      "epoch": 0.937888198757764,
      "grad_norm": 0.22019827365875244,
      "learning_rate": 5.57381523525116e-06,
      "loss": 0.3601,
      "step": 453
    },
    {
      "epoch": 0.9399585921325052,
      "grad_norm": 0.2608383595943451,
      "learning_rate": 5.557489803238934e-06,
      "loss": 0.3874,
      "step": 454
    },
    {
      "epoch": 0.9420289855072463,
      "grad_norm": 0.13737766444683075,
      "learning_rate": 5.541158350893922e-06,
      "loss": 0.1811,
      "step": 455
    },
    {
      "epoch": 0.9440993788819876,
      "grad_norm": 0.2050437182188034,
      "learning_rate": 5.524821054579491e-06,
      "loss": 0.3328,
      "step": 456
    },
    {
      "epoch": 0.9461697722567288,
      "grad_norm": 0.1720466911792755,
      "learning_rate": 5.508478090722116e-06,
      "loss": 0.3377,
      "step": 457
    },
    {
      "epoch": 0.94824016563147,
      "grad_norm": 0.16786806285381317,
      "learning_rate": 5.492129635809473e-06,
      "loss": 0.319,
      "step": 458
    },
    {
      "epoch": 0.9503105590062112,
      "grad_norm": 0.16816474497318268,
      "learning_rate": 5.475775866388542e-06,
      "loss": 0.3468,
      "step": 459
    },
    {
      "epoch": 0.9523809523809523,
      "grad_norm": 0.22582465410232544,
      "learning_rate": 5.459416959063688e-06,
      "loss": 0.3769,
      "step": 460
    },
    {
      "epoch": 0.9544513457556936,
      "grad_norm": 0.24350348114967346,
      "learning_rate": 5.443053090494763e-06,
      "loss": 0.3339,
      "step": 461
    },
    {
      "epoch": 0.9565217391304348,
      "grad_norm": 0.17309525609016418,
      "learning_rate": 5.426684437395196e-06,
      "loss": 0.2794,
      "step": 462
    },
    {
      "epoch": 0.9585921325051759,
      "grad_norm": 0.18608134984970093,
      "learning_rate": 5.4103111765300855e-06,
      "loss": 0.3433,
      "step": 463
    },
    {
      "epoch": 0.9606625258799172,
      "grad_norm": 0.16802716255187988,
      "learning_rate": 5.393933484714284e-06,
      "loss": 0.2962,
      "step": 464
    },
    {
      "epoch": 0.9627329192546584,
      "grad_norm": 0.22532106935977936,
      "learning_rate": 5.377551538810499e-06,
      "loss": 0.2857,
      "step": 465
    },
    {
      "epoch": 0.9648033126293996,
      "grad_norm": 0.23656564950942993,
      "learning_rate": 5.361165515727374e-06,
      "loss": 0.2993,
      "step": 466
    },
    {
      "epoch": 0.9668737060041408,
      "grad_norm": 0.1953495591878891,
      "learning_rate": 5.3447755924175885e-06,
      "loss": 0.3035,
      "step": 467
    },
    {
      "epoch": 0.968944099378882,
      "grad_norm": 0.21909025311470032,
      "learning_rate": 5.328381945875933e-06,
      "loss": 0.347,
      "step": 468
    },
    {
      "epoch": 0.9710144927536232,
      "grad_norm": 0.24978509545326233,
      "learning_rate": 5.311984753137407e-06,
      "loss": 0.2855,
      "step": 469
    },
    {
      "epoch": 0.9730848861283644,
      "grad_norm": 0.14834006130695343,
      "learning_rate": 5.295584191275308e-06,
      "loss": 0.1876,
      "step": 470
    },
    {
      "epoch": 0.9751552795031055,
      "grad_norm": 0.16186289489269257,
      "learning_rate": 5.279180437399316e-06,
      "loss": 0.225,
      "step": 471
    },
    {
      "epoch": 0.9772256728778468,
      "grad_norm": 0.206668421626091,
      "learning_rate": 5.26277366865358e-06,
      "loss": 0.2826,
      "step": 472
    },
    {
      "epoch": 0.979296066252588,
      "grad_norm": 0.18108750879764557,
      "learning_rate": 5.246364062214809e-06,
      "loss": 0.2851,
      "step": 473
    },
    {
      "epoch": 0.9813664596273292,
      "grad_norm": 0.1724204570055008,
      "learning_rate": 5.229951795290353e-06,
      "loss": 0.3175,
      "step": 474
    },
    {
      "epoch": 0.9834368530020704,
      "grad_norm": 0.22895722091197968,
      "learning_rate": 5.213537045116296e-06,
      "loss": 0.5164,
      "step": 475
    },
    {
      "epoch": 0.9855072463768116,
      "grad_norm": 0.204612135887146,
      "learning_rate": 5.197119988955534e-06,
      "loss": 0.491,
      "step": 476
    },
    {
      "epoch": 0.9875776397515528,
      "grad_norm": 0.26122862100601196,
      "learning_rate": 5.180700804095871e-06,
      "loss": 0.2628,
      "step": 477
    },
    {
      "epoch": 0.989648033126294,
      "grad_norm": 0.2136368602514267,
      "learning_rate": 5.164279667848094e-06,
      "loss": 0.3404,
      "step": 478
    },
    {
      "epoch": 0.9917184265010351,
      "grad_norm": 0.18253998458385468,
      "learning_rate": 5.147856757544067e-06,
      "loss": 0.2777,
      "step": 479
    },
    {
      "epoch": 0.9937888198757764,
      "grad_norm": 0.19649861752986908,
      "learning_rate": 5.131432250534809e-06,
      "loss": 0.3999,
      "step": 480
    },
    {
      "epoch": 0.9958592132505176,
      "grad_norm": 0.18403136730194092,
      "learning_rate": 5.1150063241885815e-06,
      "loss": 0.3778,
      "step": 481
    },
    {
      "epoch": 0.9979296066252588,
      "grad_norm": 0.19857674837112427,
      "learning_rate": 5.0985791558889785e-06,
      "loss": 0.2799,
      "step": 482
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.13328629732131958,
      "learning_rate": 5.082150923033001e-06,
      "loss": 0.1996,
      "step": 483
    },
    {
      "epoch": 1.0020703933747412,
      "grad_norm": 0.1862826645374298,
      "learning_rate": 5.065721803029146e-06,
      "loss": 0.411,
      "step": 484
    },
    {
      "epoch": 1.0020703933747412,
      "eval_loss": 0.3763667643070221,
      "eval_runtime": 93.2875,
      "eval_samples_per_second": 5.917,
      "eval_steps_per_second": 0.74,
      "step": 484
    },
    {
      "epoch": 1.0041407867494825,
      "grad_norm": 0.23797769844532013,
      "learning_rate": 5.049291973295492e-06,
      "loss": 0.4413,
      "step": 485
    },
    {
      "epoch": 1.0062111801242235,
      "grad_norm": 0.1646006554365158,
      "learning_rate": 5.032861611257783e-06,
      "loss": 0.2354,
      "step": 486
    },
    {
      "epoch": 1.0082815734989647,
      "grad_norm": 0.1261710673570633,
      "learning_rate": 5.016430894347513e-06,
      "loss": 0.2208,
      "step": 487
    },
    {
      "epoch": 1.010351966873706,
      "grad_norm": 0.16026055812835693,
      "learning_rate": 5e-06,
      "loss": 0.2227,
      "step": 488
    },
    {
      "epoch": 1.0124223602484472,
      "grad_norm": 0.1686556041240692,
      "learning_rate": 4.983569105652489e-06,
      "loss": 0.3089,
      "step": 489
    },
    {
      "epoch": 1.0144927536231885,
      "grad_norm": 0.18485727906227112,
      "learning_rate": 4.967138388742218e-06,
      "loss": 0.3351,
      "step": 490
    },
    {
      "epoch": 1.0165631469979297,
      "grad_norm": 0.24687248468399048,
      "learning_rate": 4.9507080267045095e-06,
      "loss": 0.3287,
      "step": 491
    },
    {
      "epoch": 1.0186335403726707,
      "grad_norm": 0.1875036209821701,
      "learning_rate": 4.934278196970857e-06,
      "loss": 0.3356,
      "step": 492
    },
    {
      "epoch": 1.020703933747412,
      "grad_norm": 0.2408786565065384,
      "learning_rate": 4.917849076966999e-06,
      "loss": 0.3066,
      "step": 493
    },
    {
      "epoch": 1.0227743271221532,
      "grad_norm": 0.17887943983078003,
      "learning_rate": 4.9014208441110215e-06,
      "loss": 0.2968,
      "step": 494
    },
    {
      "epoch": 1.0248447204968945,
      "grad_norm": 0.1789269596338272,
      "learning_rate": 4.8849936758114184e-06,
      "loss": 0.3354,
      "step": 495
    },
    {
      "epoch": 1.0269151138716357,
      "grad_norm": 0.14091640710830688,
      "learning_rate": 4.868567749465192e-06,
      "loss": 0.2079,
      "step": 496
    },
    {
      "epoch": 1.0289855072463767,
      "grad_norm": 0.27091121673583984,
      "learning_rate": 4.852143242455935e-06,
      "loss": 0.3013,
      "step": 497
    },
    {
      "epoch": 1.0020703933747412,
      "grad_norm": 0.23239950835704803,
      "learning_rate": 4.835720332151907e-06,
      "loss": 0.4112,
      "step": 498
    },
    {
      "epoch": 1.0041407867494825,
      "grad_norm": 0.2228323519229889,
      "learning_rate": 4.819299195904131e-06,
      "loss": 0.3032,
      "step": 499
    },
    {
      "epoch": 1.0062111801242235,
      "grad_norm": 0.2078544944524765,
      "learning_rate": 4.802880011044467e-06,
      "loss": 0.2802,
      "step": 500
    },
    {
      "epoch": 1.0082815734989647,
      "grad_norm": 0.20137901604175568,
      "learning_rate": 4.786462954883706e-06,
      "loss": 0.3043,
      "step": 501
    },
    {
      "epoch": 1.010351966873706,
      "grad_norm": 0.22740206122398376,
      "learning_rate": 4.770048204709648e-06,
      "loss": 0.3051,
      "step": 502
    },
    {
      "epoch": 1.0124223602484472,
      "grad_norm": 0.18020185828208923,
      "learning_rate": 4.753635937785191e-06,
      "loss": 0.2257,
      "step": 503
    },
    {
      "epoch": 1.0144927536231885,
      "grad_norm": 0.18203116953372955,
      "learning_rate": 4.73722633134642e-06,
      "loss": 0.2019,
      "step": 504
    },
    {
      "epoch": 1.0165631469979297,
      "grad_norm": 0.22147925198078156,
      "learning_rate": 4.720819562600684e-06,
      "loss": 0.2875,
      "step": 505
    },
    {
      "epoch": 1.0186335403726707,
      "grad_norm": 0.28298357129096985,
      "learning_rate": 4.7044158087246926e-06,
      "loss": 0.3057,
      "step": 506
    },
    {
      "epoch": 1.020703933747412,
      "grad_norm": 0.25004157423973083,
      "learning_rate": 4.688015246862595e-06,
      "loss": 0.2947,
      "step": 507
    },
    {
      "epoch": 1.0227743271221532,
      "grad_norm": 0.29407742619514465,
      "learning_rate": 4.67161805412407e-06,
      "loss": 0.475,
      "step": 508
    },
    {
      "epoch": 1.0248447204968945,
      "grad_norm": 0.2174006849527359,
      "learning_rate": 4.655224407582413e-06,
      "loss": 0.3527,
      "step": 509
    },
    {
      "epoch": 1.0269151138716357,
      "grad_norm": 0.17737045884132385,
      "learning_rate": 4.6388344842726266e-06,
      "loss": 0.2289,
      "step": 510
    },
    {
      "epoch": 1.0289855072463767,
      "grad_norm": 0.2775312662124634,
      "learning_rate": 4.622448461189504e-06,
      "loss": 0.2495,
      "step": 511
    },
    {
      "epoch": 1.031055900621118,
      "grad_norm": 0.2338017374277115,
      "learning_rate": 4.606066515285719e-06,
      "loss": 0.347,
      "step": 512
    },
    {
      "epoch": 1.0331262939958592,
      "grad_norm": 0.18186333775520325,
      "learning_rate": 4.589688823469915e-06,
      "loss": 0.4323,
      "step": 513
    },
    {
      "epoch": 1.0351966873706004,
      "grad_norm": 0.15205258131027222,
      "learning_rate": 4.573315562604804e-06,
      "loss": 0.2887,
      "step": 514
    },
    {
      "epoch": 1.0372670807453417,
      "grad_norm": 0.19091299176216125,
      "learning_rate": 4.5569469095052375e-06,
      "loss": 0.3767,
      "step": 515
    },
    {
      "epoch": 1.039337474120083,
      "grad_norm": 0.1971992403268814,
      "learning_rate": 4.540583040936313e-06,
      "loss": 0.3108,
      "step": 516
    },
    {
      "epoch": 1.041407867494824,
      "grad_norm": 0.19016583263874054,
      "learning_rate": 4.5242241336114595e-06,
      "loss": 0.3295,
      "step": 517
    },
    {
      "epoch": 1.0434782608695652,
      "grad_norm": 0.19993621110916138,
      "learning_rate": 4.5078703641905275e-06,
      "loss": 0.2432,
      "step": 518
    },
    {
      "epoch": 1.0455486542443064,
      "grad_norm": 0.160321444272995,
      "learning_rate": 4.491521909277886e-06,
      "loss": 0.2274,
      "step": 519
    },
    {
      "epoch": 1.0476190476190477,
      "grad_norm": 0.13464973866939545,
      "learning_rate": 4.4751789454205105e-06,
      "loss": 0.1574,
      "step": 520
    },
    {
      "epoch": 1.049689440993789,
      "grad_norm": 0.2205759882926941,
      "learning_rate": 4.458841649106079e-06,
      "loss": 0.1798,
      "step": 521
    },
    {
      "epoch": 1.05175983436853,
      "grad_norm": 0.1981126219034195,
      "learning_rate": 4.442510196761068e-06,
      "loss": 0.2323,
      "step": 522
    },
    {
      "epoch": 1.0538302277432712,
      "grad_norm": 0.14194904267787933,
      "learning_rate": 4.42618476474884e-06,
      "loss": 0.2363,
      "step": 523
    },
    {
      "epoch": 1.0559006211180124,
      "grad_norm": 0.2177942842245102,
      "learning_rate": 4.409865529367751e-06,
      "loss": 0.2997,
      "step": 524
    },
    {
      "epoch": 1.0579710144927537,
      "grad_norm": 0.18768401443958282,
      "learning_rate": 4.3935526668492325e-06,
      "loss": 0.2711,
      "step": 525
    },
    {
      "epoch": 1.060041407867495,
      "grad_norm": 0.23028354346752167,
      "learning_rate": 4.377246353355899e-06,
      "loss": 0.3507,
      "step": 526
    },
    {
      "epoch": 1.062111801242236,
      "grad_norm": 0.18663090467453003,
      "learning_rate": 4.3609467649796434e-06,
      "loss": 0.3046,
      "step": 527
    },
    {
      "epoch": 1.0641821946169772,
      "grad_norm": 0.18047387897968292,
      "learning_rate": 4.34465407773973e-06,
      "loss": 0.2462,
      "step": 528
    },
    {
      "epoch": 1.0662525879917184,
      "grad_norm": 0.17824740707874298,
      "learning_rate": 4.328368467580901e-06,
      "loss": 0.3397,
      "step": 529
    },
    {
      "epoch": 1.0683229813664596,
      "grad_norm": 0.1294453889131546,
      "learning_rate": 4.312090110371473e-06,
      "loss": 0.1294,
      "step": 530
    },
    {
      "epoch": 1.0703933747412009,
      "grad_norm": 0.1727997064590454,
      "learning_rate": 4.295819181901436e-06,
      "loss": 0.2847,
      "step": 531
    },
    {
      "epoch": 1.0724637681159421,
      "grad_norm": 0.20239372551441193,
      "learning_rate": 4.279555857880558e-06,
      "loss": 0.2757,
      "step": 532
    },
    {
      "epoch": 1.0745341614906831,
      "grad_norm": 0.23195482790470123,
      "learning_rate": 4.263300313936485e-06,
      "loss": 0.2639,
      "step": 533
    },
    {
      "epoch": 1.0766045548654244,
      "grad_norm": 0.1517760008573532,
      "learning_rate": 4.247052725612853e-06,
      "loss": 0.2448,
      "step": 534
    },
    {
      "epoch": 1.0786749482401656,
      "grad_norm": 0.2006104290485382,
      "learning_rate": 4.2308132683673765e-06,
      "loss": 0.3653,
      "step": 535
    },
    {
      "epoch": 1.0807453416149069,
      "grad_norm": 0.20593738555908203,
      "learning_rate": 4.214582117569966e-06,
      "loss": 0.3572,
      "step": 536
    },
    {
      "epoch": 1.0828157349896481,
      "grad_norm": 0.17792487144470215,
      "learning_rate": 4.198359448500832e-06,
      "loss": 0.3457,
      "step": 537
    },
    {
      "epoch": 1.0848861283643891,
      "grad_norm": 0.1635383516550064,
      "learning_rate": 4.182145436348587e-06,
      "loss": 0.3257,
      "step": 538
    },
    {
      "epoch": 1.0869565217391304,
      "grad_norm": 0.2741135060787201,
      "learning_rate": 4.165940256208361e-06,
      "loss": 0.3041,
      "step": 539
    },
    {
      "epoch": 1.0890269151138716,
      "grad_norm": 0.2326912134885788,
      "learning_rate": 4.1497440830799084e-06,
      "loss": 0.3032,
      "step": 540
    },
    {
      "epoch": 1.0910973084886129,
      "grad_norm": 0.18751777708530426,
      "learning_rate": 4.133557091865711e-06,
      "loss": 0.2383,
      "step": 541
    },
    {
      "epoch": 1.093167701863354,
      "grad_norm": 0.16328708827495575,
      "learning_rate": 4.1173794573691e-06,
      "loss": 0.2221,
      "step": 542
    },
    {
      "epoch": 1.0952380952380953,
      "grad_norm": 0.20164957642555237,
      "learning_rate": 4.1012113542923595e-06,
      "loss": 0.3223,
      "step": 543
    },
    {
      "epoch": 1.0973084886128364,
      "grad_norm": 0.1946948915719986,
      "learning_rate": 4.0850529572348505e-06,
      "loss": 0.3173,
      "step": 544
    },
    {
      "epoch": 1.0993788819875776,
      "grad_norm": 0.1779978722333908,
      "learning_rate": 4.068904440691113e-06,
      "loss": 0.3294,
      "step": 545
    },
    {
      "epoch": 1.1014492753623188,
      "grad_norm": 0.20775990188121796,
      "learning_rate": 4.052765979048986e-06,
      "loss": 0.2444,
      "step": 546
    },
    {
      "epoch": 1.10351966873706,
      "grad_norm": 0.1465255469083786,
      "learning_rate": 4.036637746587732e-06,
      "loss": 0.1946,
      "step": 547
    },
    {
      "epoch": 1.1055900621118013,
      "grad_norm": 0.1578349471092224,
      "learning_rate": 4.0205199174761435e-06,
      "loss": 0.3166,
      "step": 548
    },
    {
      "epoch": 1.1076604554865424,
      "grad_norm": 0.16846297681331635,
      "learning_rate": 4.004412665770667e-06,
      "loss": 0.2747,
      "step": 549
    },
    {
      "epoch": 1.1097308488612836,
      "grad_norm": 0.17560021579265594,
      "learning_rate": 3.988316165413528e-06,
      "loss": 0.2477,
      "step": 550
    },
    {
      "epoch": 1.1118012422360248,
      "grad_norm": 0.3064589202404022,
      "learning_rate": 3.972230590230844e-06,
      "loss": 0.384,
      "step": 551
    },
    {
      "epoch": 1.113871635610766,
      "grad_norm": 0.28778812289237976,
      "learning_rate": 3.956156113930752e-06,
      "loss": 0.2753,
      "step": 552
    },
    {
      "epoch": 1.1159420289855073,
      "grad_norm": 0.18141014873981476,
      "learning_rate": 3.940092910101533e-06,
      "loss": 0.3229,
      "step": 553
    },
    {
      "epoch": 1.1180124223602483,
      "grad_norm": 0.2242807000875473,
      "learning_rate": 3.924041152209739e-06,
      "loss": 0.3954,
      "step": 554
    },
    {
      "epoch": 1.1200828157349896,
      "grad_norm": 0.17454344034194946,
      "learning_rate": 3.9080010135983134e-06,
      "loss": 0.2391,
      "step": 555
    },
    {
      "epoch": 1.1221532091097308,
      "grad_norm": 0.22734415531158447,
      "learning_rate": 3.891972667484726e-06,
      "loss": 0.3418,
      "step": 556
    },
    {
      "epoch": 1.124223602484472,
      "grad_norm": 0.16846442222595215,
      "learning_rate": 3.875956286959096e-06,
      "loss": 0.3489,
      "step": 557
    },
    {
      "epoch": 1.1262939958592133,
      "grad_norm": 0.1894855499267578,
      "learning_rate": 3.859952044982329e-06,
      "loss": 0.312,
      "step": 558
    },
    {
      "epoch": 1.1283643892339545,
      "grad_norm": 0.17569231986999512,
      "learning_rate": 3.843960114384246e-06,
      "loss": 0.2616,
      "step": 559
    },
    {
      "epoch": 1.1304347826086956,
      "grad_norm": 0.26564398407936096,
      "learning_rate": 3.827980667861716e-06,
      "loss": 0.3152,
      "step": 560
    },
    {
      "epoch": 1.1325051759834368,
      "grad_norm": 0.19927287101745605,
      "learning_rate": 3.8120138779767958e-06,
      "loss": 0.2786,
      "step": 561
    },
    {
      "epoch": 1.134575569358178,
      "grad_norm": 0.2221873551607132,
      "learning_rate": 3.7960599171548572e-06,
      "loss": 0.3006,
      "step": 562
    },
    {
      "epoch": 1.1366459627329193,
      "grad_norm": 0.22802579402923584,
      "learning_rate": 3.780118957682736e-06,
      "loss": 0.4023,
      "step": 563
    },
    {
      "epoch": 1.1387163561076605,
      "grad_norm": 0.21268118917942047,
      "learning_rate": 3.764191171706867e-06,
      "loss": 0.3199,
      "step": 564
    },
    {
      "epoch": 1.1407867494824018,
      "grad_norm": 0.18768849968910217,
      "learning_rate": 3.7482767312314206e-06,
      "loss": 0.2721,
      "step": 565
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 0.19857840240001678,
      "learning_rate": 3.732375808116451e-06,
      "loss": 0.3533,
      "step": 566
    },
    {
      "epoch": 1.144927536231884,
      "grad_norm": 0.23903526365756989,
      "learning_rate": 3.7164885740760375e-06,
      "loss": 0.3196,
      "step": 567
    },
    {
      "epoch": 1.1469979296066253,
      "grad_norm": 0.21161702275276184,
      "learning_rate": 3.7006152006764336e-06,
      "loss": 0.3384,
      "step": 568
    },
    {
      "epoch": 1.1490683229813665,
      "grad_norm": 0.180229052901268,
      "learning_rate": 3.684755859334209e-06,
      "loss": 0.3353,
      "step": 569
    },
    {
      "epoch": 1.1511387163561078,
      "grad_norm": 0.19796229898929596,
      "learning_rate": 3.6689107213144025e-06,
      "loss": 0.3109,
      "step": 570
    },
    {
      "epoch": 1.1532091097308488,
      "grad_norm": 0.209706112742424,
      "learning_rate": 3.653079957728671e-06,
      "loss": 0.2478,
      "step": 571
    },
    {
      "epoch": 1.15527950310559,
      "grad_norm": 0.24063701927661896,
      "learning_rate": 3.6372637395334416e-06,
      "loss": 0.4906,
      "step": 572
    },
    {
      "epoch": 1.1573498964803313,
      "grad_norm": 0.17047333717346191,
      "learning_rate": 3.6214622375280637e-06,
      "loss": 0.2871,
      "step": 573
    },
    {
      "epoch": 1.1594202898550725,
      "grad_norm": 0.1966450810432434,
      "learning_rate": 3.6056756223529734e-06,
      "loss": 0.3539,
      "step": 574
    },
    {
      "epoch": 1.1614906832298137,
      "grad_norm": 0.33538636565208435,
      "learning_rate": 3.589904064487837e-06,
      "loss": 0.3929,
      "step": 575
    },
    {
      "epoch": 1.1635610766045548,
      "grad_norm": 0.17587830126285553,
      "learning_rate": 3.574147734249719e-06,
      "loss": 0.2755,
      "step": 576
    },
    {
      "epoch": 1.165631469979296,
      "grad_norm": 0.1887330859899521,
      "learning_rate": 3.5584068017912415e-06,
      "loss": 0.2438,
      "step": 577
    },
    {
      "epoch": 1.1677018633540373,
      "grad_norm": 0.17411023378372192,
      "learning_rate": 3.542681437098745e-06,
      "loss": 0.2861,
      "step": 578
    },
    {
      "epoch": 1.1697722567287785,
      "grad_norm": 0.18099556863307953,
      "learning_rate": 3.526971809990454e-06,
      "loss": 0.2884,
      "step": 579
    },
    {
      "epoch": 1.1718426501035197,
      "grad_norm": 0.1913526952266693,
      "learning_rate": 3.5112780901146426e-06,
      "loss": 0.2879,
      "step": 580
    },
    {
      "epoch": 1.1739130434782608,
      "grad_norm": 0.19618435204029083,
      "learning_rate": 3.4956004469478038e-06,
      "loss": 0.2704,
      "step": 581
    },
    {
      "epoch": 1.175983436853002,
      "grad_norm": 0.24511279165744781,
      "learning_rate": 3.479939049792817e-06,
      "loss": 0.3838,
      "step": 582
    },
    {
      "epoch": 1.1780538302277432,
      "grad_norm": 0.2790193259716034,
      "learning_rate": 3.4642940677771203e-06,
      "loss": 0.3175,
      "step": 583
    },
    {
      "epoch": 1.1801242236024845,
      "grad_norm": 0.19622787833213806,
      "learning_rate": 3.448665669850888e-06,
      "loss": 0.2721,
      "step": 584
    },
    {
      "epoch": 1.1821946169772257,
      "grad_norm": 0.2602899968624115,
      "learning_rate": 3.433054024785199e-06,
      "loss": 0.4717,
      "step": 585
    },
    {
      "epoch": 1.184265010351967,
      "grad_norm": 0.2367587685585022,
      "learning_rate": 3.4174593011702197e-06,
      "loss": 0.2121,
      "step": 586
    },
    {
      "epoch": 1.186335403726708,
      "grad_norm": 0.17743825912475586,
      "learning_rate": 3.4018816674133814e-06,
      "loss": 0.3188,
      "step": 587
    },
    {
      "epoch": 1.1884057971014492,
      "grad_norm": 0.18005116283893585,
      "learning_rate": 3.386321291737563e-06,
      "loss": 0.222,
      "step": 588
    },
    {
      "epoch": 1.1904761904761905,
      "grad_norm": 0.17598001658916473,
      "learning_rate": 3.3707783421792724e-06,
      "loss": 0.3,
      "step": 589
    },
    {
      "epoch": 1.1925465838509317,
      "grad_norm": 0.24181915819644928,
      "learning_rate": 3.3552529865868323e-06,
      "loss": 0.2492,
      "step": 590
    },
    {
      "epoch": 1.194616977225673,
      "grad_norm": 0.16726148128509521,
      "learning_rate": 3.339745392618569e-06,
      "loss": 0.2398,
      "step": 591
    },
    {
      "epoch": 1.1966873706004142,
      "grad_norm": 0.20498837530612946,
      "learning_rate": 3.3242557277410015e-06,
      "loss": 0.427,
      "step": 592
    },
    {
      "epoch": 1.1987577639751552,
      "grad_norm": 0.14857308566570282,
      "learning_rate": 3.3087841592270296e-06,
      "loss": 0.2312,
      "step": 593
    },
    {
      "epoch": 1.2008281573498965,
      "grad_norm": 0.20990735292434692,
      "learning_rate": 3.2933308541541365e-06,
      "loss": 0.3369,
      "step": 594
    },
    {
      "epoch": 1.2028985507246377,
      "grad_norm": 0.21872520446777344,
      "learning_rate": 3.2778959794025735e-06,
      "loss": 0.3638,
      "step": 595
    },
    {
      "epoch": 1.204968944099379,
      "grad_norm": 0.20835131406784058,
      "learning_rate": 3.2624797016535626e-06,
      "loss": 0.3235,
      "step": 596
    },
    {
      "epoch": 1.2070393374741202,
      "grad_norm": 0.3120534420013428,
      "learning_rate": 3.2470821873875003e-06,
      "loss": 0.4174,
      "step": 597
    },
    {
      "epoch": 1.2091097308488612,
      "grad_norm": 0.176408052444458,
      "learning_rate": 3.2317036028821523e-06,
      "loss": 0.2244,
      "step": 598
    },
    {
      "epoch": 1.2111801242236024,
      "grad_norm": 0.47249072790145874,
      "learning_rate": 3.216344114210865e-06,
      "loss": 0.459,
      "step": 599
    },
    {
      "epoch": 1.2132505175983437,
      "grad_norm": 0.21692520380020142,
      "learning_rate": 3.201003887240768e-06,
      "loss": 0.329,
      "step": 600
    },
    {
      "epoch": 1.215320910973085,
      "grad_norm": 0.2014329880475998,
      "learning_rate": 3.185683087630982e-06,
      "loss": 0.3236,
      "step": 601
    },
    {
      "epoch": 1.2173913043478262,
      "grad_norm": 0.17368103563785553,
      "learning_rate": 3.1703818808308327e-06,
      "loss": 0.2636,
      "step": 602
    },
    {
      "epoch": 1.2194616977225672,
      "grad_norm": 0.25841301679611206,
      "learning_rate": 3.1551004320780634e-06,
      "loss": 0.3909,
      "step": 603
    },
    {
      "epoch": 1.2215320910973084,
      "grad_norm": 0.15552645921707153,
      "learning_rate": 3.1398389063970512e-06,
      "loss": 0.201,
      "step": 604
    },
    {
      "epoch": 1.2236024844720497,
      "grad_norm": 0.2024988979101181,
      "learning_rate": 3.124597468597024e-06,
      "loss": 0.3928,
      "step": 605
    },
    {
      "epoch": 1.2236024844720497,
      "eval_loss": 0.3698825240135193,
      "eval_runtime": 93.2967,
      "eval_samples_per_second": 5.917,
      "eval_steps_per_second": 0.74,
      "step": 605
    },
    {
      "epoch": 1.225672877846791,
      "grad_norm": 0.25626295804977417,
      "learning_rate": 3.1093762832702775e-06,
      "loss": 0.3436,
      "step": 606
    },
    {
      "epoch": 1.2277432712215322,
      "grad_norm": 0.21667784452438354,
      "learning_rate": 3.0941755147904027e-06,
      "loss": 0.2657,
      "step": 607
    },
    {
      "epoch": 1.2298136645962732,
      "grad_norm": 0.19471991062164307,
      "learning_rate": 3.0789953273105123e-06,
      "loss": 0.2913,
      "step": 608
    },
    {
      "epoch": 1.2318840579710144,
      "grad_norm": 0.2012849897146225,
      "learning_rate": 3.063835884761458e-06,
      "loss": 0.3049,
      "step": 609
    },
    {
      "epoch": 1.2339544513457557,
      "grad_norm": 0.22813789546489716,
      "learning_rate": 3.048697350850073e-06,
      "loss": 0.3013,
      "step": 610
    },
    {
      "epoch": 1.236024844720497,
      "grad_norm": 0.15448856353759766,
      "learning_rate": 3.0335798890573944e-06,
      "loss": 0.331,
      "step": 611
    },
    {
      "epoch": 1.2380952380952381,
      "grad_norm": 0.21554410457611084,
      "learning_rate": 3.0184836626369034e-06,
      "loss": 0.4201,
      "step": 612
    },
    {
      "epoch": 1.2401656314699794,
      "grad_norm": 0.17196184396743774,
      "learning_rate": 3.003408834612759e-06,
      "loss": 0.2679,
      "step": 613
    },
    {
      "epoch": 1.2422360248447206,
      "grad_norm": 0.1990622580051422,
      "learning_rate": 2.988355567778043e-06,
      "loss": 0.2306,
      "step": 614
    },
    {
      "epoch": 1.2443064182194616,
      "grad_norm": 0.181349977850914,
      "learning_rate": 2.9733240246929927e-06,
      "loss": 0.2736,
      "step": 615
    },
    {
      "epoch": 1.2463768115942029,
      "grad_norm": 0.17092598974704742,
      "learning_rate": 2.9583143676832526e-06,
      "loss": 0.3401,
      "step": 616
    },
    {
      "epoch": 1.2484472049689441,
      "grad_norm": 0.16415558755397797,
      "learning_rate": 2.9433267588381198e-06,
      "loss": 0.2277,
      "step": 617
    },
    {
      "epoch": 1.2505175983436854,
      "grad_norm": 0.16953089833259583,
      "learning_rate": 2.9283613600087933e-06,
      "loss": 0.2741,
      "step": 618
    },
    {
      "epoch": 1.2525879917184266,
      "grad_norm": 0.1386069357395172,
      "learning_rate": 2.913418332806624e-06,
      "loss": 0.1662,
      "step": 619
    },
    {
      "epoch": 1.2546583850931676,
      "grad_norm": 0.17235419154167175,
      "learning_rate": 2.8984978386013767e-06,
      "loss": 0.2238,
      "step": 620
    },
    {
      "epoch": 1.2567287784679089,
      "grad_norm": 0.13924579322338104,
      "learning_rate": 2.8836000385194753e-06,
      "loss": 0.1699,
      "step": 621
    },
    {
      "epoch": 1.25879917184265,
      "grad_norm": 0.17973099648952484,
      "learning_rate": 2.8687250934422774e-06,
      "loss": 0.2394,
      "step": 622
    },
    {
      "epoch": 1.2608695652173914,
      "grad_norm": 0.23071612417697906,
      "learning_rate": 2.853873164004321e-06,
      "loss": 0.2409,
      "step": 623
    },
    {
      "epoch": 1.2629399585921326,
      "grad_norm": 0.17298616468906403,
      "learning_rate": 2.839044410591606e-06,
      "loss": 0.3126,
      "step": 624
    },
    {
      "epoch": 1.2650103519668736,
      "grad_norm": 0.18979336321353912,
      "learning_rate": 2.824238993339852e-06,
      "loss": 0.1662,
      "step": 625
    },
    {
      "epoch": 1.2670807453416149,
      "grad_norm": 0.14998288452625275,
      "learning_rate": 2.809457072132766e-06,
      "loss": 0.2216,
      "step": 626
    },
    {
      "epoch": 1.269151138716356,
      "grad_norm": 0.19262179732322693,
      "learning_rate": 2.794698806600331e-06,
      "loss": 0.2759,
      "step": 627
    },
    {
      "epoch": 1.2712215320910973,
      "grad_norm": 0.2046218365430832,
      "learning_rate": 2.779964356117063e-06,
      "loss": 0.4046,
      "step": 628
    },
    {
      "epoch": 1.2732919254658386,
      "grad_norm": 0.1875535100698471,
      "learning_rate": 2.765253879800307e-06,
      "loss": 0.2233,
      "step": 629
    },
    {
      "epoch": 1.2753623188405796,
      "grad_norm": 0.19307266175746918,
      "learning_rate": 2.750567536508504e-06,
      "loss": 0.2949,
      "step": 630
    },
    {
      "epoch": 1.2774327122153208,
      "grad_norm": 0.21273551881313324,
      "learning_rate": 2.735905484839488e-06,
      "loss": 0.3943,
      "step": 631
    },
    {
      "epoch": 1.279503105590062,
      "grad_norm": 0.19536952674388885,
      "learning_rate": 2.7212678831287627e-06,
      "loss": 0.3996,
      "step": 632
    },
    {
      "epoch": 1.2815734989648033,
      "grad_norm": 0.2063191533088684,
      "learning_rate": 2.7066548894478013e-06,
      "loss": 0.3424,
      "step": 633
    },
    {
      "epoch": 1.2836438923395446,
      "grad_norm": 0.2701599597930908,
      "learning_rate": 2.692066661602333e-06,
      "loss": 0.4112,
      "step": 634
    },
    {
      "epoch": 1.2857142857142856,
      "grad_norm": 0.1624588668346405,
      "learning_rate": 2.6775033571306425e-06,
      "loss": 0.2414,
      "step": 635
    },
    {
      "epoch": 1.287784679089027,
      "grad_norm": 0.18011558055877686,
      "learning_rate": 2.662965133301862e-06,
      "loss": 0.2735,
      "step": 636
    },
    {
      "epoch": 1.289855072463768,
      "grad_norm": 0.28961238265037537,
      "learning_rate": 2.6484521471142845e-06,
      "loss": 0.3408,
      "step": 637
    },
    {
      "epoch": 1.2919254658385093,
      "grad_norm": 0.1710355579853058,
      "learning_rate": 2.633964555293654e-06,
      "loss": 0.2626,
      "step": 638
    },
    {
      "epoch": 1.2939958592132506,
      "grad_norm": 0.1638377606868744,
      "learning_rate": 2.619502514291489e-06,
      "loss": 0.2677,
      "step": 639
    },
    {
      "epoch": 1.2960662525879918,
      "grad_norm": 0.2379457652568817,
      "learning_rate": 2.605066180283378e-06,
      "loss": 0.2812,
      "step": 640
    },
    {
      "epoch": 1.298136645962733,
      "grad_norm": 0.2495289295911789,
      "learning_rate": 2.590655709167304e-06,
      "loss": 0.3053,
      "step": 641
    },
    {
      "epoch": 1.300207039337474,
      "grad_norm": 0.18783414363861084,
      "learning_rate": 2.576271256561953e-06,
      "loss": 0.3656,
      "step": 642
    },
    {
      "epoch": 1.3022774327122153,
      "grad_norm": 0.1957497000694275,
      "learning_rate": 2.5619129778050395e-06,
      "loss": 0.3956,
      "step": 643
    },
    {
      "epoch": 1.3043478260869565,
      "grad_norm": 0.19549976289272308,
      "learning_rate": 2.5475810279516287e-06,
      "loss": 0.3395,
      "step": 644
    },
    {
      "epoch": 1.3064182194616978,
      "grad_norm": 0.17275187373161316,
      "learning_rate": 2.5332755617724535e-06,
      "loss": 0.2132,
      "step": 645
    },
    {
      "epoch": 1.308488612836439,
      "grad_norm": 0.218386709690094,
      "learning_rate": 2.5189967337522574e-06,
      "loss": 0.3096,
      "step": 646
    },
    {
      "epoch": 1.31055900621118,
      "grad_norm": 0.2581380009651184,
      "learning_rate": 2.5047446980881106e-06,
      "loss": 0.4621,
      "step": 647
    },
    {
      "epoch": 1.3126293995859213,
      "grad_norm": 0.1893983781337738,
      "learning_rate": 2.49051960868776e-06,
      "loss": 0.2872,
      "step": 648
    },
    {
      "epoch": 1.3146997929606625,
      "grad_norm": 0.22159774601459503,
      "learning_rate": 2.476321619167952e-06,
      "loss": 0.4122,
      "step": 649
    },
    {
      "epoch": 1.3167701863354038,
      "grad_norm": 0.1909107267856598,
      "learning_rate": 2.46215088285279e-06,
      "loss": 0.2927,
      "step": 650
    },
    {
      "epoch": 1.318840579710145,
      "grad_norm": 0.23836734890937805,
      "learning_rate": 2.448007552772062e-06,
      "loss": 0.2825,
      "step": 651
    },
    {
      "epoch": 1.320910973084886,
      "grad_norm": 0.19211657345294952,
      "learning_rate": 2.433891781659603e-06,
      "loss": 0.2591,
      "step": 652
    },
    {
      "epoch": 1.3229813664596273,
      "grad_norm": 0.16029834747314453,
      "learning_rate": 2.41980372195163e-06,
      "loss": 0.2418,
      "step": 653
    },
    {
      "epoch": 1.3250517598343685,
      "grad_norm": 0.17719939351081848,
      "learning_rate": 2.4057435257851173e-06,
      "loss": 0.3019,
      "step": 654
    },
    {
      "epoch": 1.3271221532091098,
      "grad_norm": 0.17115546762943268,
      "learning_rate": 2.391711344996128e-06,
      "loss": 0.2524,
      "step": 655
    },
    {
      "epoch": 1.329192546583851,
      "grad_norm": 0.19674405455589294,
      "learning_rate": 2.377707331118196e-06,
      "loss": 0.3467,
      "step": 656
    },
    {
      "epoch": 1.331262939958592,
      "grad_norm": 0.19814486801624298,
      "learning_rate": 2.363731635380673e-06,
      "loss": 0.3938,
      "step": 657
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.19398008286952972,
      "learning_rate": 2.349784408707112e-06,
      "loss": 0.3151,
      "step": 658
    },
    {
      "epoch": 1.3354037267080745,
      "grad_norm": 0.19555437564849854,
      "learning_rate": 2.33586580171362e-06,
      "loss": 0.3578,
      "step": 659
    },
    {
      "epoch": 1.3374741200828157,
      "grad_norm": 0.19209536910057068,
      "learning_rate": 2.3219759647072467e-06,
      "loss": 0.2102,
      "step": 660
    },
    {
      "epoch": 1.339544513457557,
      "grad_norm": 0.1797213852405548,
      "learning_rate": 2.3081150476843484e-06,
      "loss": 0.3129,
      "step": 661
    },
    {
      "epoch": 1.341614906832298,
      "grad_norm": 0.21602602303028107,
      "learning_rate": 2.2942832003289823e-06,
      "loss": 0.3156,
      "step": 662
    },
    {
      "epoch": 1.3436853002070395,
      "grad_norm": 0.15796010196208954,
      "learning_rate": 2.280480572011274e-06,
      "loss": 0.2369,
      "step": 663
    },
    {
      "epoch": 1.3457556935817805,
      "grad_norm": 0.2077028453350067,
      "learning_rate": 2.2667073117858185e-06,
      "loss": 0.3046,
      "step": 664
    },
    {
      "epoch": 1.3478260869565217,
      "grad_norm": 0.24971678853034973,
      "learning_rate": 2.252963568390064e-06,
      "loss": 0.3312,
      "step": 665
    },
    {
      "epoch": 1.349896480331263,
      "grad_norm": 0.2832271456718445,
      "learning_rate": 2.2392494902427027e-06,
      "loss": 0.3522,
      "step": 666
    },
    {
      "epoch": 1.3519668737060042,
      "grad_norm": 0.1852179914712906,
      "learning_rate": 2.2255652254420774e-06,
      "loss": 0.3036,
      "step": 667
    },
    {
      "epoch": 1.3540372670807455,
      "grad_norm": 0.21870185434818268,
      "learning_rate": 2.2119109217645697e-06,
      "loss": 0.4147,
      "step": 668
    },
    {
      "epoch": 1.3561076604554865,
      "grad_norm": 0.16570810973644257,
      "learning_rate": 2.1982867266630185e-06,
      "loss": 0.2208,
      "step": 669
    },
    {
      "epoch": 1.3581780538302277,
      "grad_norm": 0.1982434242963791,
      "learning_rate": 2.1846927872651135e-06,
      "loss": 0.2939,
      "step": 670
    },
    {
      "epoch": 1.360248447204969,
      "grad_norm": 0.17540566623210907,
      "learning_rate": 2.171129250371819e-06,
      "loss": 0.241,
      "step": 671
    },
    {
      "epoch": 1.3623188405797102,
      "grad_norm": 0.2423764020204544,
      "learning_rate": 2.1575962624557754e-06,
      "loss": 0.3772,
      "step": 672
    },
    {
      "epoch": 1.3643892339544514,
      "grad_norm": 0.15541775524616241,
      "learning_rate": 2.1440939696597323e-06,
      "loss": 0.2138,
      "step": 673
    },
    {
      "epoch": 1.3664596273291925,
      "grad_norm": 0.16889381408691406,
      "learning_rate": 2.1306225177949584e-06,
      "loss": 0.3073,
      "step": 674
    },
    {
      "epoch": 1.3685300207039337,
      "grad_norm": 0.17894522845745087,
      "learning_rate": 2.1171820523396737e-06,
      "loss": 0.3637,
      "step": 675
    },
    {
      "epoch": 1.370600414078675,
      "grad_norm": 0.17433612048625946,
      "learning_rate": 2.1037727184374705e-06,
      "loss": 0.3006,
      "step": 676
    },
    {
      "epoch": 1.3726708074534162,
      "grad_norm": 0.1843237578868866,
      "learning_rate": 2.09039466089576e-06,
      "loss": 0.3195,
      "step": 677
    },
    {
      "epoch": 1.3747412008281574,
      "grad_norm": 0.18528933823108673,
      "learning_rate": 2.07704802418419e-06,
      "loss": 0.2284,
      "step": 678
    },
    {
      "epoch": 1.3768115942028984,
      "grad_norm": 0.28900831937789917,
      "learning_rate": 2.0637329524331044e-06,
      "loss": 0.4572,
      "step": 679
    },
    {
      "epoch": 1.3788819875776397,
      "grad_norm": 0.23907282948493958,
      "learning_rate": 2.050449589431969e-06,
      "loss": 0.2616,
      "step": 680
    },
    {
      "epoch": 1.380952380952381,
      "grad_norm": 0.29129111766815186,
      "learning_rate": 2.0371980786278346e-06,
      "loss": 0.4761,
      "step": 681
    },
    {
      "epoch": 1.3830227743271222,
      "grad_norm": 0.186233788728714,
      "learning_rate": 2.023978563123771e-06,
      "loss": 0.3047,
      "step": 682
    },
    {
      "epoch": 1.3850931677018634,
      "grad_norm": 0.22705401480197906,
      "learning_rate": 2.010791185677337e-06,
      "loss": 0.3393,
      "step": 683
    },
    {
      "epoch": 1.3871635610766044,
      "grad_norm": 0.2112823873758316,
      "learning_rate": 1.997636088699035e-06,
      "loss": 0.3082,
      "step": 684
    },
    {
      "epoch": 1.3892339544513457,
      "grad_norm": 0.2420927882194519,
      "learning_rate": 1.9845134142507615e-06,
      "loss": 0.312,
      "step": 685
    },
    {
      "epoch": 1.391304347826087,
      "grad_norm": 0.18831101059913635,
      "learning_rate": 1.9714233040442915e-06,
      "loss": 0.2822,
      "step": 686
    },
    {
      "epoch": 1.3933747412008282,
      "grad_norm": 0.22394514083862305,
      "learning_rate": 1.958365899439731e-06,
      "loss": 0.2924,
      "step": 687
    },
    {
      "epoch": 1.3954451345755694,
      "grad_norm": 0.23200605809688568,
      "learning_rate": 1.9453413414440043e-06,
      "loss": 0.3346,
      "step": 688
    },
    {
      "epoch": 1.3975155279503104,
      "grad_norm": 0.2842341661453247,
      "learning_rate": 1.93234977070932e-06,
      "loss": 0.3364,
      "step": 689
    },
    {
      "epoch": 1.3995859213250519,
      "grad_norm": 0.339133083820343,
      "learning_rate": 1.919391327531663e-06,
      "loss": 0.3318,
      "step": 690
    },
    {
      "epoch": 1.401656314699793,
      "grad_norm": 0.2594052255153656,
      "learning_rate": 1.9064661518492666e-06,
      "loss": 0.3141,
      "step": 691
    },
    {
      "epoch": 1.4037267080745341,
      "grad_norm": 0.1589246690273285,
      "learning_rate": 1.8935743832411163e-06,
      "loss": 0.2184,
      "step": 692
    },
    {
      "epoch": 1.4057971014492754,
      "grad_norm": 0.21517537534236908,
      "learning_rate": 1.8807161609254254e-06,
      "loss": 0.3006,
      "step": 693
    },
    {
      "epoch": 1.4078674948240166,
      "grad_norm": 0.18755370378494263,
      "learning_rate": 1.8678916237581524e-06,
      "loss": 0.3264,
      "step": 694
    },
    {
      "epoch": 1.4099378881987579,
      "grad_norm": 0.15639418363571167,
      "learning_rate": 1.8551009102314782e-06,
      "loss": 0.2196,
      "step": 695
    },
    {
      "epoch": 1.412008281573499,
      "grad_norm": 0.2891353964805603,
      "learning_rate": 1.8423441584723312e-06,
      "loss": 0.2758,
      "step": 696
    },
    {
      "epoch": 1.4140786749482401,
      "grad_norm": 0.27669522166252136,
      "learning_rate": 1.8296215062408785e-06,
      "loss": 0.407,
      "step": 697
    },
    {
      "epoch": 1.4161490683229814,
      "grad_norm": 0.1550840139389038,
      "learning_rate": 1.816933090929055e-06,
      "loss": 0.1668,
      "step": 698
    },
    {
      "epoch": 1.4182194616977226,
      "grad_norm": 0.21120239794254303,
      "learning_rate": 1.8042790495590629e-06,
      "loss": 0.3332,
      "step": 699
    },
    {
      "epoch": 1.4202898550724639,
      "grad_norm": 0.21207259595394135,
      "learning_rate": 1.791659518781908e-06,
      "loss": 0.2253,
      "step": 700
    },
    {
      "epoch": 1.4223602484472049,
      "grad_norm": 0.15627498924732208,
      "learning_rate": 1.779074634875908e-06,
      "loss": 0.1653,
      "step": 701
    },
    {
      "epoch": 1.4244306418219461,
      "grad_norm": 0.2120233029127121,
      "learning_rate": 1.7665245337452368e-06,
      "loss": 0.2852,
      "step": 702
    },
    {
      "epoch": 1.4265010351966874,
      "grad_norm": 0.25371691584587097,
      "learning_rate": 1.7540093509184426e-06,
      "loss": 0.3425,
      "step": 703
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 0.2664846181869507,
      "learning_rate": 1.7415292215469948e-06,
      "loss": 0.4242,
      "step": 704
    },
    {
      "epoch": 1.4306418219461698,
      "grad_norm": 0.20060203969478607,
      "learning_rate": 1.7290842804038215e-06,
      "loss": 0.2486,
      "step": 705
    },
    {
      "epoch": 1.4327122153209109,
      "grad_norm": 0.17833402752876282,
      "learning_rate": 1.716674661881848e-06,
      "loss": 0.322,
      "step": 706
    },
    {
      "epoch": 1.434782608695652,
      "grad_norm": 0.15655331313610077,
      "learning_rate": 1.7043004999925561e-06,
      "loss": 0.27,
      "step": 707
    },
    {
      "epoch": 1.4368530020703933,
      "grad_norm": 0.3605753481388092,
      "learning_rate": 1.6919619283645262e-06,
      "loss": 0.2902,
      "step": 708
    },
    {
      "epoch": 1.4389233954451346,
      "grad_norm": 0.2352147102355957,
      "learning_rate": 1.6796590802420054e-06,
      "loss": 0.3429,
      "step": 709
    },
    {
      "epoch": 1.4409937888198758,
      "grad_norm": 0.2271387279033661,
      "learning_rate": 1.667392088483456e-06,
      "loss": 0.3038,
      "step": 710
    },
    {
      "epoch": 1.4430641821946169,
      "grad_norm": 0.20507076382637024,
      "learning_rate": 1.6551610855601336e-06,
      "loss": 0.3321,
      "step": 711
    },
    {
      "epoch": 1.4451345755693583,
      "grad_norm": 0.192070871591568,
      "learning_rate": 1.6429662035546451e-06,
      "loss": 0.2685,
      "step": 712
    },
    {
      "epoch": 1.4472049689440993,
      "grad_norm": 0.18269428610801697,
      "learning_rate": 1.6308075741595313e-06,
      "loss": 0.2416,
      "step": 713
    },
    {
      "epoch": 1.4492753623188406,
      "grad_norm": 0.274029403924942,
      "learning_rate": 1.6186853286758397e-06,
      "loss": 0.4039,
      "step": 714
    },
    {
      "epoch": 1.4513457556935818,
      "grad_norm": 0.19074054062366486,
      "learning_rate": 1.60659959801171e-06,
      "loss": 0.3047,
      "step": 715
    },
    {
      "epoch": 1.453416149068323,
      "grad_norm": 0.20888854563236237,
      "learning_rate": 1.5945505126809524e-06,
      "loss": 0.3014,
      "step": 716
    },
    {
      "epoch": 1.4554865424430643,
      "grad_norm": 0.29325804114341736,
      "learning_rate": 1.5825382028016516e-06,
      "loss": 0.4245,
      "step": 717
    },
    {
      "epoch": 1.4575569358178053,
      "grad_norm": 0.23209422826766968,
      "learning_rate": 1.570562798094747e-06,
      "loss": 0.2608,
      "step": 718
    },
    {
      "epoch": 1.4596273291925466,
      "grad_norm": 0.19618046283721924,
      "learning_rate": 1.5586244278826469e-06,
      "loss": 0.2987,
      "step": 719
    },
    {
      "epoch": 1.4616977225672878,
      "grad_norm": 0.24281901121139526,
      "learning_rate": 1.5467232210878153e-06,
      "loss": 0.3474,
      "step": 720
    },
    {
      "epoch": 1.463768115942029,
      "grad_norm": 0.2319604754447937,
      "learning_rate": 1.5348593062313966e-06,
      "loss": 0.3492,
      "step": 721
    },
    {
      "epoch": 1.4658385093167703,
      "grad_norm": 0.19683247804641724,
      "learning_rate": 1.5230328114318127e-06,
      "loss": 0.2443,
      "step": 722
    },
    {
      "epoch": 1.4679089026915113,
      "grad_norm": 0.2371407449245453,
      "learning_rate": 1.5112438644033917e-06,
      "loss": 0.3228,
      "step": 723
    },
    {
      "epoch": 1.4699792960662525,
      "grad_norm": 0.19741837680339813,
      "learning_rate": 1.4994925924549797e-06,
      "loss": 0.2199,
      "step": 724
    },
    {
      "epoch": 1.4720496894409938,
      "grad_norm": 0.1965048760175705,
      "learning_rate": 1.4877791224885729e-06,
      "loss": 0.2673,
      "step": 725
    },
    {
      "epoch": 1.474120082815735,
      "grad_norm": 0.20242713391780853,
      "learning_rate": 1.4761035809979395e-06,
      "loss": 0.2455,
      "step": 726
    },
    {
      "epoch": 1.474120082815735,
      "eval_loss": 0.36622509360313416,
      "eval_runtime": 93.169,
      "eval_samples_per_second": 5.925,
      "eval_steps_per_second": 0.741,
      "step": 726
    },
    {
      "epoch": 1.4761904761904763,
      "grad_norm": 0.22379763424396515,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.2963,
      "step": 727
    },
    {
      "epoch": 1.4782608695652173,
      "grad_norm": 0.20382773876190186,
      "learning_rate": 1.452866787369771e-06,
      "loss": 0.3634,
      "step": 728
    },
    {
      "epoch": 1.4803312629399585,
      "grad_norm": 0.1743982583284378,
      "learning_rate": 1.4413057861663843e-06,
      "loss": 0.2676,
      "step": 729
    },
    {
      "epoch": 1.4824016563146998,
      "grad_norm": 0.21269293129444122,
      "learning_rate": 1.4297832153043657e-06,
      "loss": 0.3427,
      "step": 730
    },
    {
      "epoch": 1.484472049689441,
      "grad_norm": 0.15802772343158722,
      "learning_rate": 1.418299199215963e-06,
      "loss": 0.1784,
      "step": 731
    },
    {
      "epoch": 1.4865424430641823,
      "grad_norm": 0.1924794614315033,
      "learning_rate": 1.4068538619170763e-06,
      "loss": 0.2559,
      "step": 732
    },
    {
      "epoch": 1.4886128364389233,
      "grad_norm": 0.19855406880378723,
      "learning_rate": 1.3954473270059104e-06,
      "loss": 0.3105,
      "step": 733
    },
    {
      "epoch": 1.4906832298136645,
      "grad_norm": 0.17971496284008026,
      "learning_rate": 1.3840797176616467e-06,
      "loss": 0.1931,
      "step": 734
    },
    {
      "epoch": 1.4927536231884058,
      "grad_norm": 0.23061345517635345,
      "learning_rate": 1.3727511566431024e-06,
      "loss": 0.3635,
      "step": 735
    },
    {
      "epoch": 1.494824016563147,
      "grad_norm": 0.21202407777309418,
      "learning_rate": 1.3614617662874197e-06,
      "loss": 0.3096,
      "step": 736
    },
    {
      "epoch": 1.4968944099378882,
      "grad_norm": 0.19486172497272491,
      "learning_rate": 1.3502116685087303e-06,
      "loss": 0.2196,
      "step": 737
    },
    {
      "epoch": 1.4989648033126293,
      "grad_norm": 0.3029009699821472,
      "learning_rate": 1.3390009847968505e-06,
      "loss": 0.2623,
      "step": 738
    },
    {
      "epoch": 1.5010351966873707,
      "grad_norm": 0.21870052814483643,
      "learning_rate": 1.3278298362159598e-06,
      "loss": 0.3429,
      "step": 739
    },
    {
      "epoch": 1.5031055900621118,
      "grad_norm": 0.21567311882972717,
      "learning_rate": 1.316698343403302e-06,
      "loss": 0.2831,
      "step": 740
    },
    {
      "epoch": 1.505175983436853,
      "grad_norm": 0.20199137926101685,
      "learning_rate": 1.305606626567873e-06,
      "loss": 0.3005,
      "step": 741
    },
    {
      "epoch": 1.5072463768115942,
      "grad_norm": 0.17147856950759888,
      "learning_rate": 1.2945548054891322e-06,
      "loss": 0.235,
      "step": 742
    },
    {
      "epoch": 1.5093167701863353,
      "grad_norm": 0.1645573228597641,
      "learning_rate": 1.283542999515704e-06,
      "loss": 0.1795,
      "step": 743
    },
    {
      "epoch": 1.5113871635610767,
      "grad_norm": 0.1960628777742386,
      "learning_rate": 1.27257132756409e-06,
      "loss": 0.2717,
      "step": 744
    },
    {
      "epoch": 1.5134575569358177,
      "grad_norm": 0.2242807298898697,
      "learning_rate": 1.2616399081173802e-06,
      "loss": 0.2269,
      "step": 745
    },
    {
      "epoch": 1.515527950310559,
      "grad_norm": 0.20267319679260254,
      "learning_rate": 1.2507488592239848e-06,
      "loss": 0.2729,
      "step": 746
    },
    {
      "epoch": 1.5175983436853002,
      "grad_norm": 0.16490796208381653,
      "learning_rate": 1.2398982984963454e-06,
      "loss": 0.1948,
      "step": 747
    },
    {
      "epoch": 1.5196687370600412,
      "grad_norm": 0.21590249240398407,
      "learning_rate": 1.2290883431096778e-06,
      "loss": 0.4181,
      "step": 748
    },
    {
      "epoch": 1.5217391304347827,
      "grad_norm": 0.1928826868534088,
      "learning_rate": 1.2183191098006948e-06,
      "loss": 0.2692,
      "step": 749
    },
    {
      "epoch": 1.5238095238095237,
      "grad_norm": 0.2254112809896469,
      "learning_rate": 1.2075907148663579e-06,
      "loss": 0.2489,
      "step": 750
    },
    {
      "epoch": 1.525879917184265,
      "grad_norm": 0.20881828665733337,
      "learning_rate": 1.196903274162609e-06,
      "loss": 0.2952,
      "step": 751
    },
    {
      "epoch": 1.5279503105590062,
      "grad_norm": 0.1981811821460724,
      "learning_rate": 1.186256903103129e-06,
      "loss": 0.2278,
      "step": 752
    },
    {
      "epoch": 1.5300207039337475,
      "grad_norm": 0.1390255093574524,
      "learning_rate": 1.1756517166580883e-06,
      "loss": 0.1711,
      "step": 753
    },
    {
      "epoch": 1.5320910973084887,
      "grad_norm": 0.20273476839065552,
      "learning_rate": 1.1650878293528994e-06,
      "loss": 0.3948,
      "step": 754
    },
    {
      "epoch": 1.5341614906832297,
      "grad_norm": 0.2850019633769989,
      "learning_rate": 1.1545653552669928e-06,
      "loss": 0.3879,
      "step": 755
    },
    {
      "epoch": 1.5362318840579712,
      "grad_norm": 0.1915498822927475,
      "learning_rate": 1.1440844080325703e-06,
      "loss": 0.289,
      "step": 756
    },
    {
      "epoch": 1.5383022774327122,
      "grad_norm": 0.2492920160293579,
      "learning_rate": 1.133645100833391e-06,
      "loss": 0.3461,
      "step": 757
    },
    {
      "epoch": 1.5403726708074534,
      "grad_norm": 0.1911374032497406,
      "learning_rate": 1.1232475464035386e-06,
      "loss": 0.2607,
      "step": 758
    },
    {
      "epoch": 1.5424430641821947,
      "grad_norm": 0.18775556981563568,
      "learning_rate": 1.1128918570262137e-06,
      "loss": 0.3516,
      "step": 759
    },
    {
      "epoch": 1.5445134575569357,
      "grad_norm": 0.24565483629703522,
      "learning_rate": 1.10257814453251e-06,
      "loss": 0.2813,
      "step": 760
    },
    {
      "epoch": 1.5465838509316772,
      "grad_norm": 0.20192737877368927,
      "learning_rate": 1.0923065203002199e-06,
      "loss": 0.3487,
      "step": 761
    },
    {
      "epoch": 1.5486542443064182,
      "grad_norm": 0.2754063606262207,
      "learning_rate": 1.0820770952526155e-06,
      "loss": 0.2709,
      "step": 762
    },
    {
      "epoch": 1.5507246376811594,
      "grad_norm": 0.19039547443389893,
      "learning_rate": 1.0718899798572713e-06,
      "loss": 0.3251,
      "step": 763
    },
    {
      "epoch": 1.5527950310559007,
      "grad_norm": 0.19267532229423523,
      "learning_rate": 1.0617452841248494e-06,
      "loss": 0.2611,
      "step": 764
    },
    {
      "epoch": 1.5548654244306417,
      "grad_norm": 0.2473609745502472,
      "learning_rate": 1.0516431176079296e-06,
      "loss": 0.2622,
      "step": 765
    },
    {
      "epoch": 1.5569358178053831,
      "grad_norm": 0.19303014874458313,
      "learning_rate": 1.0415835893998116e-06,
      "loss": 0.2595,
      "step": 766
    },
    {
      "epoch": 1.5590062111801242,
      "grad_norm": 0.24576401710510254,
      "learning_rate": 1.0315668081333519e-06,
      "loss": 0.3033,
      "step": 767
    },
    {
      "epoch": 1.5610766045548654,
      "grad_norm": 0.15718652307987213,
      "learning_rate": 1.0215928819797744e-06,
      "loss": 0.2193,
      "step": 768
    },
    {
      "epoch": 1.5631469979296067,
      "grad_norm": 0.22884951531887054,
      "learning_rate": 1.0116619186475185e-06,
      "loss": 0.2461,
      "step": 769
    },
    {
      "epoch": 1.5652173913043477,
      "grad_norm": 0.14403100311756134,
      "learning_rate": 1.0017740253810608e-06,
      "loss": 0.12,
      "step": 770
    },
    {
      "epoch": 1.5672877846790891,
      "grad_norm": 0.2506249248981476,
      "learning_rate": 9.919293089597715e-07,
      "loss": 0.2952,
      "step": 771
    },
    {
      "epoch": 1.5693581780538302,
      "grad_norm": 0.23279593884944916,
      "learning_rate": 9.821278756967467e-07,
      "loss": 0.2258,
      "step": 772
    },
    {
      "epoch": 1.5714285714285714,
      "grad_norm": 0.4160136282444,
      "learning_rate": 9.723698314376733e-07,
      "loss": 0.4735,
      "step": 773
    },
    {
      "epoch": 1.5734989648033126,
      "grad_norm": 0.22142386436462402,
      "learning_rate": 9.62655281559679e-07,
      "loss": 0.2525,
      "step": 774
    },
    {
      "epoch": 1.5755693581780539,
      "grad_norm": 0.20253802835941315,
      "learning_rate": 9.529843309701925e-07,
      "loss": 0.3334,
      "step": 775
    },
    {
      "epoch": 1.5776397515527951,
      "grad_norm": 0.17480939626693726,
      "learning_rate": 9.433570841058187e-07,
      "loss": 0.2547,
      "step": 776
    },
    {
      "epoch": 1.5797101449275361,
      "grad_norm": 0.1820262223482132,
      "learning_rate": 9.337736449312007e-07,
      "loss": 0.2979,
      "step": 777
    },
    {
      "epoch": 1.5817805383022774,
      "grad_norm": 0.23739475011825562,
      "learning_rate": 9.242341169379077e-07,
      "loss": 0.3483,
      "step": 778
    },
    {
      "epoch": 1.5838509316770186,
      "grad_norm": 0.2090044617652893,
      "learning_rate": 9.14738603143307e-07,
      "loss": 0.3042,
      "step": 779
    },
    {
      "epoch": 1.5859213250517599,
      "grad_norm": 0.33553430438041687,
      "learning_rate": 9.052872060894613e-07,
      "loss": 0.266,
      "step": 780
    },
    {
      "epoch": 1.587991718426501,
      "grad_norm": 0.19441604614257812,
      "learning_rate": 8.958800278420116e-07,
      "loss": 0.3754,
      "step": 781
    },
    {
      "epoch": 1.5900621118012421,
      "grad_norm": 0.18226413428783417,
      "learning_rate": 8.865171699890835e-07,
      "loss": 0.2289,
      "step": 782
    },
    {
      "epoch": 1.5921325051759836,
      "grad_norm": 0.1726485639810562,
      "learning_rate": 8.77198733640186e-07,
      "loss": 0.2424,
      "step": 783
    },
    {
      "epoch": 1.5942028985507246,
      "grad_norm": 0.19749626517295837,
      "learning_rate": 8.679248194251211e-07,
      "loss": 0.2693,
      "step": 784
    },
    {
      "epoch": 1.5962732919254659,
      "grad_norm": 0.3344188332557678,
      "learning_rate": 8.586955274928926e-07,
      "loss": 0.4119,
      "step": 785
    },
    {
      "epoch": 1.598343685300207,
      "grad_norm": 0.1941257119178772,
      "learning_rate": 8.495109575106331e-07,
      "loss": 0.2691,
      "step": 786
    },
    {
      "epoch": 1.6004140786749481,
      "grad_norm": 0.23682045936584473,
      "learning_rate": 8.403712086625176e-07,
      "loss": 0.399,
      "step": 787
    },
    {
      "epoch": 1.6024844720496896,
      "grad_norm": 0.17793674767017365,
      "learning_rate": 8.312763796487038e-07,
      "loss": 0.2604,
      "step": 788
    },
    {
      "epoch": 1.6045548654244306,
      "grad_norm": 0.20170265436172485,
      "learning_rate": 8.22226568684254e-07,
      "loss": 0.2313,
      "step": 789
    },
    {
      "epoch": 1.6066252587991718,
      "grad_norm": 0.22592318058013916,
      "learning_rate": 8.132218734980852e-07,
      "loss": 0.2609,
      "step": 790
    },
    {
      "epoch": 1.608695652173913,
      "grad_norm": 0.1931549608707428,
      "learning_rate": 8.042623913319048e-07,
      "loss": 0.2652,
      "step": 791
    },
    {
      "epoch": 1.610766045548654,
      "grad_norm": 0.2012760043144226,
      "learning_rate": 7.953482189391687e-07,
      "loss": 0.2863,
      "step": 792
    },
    {
      "epoch": 1.6128364389233956,
      "grad_norm": 0.17883974313735962,
      "learning_rate": 7.864794525840325e-07,
      "loss": 0.2197,
      "step": 793
    },
    {
      "epoch": 1.6149068322981366,
      "grad_norm": 0.2195083647966385,
      "learning_rate": 7.776561880403072e-07,
      "loss": 0.3113,
      "step": 794
    },
    {
      "epoch": 1.6169772256728778,
      "grad_norm": 0.27091073989868164,
      "learning_rate": 7.688785205904359e-07,
      "loss": 0.3241,
      "step": 795
    },
    {
      "epoch": 1.619047619047619,
      "grad_norm": 0.2131056785583496,
      "learning_rate": 7.601465450244528e-07,
      "loss": 0.3203,
      "step": 796
    },
    {
      "epoch": 1.62111801242236,
      "grad_norm": 0.16343261301517487,
      "learning_rate": 7.514603556389716e-07,
      "loss": 0.1965,
      "step": 797
    },
    {
      "epoch": 1.6231884057971016,
      "grad_norm": 0.17004933953285217,
      "learning_rate": 7.42820046236154e-07,
      "loss": 0.2833,
      "step": 798
    },
    {
      "epoch": 1.6252587991718426,
      "grad_norm": 0.17998455464839935,
      "learning_rate": 7.342257101227112e-07,
      "loss": 0.2284,
      "step": 799
    },
    {
      "epoch": 1.6273291925465838,
      "grad_norm": 0.16908283531665802,
      "learning_rate": 7.256774401088817e-07,
      "loss": 0.2292,
      "step": 800
    },
    {
      "epoch": 1.629399585921325,
      "grad_norm": 0.2141779363155365,
      "learning_rate": 7.171753285074424e-07,
      "loss": 0.332,
      "step": 801
    },
    {
      "epoch": 1.6314699792960663,
      "grad_norm": 0.26314640045166016,
      "learning_rate": 7.087194671326986e-07,
      "loss": 0.4573,
      "step": 802
    },
    {
      "epoch": 1.6335403726708075,
      "grad_norm": 0.20474179089069366,
      "learning_rate": 7.003099472995084e-07,
      "loss": 0.3609,
      "step": 803
    },
    {
      "epoch": 1.6356107660455486,
      "grad_norm": 0.2234301120042801,
      "learning_rate": 6.91946859822279e-07,
      "loss": 0.315,
      "step": 804
    },
    {
      "epoch": 1.6376811594202898,
      "grad_norm": 0.22140094637870789,
      "learning_rate": 6.83630295014e-07,
      "loss": 0.3776,
      "step": 805
    },
    {
      "epoch": 1.639751552795031,
      "grad_norm": 0.35324928164482117,
      "learning_rate": 6.753603426852589e-07,
      "loss": 0.3518,
      "step": 806
    },
    {
      "epoch": 1.6418219461697723,
      "grad_norm": 0.23616626858711243,
      "learning_rate": 6.67137092143279e-07,
      "loss": 0.2792,
      "step": 807
    },
    {
      "epoch": 1.6438923395445135,
      "grad_norm": 0.21780608594417572,
      "learning_rate": 6.589606321909464e-07,
      "loss": 0.3164,
      "step": 808
    },
    {
      "epoch": 1.6459627329192545,
      "grad_norm": 0.224593847990036,
      "learning_rate": 6.508310511258603e-07,
      "loss": 0.2601,
      "step": 809
    },
    {
      "epoch": 1.648033126293996,
      "grad_norm": 0.22506175935268402,
      "learning_rate": 6.427484367393699e-07,
      "loss": 0.3016,
      "step": 810
    },
    {
      "epoch": 1.650103519668737,
      "grad_norm": 0.2013794630765915,
      "learning_rate": 6.347128763156363e-07,
      "loss": 0.2155,
      "step": 811
    },
    {
      "epoch": 1.6521739130434783,
      "grad_norm": 0.210536926984787,
      "learning_rate": 6.267244566306801e-07,
      "loss": 0.3181,
      "step": 812
    },
    {
      "epoch": 1.6542443064182195,
      "grad_norm": 0.2421325147151947,
      "learning_rate": 6.187832639514524e-07,
      "loss": 0.2719,
      "step": 813
    },
    {
      "epoch": 1.6563146997929605,
      "grad_norm": 0.2348569929599762,
      "learning_rate": 6.108893840348995e-07,
      "loss": 0.3543,
      "step": 814
    },
    {
      "epoch": 1.658385093167702,
      "grad_norm": 0.20567581057548523,
      "learning_rate": 6.030429021270346e-07,
      "loss": 0.3187,
      "step": 815
    },
    {
      "epoch": 1.660455486542443,
      "grad_norm": 0.22527620196342468,
      "learning_rate": 5.952439029620222e-07,
      "loss": 0.3088,
      "step": 816
    },
    {
      "epoch": 1.6625258799171843,
      "grad_norm": 0.15716291964054108,
      "learning_rate": 5.874924707612595e-07,
      "loss": 0.2427,
      "step": 817
    },
    {
      "epoch": 1.6645962732919255,
      "grad_norm": 0.2059607058763504,
      "learning_rate": 5.797886892324695e-07,
      "loss": 0.3054,
      "step": 818
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.21156099438667297,
      "learning_rate": 5.721326415687928e-07,
      "loss": 0.3218,
      "step": 819
    },
    {
      "epoch": 1.668737060041408,
      "grad_norm": 0.2147274613380432,
      "learning_rate": 5.645244104478947e-07,
      "loss": 0.2965,
      "step": 820
    },
    {
      "epoch": 1.670807453416149,
      "grad_norm": 0.24699914455413818,
      "learning_rate": 5.569640780310681e-07,
      "loss": 0.3731,
      "step": 821
    },
    {
      "epoch": 1.6728778467908902,
      "grad_norm": 0.1968562752008438,
      "learning_rate": 5.494517259623478e-07,
      "loss": 0.2637,
      "step": 822
    },
    {
      "epoch": 1.6749482401656315,
      "grad_norm": 0.19729909300804138,
      "learning_rate": 5.419874353676302e-07,
      "loss": 0.2993,
      "step": 823
    },
    {
      "epoch": 1.6770186335403725,
      "grad_norm": 0.21764995157718658,
      "learning_rate": 5.34571286853795e-07,
      "loss": 0.4701,
      "step": 824
    },
    {
      "epoch": 1.679089026915114,
      "grad_norm": 0.2567196190357208,
      "learning_rate": 5.272033605078336e-07,
      "loss": 0.306,
      "step": 825
    },
    {
      "epoch": 1.681159420289855,
      "grad_norm": 0.19084686040878296,
      "learning_rate": 5.198837358959901e-07,
      "loss": 0.3407,
      "step": 826
    },
    {
      "epoch": 1.6832298136645962,
      "grad_norm": 0.18046824634075165,
      "learning_rate": 5.126124920628939e-07,
      "loss": 0.1684,
      "step": 827
    },
    {
      "epoch": 1.6853002070393375,
      "grad_norm": 0.2360648661851883,
      "learning_rate": 5.05389707530714e-07,
      "loss": 0.3395,
      "step": 828
    },
    {
      "epoch": 1.6873706004140787,
      "grad_norm": 0.2442813515663147,
      "learning_rate": 4.982154602983042e-07,
      "loss": 0.3327,
      "step": 829
    },
    {
      "epoch": 1.68944099378882,
      "grad_norm": 0.24826480448246002,
      "learning_rate": 4.91089827840367e-07,
      "loss": 0.3129,
      "step": 830
    },
    {
      "epoch": 1.691511387163561,
      "grad_norm": 0.35874247550964355,
      "learning_rate": 4.840128871066107e-07,
      "loss": 0.3476,
      "step": 831
    },
    {
      "epoch": 1.6935817805383024,
      "grad_norm": 0.2496299296617508,
      "learning_rate": 4.769847145209244e-07,
      "loss": 0.4227,
      "step": 832
    },
    {
      "epoch": 1.6956521739130435,
      "grad_norm": 0.20718629658222198,
      "learning_rate": 4.700053859805498e-07,
      "loss": 0.2448,
      "step": 833
    },
    {
      "epoch": 1.6977225672877847,
      "grad_norm": 0.1962181031703949,
      "learning_rate": 4.6307497685525894e-07,
      "loss": 0.3321,
      "step": 834
    },
    {
      "epoch": 1.699792960662526,
      "grad_norm": 0.20894816517829895,
      "learning_rate": 4.5619356198654654e-07,
      "loss": 0.3279,
      "step": 835
    },
    {
      "epoch": 1.701863354037267,
      "grad_norm": 0.20937037467956543,
      "learning_rate": 4.4936121568681546e-07,
      "loss": 0.2815,
      "step": 836
    },
    {
      "epoch": 1.7039337474120084,
      "grad_norm": 0.220305398106575,
      "learning_rate": 4.425780117385797e-07,
      "loss": 0.2257,
      "step": 837
    },
    {
      "epoch": 1.7060041407867494,
      "grad_norm": 0.2656574249267578,
      "learning_rate": 4.3584402339366174e-07,
      "loss": 0.3553,
      "step": 838
    },
    {
      "epoch": 1.7080745341614907,
      "grad_norm": 0.23722606897354126,
      "learning_rate": 4.291593233724084e-07,
      "loss": 0.4194,
      "step": 839
    },
    {
      "epoch": 1.710144927536232,
      "grad_norm": 0.21551856398582458,
      "learning_rate": 4.225239838628981e-07,
      "loss": 0.3319,
      "step": 840
    },
    {
      "epoch": 1.712215320910973,
      "grad_norm": 0.24776722490787506,
      "learning_rate": 4.159380765201687e-07,
      "loss": 0.3984,
      "step": 841
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 0.20069292187690735,
      "learning_rate": 4.0940167246543595e-07,
      "loss": 0.3096,
      "step": 842
    },
    {
      "epoch": 1.7163561076604554,
      "grad_norm": 0.19585564732551575,
      "learning_rate": 4.0291484228533574e-07,
      "loss": 0.2744,
      "step": 843
    },
    {
      "epoch": 1.7184265010351967,
      "grad_norm": 0.2340921312570572,
      "learning_rate": 3.964776560311484e-07,
      "loss": 0.3995,
      "step": 844
    },
    {
      "epoch": 1.720496894409938,
      "grad_norm": 0.18698614835739136,
      "learning_rate": 3.900901832180548e-07,
      "loss": 0.2305,
      "step": 845
    },
    {
      "epoch": 1.722567287784679,
      "grad_norm": 0.16317184269428253,
      "learning_rate": 3.8375249282437743e-07,
      "loss": 0.1814,
      "step": 846
    },
    {
      "epoch": 1.7246376811594204,
      "grad_norm": 0.1703299582004547,
      "learning_rate": 3.774646532908405e-07,
      "loss": 0.2401,
      "step": 847
    },
    {
      "epoch": 1.7246376811594204,
      "eval_loss": 0.36501544713974,
      "eval_runtime": 93.4662,
      "eval_samples_per_second": 5.906,
      "eval_steps_per_second": 0.738,
      "step": 847
    },
    {
      "epoch": 1.7267080745341614,
      "grad_norm": 0.22787785530090332,
      "learning_rate": 3.71226732519826e-07,
      "loss": 0.2781,
      "step": 848
    },
    {
      "epoch": 1.7287784679089027,
      "grad_norm": 0.2101709544658661,
      "learning_rate": 3.6503879787464614e-07,
      "loss": 0.2191,
      "step": 849
    },
    {
      "epoch": 1.730848861283644,
      "grad_norm": 0.18964454531669617,
      "learning_rate": 3.589009161788104e-07,
      "loss": 0.2831,
      "step": 850
    },
    {
      "epoch": 1.7329192546583851,
      "grad_norm": 0.2441266030073166,
      "learning_rate": 3.5281315371530953e-07,
      "loss": 0.3778,
      "step": 851
    },
    {
      "epoch": 1.7349896480331264,
      "grad_norm": 0.20503799617290497,
      "learning_rate": 3.4677557622589175e-07,
      "loss": 0.3275,
      "step": 852
    },
    {
      "epoch": 1.7370600414078674,
      "grad_norm": 0.28412970900535583,
      "learning_rate": 3.407882489103642e-07,
      "loss": 0.2926,
      "step": 853
    },
    {
      "epoch": 1.7391304347826086,
      "grad_norm": 0.20368856191635132,
      "learning_rate": 3.3485123642587657e-07,
      "loss": 0.2754,
      "step": 854
    },
    {
      "epoch": 1.74120082815735,
      "grad_norm": 0.18978296220302582,
      "learning_rate": 3.28964602886232e-07,
      "loss": 0.3035,
      "step": 855
    },
    {
      "epoch": 1.7432712215320911,
      "grad_norm": 0.26132866740226746,
      "learning_rate": 3.2312841186118937e-07,
      "loss": 0.3749,
      "step": 856
    },
    {
      "epoch": 1.7453416149068324,
      "grad_norm": 0.21782605350017548,
      "learning_rate": 3.1734272637578113e-07,
      "loss": 0.2398,
      "step": 857
    },
    {
      "epoch": 1.7474120082815734,
      "grad_norm": 0.24689531326293945,
      "learning_rate": 3.116076089096265e-07,
      "loss": 0.3806,
      "step": 858
    },
    {
      "epoch": 1.7494824016563149,
      "grad_norm": 0.27699175477027893,
      "learning_rate": 3.05923121396266e-07,
      "loss": 0.4021,
      "step": 859
    },
    {
      "epoch": 1.7515527950310559,
      "grad_norm": 0.263994425535202,
      "learning_rate": 3.0028932522248256e-07,
      "loss": 0.3092,
      "step": 860
    },
    {
      "epoch": 1.7536231884057971,
      "grad_norm": 0.21080216765403748,
      "learning_rate": 2.9470628122764575e-07,
      "loss": 0.4134,
      "step": 861
    },
    {
      "epoch": 1.7556935817805384,
      "grad_norm": 0.21831892430782318,
      "learning_rate": 2.8917404970305096e-07,
      "loss": 0.2616,
      "step": 862
    },
    {
      "epoch": 1.7577639751552794,
      "grad_norm": 0.22182129323482513,
      "learning_rate": 2.8369269039127124e-07,
      "loss": 0.3379,
      "step": 863
    },
    {
      "epoch": 1.7598343685300208,
      "grad_norm": 0.20161934196949005,
      "learning_rate": 2.782622624855097e-07,
      "loss": 0.3677,
      "step": 864
    },
    {
      "epoch": 1.7619047619047619,
      "grad_norm": 0.20959687232971191,
      "learning_rate": 2.728828246289605e-07,
      "loss": 0.328,
      "step": 865
    },
    {
      "epoch": 1.763975155279503,
      "grad_norm": 0.22280235588550568,
      "learning_rate": 2.6755443491417786e-07,
      "loss": 0.341,
      "step": 866
    },
    {
      "epoch": 1.7660455486542443,
      "grad_norm": 0.19911250472068787,
      "learning_rate": 2.622771508824451e-07,
      "loss": 0.2035,
      "step": 867
    },
    {
      "epoch": 1.7681159420289854,
      "grad_norm": 0.2472066879272461,
      "learning_rate": 2.570510295231571e-07,
      "loss": 0.3076,
      "step": 868
    },
    {
      "epoch": 1.7701863354037268,
      "grad_norm": 0.2838335335254669,
      "learning_rate": 2.518761272732012e-07,
      "loss": 0.3175,
      "step": 869
    },
    {
      "epoch": 1.7722567287784678,
      "grad_norm": 0.1958410143852234,
      "learning_rate": 2.467525000163523e-07,
      "loss": 0.328,
      "step": 870
    },
    {
      "epoch": 1.774327122153209,
      "grad_norm": 0.21654951572418213,
      "learning_rate": 2.416802030826626e-07,
      "loss": 0.3124,
      "step": 871
    },
    {
      "epoch": 1.7763975155279503,
      "grad_norm": 0.21129873394966125,
      "learning_rate": 2.36659291247871e-07,
      "loss": 0.3465,
      "step": 872
    },
    {
      "epoch": 1.7784679089026914,
      "grad_norm": 0.25625911355018616,
      "learning_rate": 2.316898187328087e-07,
      "loss": 0.3733,
      "step": 873
    },
    {
      "epoch": 1.7805383022774328,
      "grad_norm": 0.20912493765354156,
      "learning_rate": 2.2677183920281342e-07,
      "loss": 0.2978,
      "step": 874
    },
    {
      "epoch": 1.7826086956521738,
      "grad_norm": 0.1701553463935852,
      "learning_rate": 2.2190540576714825e-07,
      "loss": 0.2677,
      "step": 875
    },
    {
      "epoch": 1.784679089026915,
      "grad_norm": 0.1535242795944214,
      "learning_rate": 2.1709057097843266e-07,
      "loss": 0.1532,
      "step": 876
    },
    {
      "epoch": 1.7867494824016563,
      "grad_norm": 0.19895993173122406,
      "learning_rate": 2.123273868320691e-07,
      "loss": 0.3049,
      "step": 877
    },
    {
      "epoch": 1.7888198757763976,
      "grad_norm": 0.15035057067871094,
      "learning_rate": 2.0761590476568893e-07,
      "loss": 0.2621,
      "step": 878
    },
    {
      "epoch": 1.7908902691511388,
      "grad_norm": 0.21586264669895172,
      "learning_rate": 2.029561756585885e-07,
      "loss": 0.3422,
      "step": 879
    },
    {
      "epoch": 1.7929606625258798,
      "grad_norm": 0.18950589001178741,
      "learning_rate": 1.9834824983118673e-07,
      "loss": 0.2516,
      "step": 880
    },
    {
      "epoch": 1.795031055900621,
      "grad_norm": 0.163870170712471,
      "learning_rate": 1.9379217704447728e-07,
      "loss": 0.2275,
      "step": 881
    },
    {
      "epoch": 1.7971014492753623,
      "grad_norm": 0.206380695104599,
      "learning_rate": 1.892880064994934e-07,
      "loss": 0.305,
      "step": 882
    },
    {
      "epoch": 1.7991718426501035,
      "grad_norm": 0.17586331069469452,
      "learning_rate": 1.8483578683677783e-07,
      "loss": 0.2527,
      "step": 883
    },
    {
      "epoch": 1.8012422360248448,
      "grad_norm": 0.18862633407115936,
      "learning_rate": 1.8043556613585143e-07,
      "loss": 0.2868,
      "step": 884
    },
    {
      "epoch": 1.8033126293995858,
      "grad_norm": 0.2271614372730255,
      "learning_rate": 1.7608739191470214e-07,
      "loss": 0.3503,
      "step": 885
    },
    {
      "epoch": 1.8053830227743273,
      "grad_norm": 0.20493744313716888,
      "learning_rate": 1.7179131112926628e-07,
      "loss": 0.2666,
      "step": 886
    },
    {
      "epoch": 1.8074534161490683,
      "grad_norm": 0.2284647822380066,
      "learning_rate": 1.67547370172923e-07,
      "loss": 0.2872,
      "step": 887
    },
    {
      "epoch": 1.8095238095238095,
      "grad_norm": 0.18809698522090912,
      "learning_rate": 1.6335561487599406e-07,
      "loss": 0.2607,
      "step": 888
    },
    {
      "epoch": 1.8115942028985508,
      "grad_norm": 0.23940181732177734,
      "learning_rate": 1.592160905052481e-07,
      "loss": 0.3169,
      "step": 889
    },
    {
      "epoch": 1.8136645962732918,
      "grad_norm": 0.16910064220428467,
      "learning_rate": 1.551288417634106e-07,
      "loss": 0.1735,
      "step": 890
    },
    {
      "epoch": 1.8157349896480333,
      "grad_norm": 0.21867364645004272,
      "learning_rate": 1.5109391278868568e-07,
      "loss": 0.3319,
      "step": 891
    },
    {
      "epoch": 1.8178053830227743,
      "grad_norm": 0.22258295118808746,
      "learning_rate": 1.471113471542712e-07,
      "loss": 0.2971,
      "step": 892
    },
    {
      "epoch": 1.8198757763975155,
      "grad_norm": 0.19150149822235107,
      "learning_rate": 1.4318118786789958e-07,
      "loss": 0.4353,
      "step": 893
    },
    {
      "epoch": 1.8219461697722568,
      "grad_norm": 0.21853388845920563,
      "learning_rate": 1.3930347737136195e-07,
      "loss": 0.2208,
      "step": 894
    },
    {
      "epoch": 1.8240165631469978,
      "grad_norm": 0.13483737409114838,
      "learning_rate": 1.3547825754005762e-07,
      "loss": 0.142,
      "step": 895
    },
    {
      "epoch": 1.8260869565217392,
      "grad_norm": 0.1933605819940567,
      "learning_rate": 1.3170556968253756e-07,
      "loss": 0.3566,
      "step": 896
    },
    {
      "epoch": 1.8281573498964803,
      "grad_norm": 0.17729544639587402,
      "learning_rate": 1.2798545454006106e-07,
      "loss": 0.204,
      "step": 897
    },
    {
      "epoch": 1.8302277432712215,
      "grad_norm": 0.2337586134672165,
      "learning_rate": 1.2431795228615372e-07,
      "loss": 0.2763,
      "step": 898
    },
    {
      "epoch": 1.8322981366459627,
      "grad_norm": 0.2700212597846985,
      "learning_rate": 1.2070310252617567e-07,
      "loss": 0.3541,
      "step": 899
    },
    {
      "epoch": 1.8343685300207038,
      "grad_norm": 0.247345969080925,
      "learning_rate": 1.1714094429689127e-07,
      "loss": 0.3693,
      "step": 900
    },
    {
      "epoch": 1.8364389233954452,
      "grad_norm": 0.21636363863945007,
      "learning_rate": 1.1363151606605117e-07,
      "loss": 0.3488,
      "step": 901
    },
    {
      "epoch": 1.8385093167701863,
      "grad_norm": 0.19248615205287933,
      "learning_rate": 1.1017485573197151e-07,
      "loss": 0.346,
      "step": 902
    },
    {
      "epoch": 1.8405797101449275,
      "grad_norm": 0.2149479240179062,
      "learning_rate": 1.0677100062313095e-07,
      "loss": 0.2206,
      "step": 903
    },
    {
      "epoch": 1.8426501035196687,
      "grad_norm": 0.1936839520931244,
      "learning_rate": 1.0341998749776316e-07,
      "loss": 0.2105,
      "step": 904
    },
    {
      "epoch": 1.84472049689441,
      "grad_norm": 0.24326792359352112,
      "learning_rate": 1.0012185254346052e-07,
      "loss": 0.3633,
      "step": 905
    },
    {
      "epoch": 1.8467908902691512,
      "grad_norm": 0.17744815349578857,
      "learning_rate": 9.687663137678605e-08,
      "loss": 0.3085,
      "step": 906
    },
    {
      "epoch": 1.8488612836438922,
      "grad_norm": 0.18408702313899994,
      "learning_rate": 9.368435904288431e-08,
      "loss": 0.2346,
      "step": 907
    },
    {
      "epoch": 1.8509316770186337,
      "grad_norm": 0.27747613191604614,
      "learning_rate": 9.054507001510727e-08,
      "loss": 0.3265,
      "step": 908
    },
    {
      "epoch": 1.8530020703933747,
      "grad_norm": 0.1913614422082901,
      "learning_rate": 8.745879819463843e-08,
      "loss": 0.2061,
      "step": 909
    },
    {
      "epoch": 1.855072463768116,
      "grad_norm": 0.28982409834861755,
      "learning_rate": 8.442557691013042e-08,
      "loss": 0.5099,
      "step": 910
    },
    {
      "epoch": 1.8571428571428572,
      "grad_norm": 0.2792413830757141,
      "learning_rate": 8.144543891734135e-08,
      "loss": 0.4041,
      "step": 911
    },
    {
      "epoch": 1.8592132505175982,
      "grad_norm": 0.22993594408035278,
      "learning_rate": 7.851841639878399e-08,
      "loss": 0.3219,
      "step": 912
    },
    {
      "epoch": 1.8612836438923397,
      "grad_norm": 0.31364697217941284,
      "learning_rate": 7.564454096337659e-08,
      "loss": 0.377,
      "step": 913
    },
    {
      "epoch": 1.8633540372670807,
      "grad_norm": 0.1780468225479126,
      "learning_rate": 7.282384364610207e-08,
      "loss": 0.3827,
      "step": 914
    },
    {
      "epoch": 1.865424430641822,
      "grad_norm": 0.2630448341369629,
      "learning_rate": 7.005635490767216e-08,
      "loss": 0.4095,
      "step": 915
    },
    {
      "epoch": 1.8674948240165632,
      "grad_norm": 0.15141789615154266,
      "learning_rate": 6.734210463420099e-08,
      "loss": 0.1393,
      "step": 916
    },
    {
      "epoch": 1.8695652173913042,
      "grad_norm": 0.20141035318374634,
      "learning_rate": 6.46811221368776e-08,
      "loss": 0.3595,
      "step": 917
    },
    {
      "epoch": 1.8716356107660457,
      "grad_norm": 0.2934320867061615,
      "learning_rate": 6.207343615165562e-08,
      "loss": 0.3961,
      "step": 918
    },
    {
      "epoch": 1.8737060041407867,
      "grad_norm": 0.2153390347957611,
      "learning_rate": 5.951907483893626e-08,
      "loss": 0.2565,
      "step": 919
    },
    {
      "epoch": 1.875776397515528,
      "grad_norm": 0.33150872588157654,
      "learning_rate": 5.701806578327029e-08,
      "loss": 0.392,
      "step": 920
    },
    {
      "epoch": 1.8778467908902692,
      "grad_norm": 0.26579582691192627,
      "learning_rate": 5.4570435993055446e-08,
      "loss": 0.2612,
      "step": 921
    },
    {
      "epoch": 1.8799171842650102,
      "grad_norm": 0.1782764196395874,
      "learning_rate": 5.21762119002478e-08,
      "loss": 0.1443,
      "step": 922
    },
    {
      "epoch": 1.8819875776397517,
      "grad_norm": 0.347116619348526,
      "learning_rate": 4.9835419360076406e-08,
      "loss": 0.3098,
      "step": 923
    },
    {
      "epoch": 1.8840579710144927,
      "grad_norm": 0.20360776782035828,
      "learning_rate": 4.7548083650759134e-08,
      "loss": 0.2233,
      "step": 924
    },
    {
      "epoch": 1.886128364389234,
      "grad_norm": 0.18242114782333374,
      "learning_rate": 4.531422947323838e-08,
      "loss": 0.247,
      "step": 925
    },
    {
      "epoch": 1.8881987577639752,
      "grad_norm": 0.25464263558387756,
      "learning_rate": 4.31338809509052e-08,
      "loss": 0.3339,
      "step": 926
    },
    {
      "epoch": 1.8902691511387164,
      "grad_norm": 0.1700025200843811,
      "learning_rate": 4.10070616293462e-08,
      "loss": 0.2131,
      "step": 927
    },
    {
      "epoch": 1.8923395445134576,
      "grad_norm": 0.14673881232738495,
      "learning_rate": 3.8933794476083143e-08,
      "loss": 0.153,
      "step": 928
    },
    {
      "epoch": 1.8944099378881987,
      "grad_norm": 0.22516903281211853,
      "learning_rate": 3.691410188033151e-08,
      "loss": 0.251,
      "step": 929
    },
    {
      "epoch": 1.89648033126294,
      "grad_norm": 0.16170381009578705,
      "learning_rate": 3.494800565275125e-08,
      "loss": 0.2359,
      "step": 930
    },
    {
      "epoch": 1.8985507246376812,
      "grad_norm": 0.20099857449531555,
      "learning_rate": 3.303552702521806e-08,
      "loss": 0.3992,
      "step": 931
    },
    {
      "epoch": 1.9006211180124224,
      "grad_norm": 0.22314739227294922,
      "learning_rate": 3.1176686650589147e-08,
      "loss": 0.2455,
      "step": 932
    },
    {
      "epoch": 1.9026915113871636,
      "grad_norm": 0.236553356051445,
      "learning_rate": 2.93715046024845e-08,
      "loss": 0.2531,
      "step": 933
    },
    {
      "epoch": 1.9047619047619047,
      "grad_norm": 0.21454867720603943,
      "learning_rate": 2.7620000375064848e-08,
      "loss": 0.275,
      "step": 934
    },
    {
      "epoch": 1.9068322981366461,
      "grad_norm": 0.24925491213798523,
      "learning_rate": 2.592219288282627e-08,
      "loss": 0.3424,
      "step": 935
    },
    {
      "epoch": 1.9089026915113871,
      "grad_norm": 0.21076105535030365,
      "learning_rate": 2.4278100460393138e-08,
      "loss": 0.3642,
      "step": 936
    },
    {
      "epoch": 1.9109730848861284,
      "grad_norm": 0.1848498284816742,
      "learning_rate": 2.2687740862320506e-08,
      "loss": 0.2639,
      "step": 937
    },
    {
      "epoch": 1.9130434782608696,
      "grad_norm": 0.21795926988124847,
      "learning_rate": 2.115113126290258e-08,
      "loss": 0.4229,
      "step": 938
    },
    {
      "epoch": 1.9151138716356106,
      "grad_norm": 0.2156882882118225,
      "learning_rate": 1.966828825598843e-08,
      "loss": 0.2438,
      "step": 939
    },
    {
      "epoch": 1.917184265010352,
      "grad_norm": 0.24736322462558746,
      "learning_rate": 1.8239227854799368e-08,
      "loss": 0.4286,
      "step": 940
    },
    {
      "epoch": 1.9192546583850931,
      "grad_norm": 0.2588346004486084,
      "learning_rate": 1.686396549176128e-08,
      "loss": 0.2705,
      "step": 941
    },
    {
      "epoch": 1.9213250517598344,
      "grad_norm": 0.19903713464736938,
      "learning_rate": 1.554251601833201e-08,
      "loss": 0.1754,
      "step": 942
    },
    {
      "epoch": 1.9233954451345756,
      "grad_norm": 0.21781648695468903,
      "learning_rate": 1.4274893704845916e-08,
      "loss": 0.2451,
      "step": 943
    },
    {
      "epoch": 1.9254658385093166,
      "grad_norm": 0.23910443484783173,
      "learning_rate": 1.3061112240357887e-08,
      "loss": 0.3731,
      "step": 944
    },
    {
      "epoch": 1.927536231884058,
      "grad_norm": 0.3607318699359894,
      "learning_rate": 1.1901184732493465e-08,
      "loss": 0.3392,
      "step": 945
    },
    {
      "epoch": 1.9296066252587991,
      "grad_norm": 0.18589964509010315,
      "learning_rate": 1.0795123707312283e-08,
      "loss": 0.3498,
      "step": 946
    },
    {
      "epoch": 1.9316770186335404,
      "grad_norm": 0.20047861337661743,
      "learning_rate": 9.742941109166515e-09,
      "loss": 0.3472,
      "step": 947
    },
    {
      "epoch": 1.9337474120082816,
      "grad_norm": 0.1777239441871643,
      "learning_rate": 8.744648300578196e-09,
      "loss": 0.3169,
      "step": 948
    },
    {
      "epoch": 1.9358178053830226,
      "grad_norm": 0.22908566892147064,
      "learning_rate": 7.80025606211099e-09,
      "loss": 0.3575,
      "step": 949
    },
    {
      "epoch": 1.937888198757764,
      "grad_norm": 0.20297054946422577,
      "learning_rate": 6.9097745922580564e-09,
      "loss": 0.3494,
      "step": 950
    },
    {
      "epoch": 1.939958592132505,
      "grad_norm": 0.25517526268959045,
      "learning_rate": 6.073213507328812e-09,
      "loss": 0.2926,
      "step": 951
    },
    {
      "epoch": 1.9420289855072463,
      "grad_norm": 0.21190893650054932,
      "learning_rate": 5.2905818413478975e-09,
      "loss": 0.309,
      "step": 952
    },
    {
      "epoch": 1.9440993788819876,
      "grad_norm": 0.22347129881381989,
      "learning_rate": 4.5618880459541435e-09,
      "loss": 0.2969,
      "step": 953
    },
    {
      "epoch": 1.9461697722567288,
      "grad_norm": 0.23573170602321625,
      "learning_rate": 3.887139990313427e-09,
      "loss": 0.2966,
      "step": 954
    },
    {
      "epoch": 1.94824016563147,
      "grad_norm": 0.18915671110153198,
      "learning_rate": 3.2663449610298435e-09,
      "loss": 0.2483,
      "step": 955
    },
    {
      "epoch": 1.950310559006211,
      "grad_norm": 0.2534366250038147,
      "learning_rate": 2.699509662069666e-09,
      "loss": 0.2978,
      "step": 956
    },
    {
      "epoch": 1.9523809523809523,
      "grad_norm": 0.17050714790821075,
      "learning_rate": 2.1866402146875077e-09,
      "loss": 0.2408,
      "step": 957
    },
    {
      "epoch": 1.9544513457556936,
      "grad_norm": 0.2743813097476959,
      "learning_rate": 1.7277421573608234e-09,
      "loss": 0.3852,
      "step": 958
    },
    {
      "epoch": 1.9565217391304348,
      "grad_norm": 0.20148064196109772,
      "learning_rate": 1.3228204457299555e-09,
      "loss": 0.2587,
      "step": 959
    },
    {
      "epoch": 1.958592132505176,
      "grad_norm": 0.2194848358631134,
      "learning_rate": 9.71879452545399e-10,
      "loss": 0.2841,
      "step": 960
    },
    {
      "epoch": 1.960662525879917,
      "grad_norm": 0.2698388993740082,
      "learning_rate": 6.749229676183965e-10,
      "loss": 0.4823,
      "step": 961
    },
    {
      "epoch": 1.9627329192546585,
      "grad_norm": 0.23014101386070251,
      "learning_rate": 4.3195419778319095e-10,
      "loss": 0.3208,
      "step": 962
    },
    {
      "epoch": 1.9648033126293996,
      "grad_norm": 0.16867674887180328,
      "learning_rate": 2.429757668587218e-10,
      "loss": 0.1936,
      "step": 963
    },
    {
      "epoch": 1.9668737060041408,
      "grad_norm": 0.18036042153835297,
      "learning_rate": 1.0798971562364647e-10,
      "loss": 0.2331,
      "step": 964
    },
    {
      "epoch": 1.968944099378882,
      "grad_norm": 0.1979045867919922,
      "learning_rate": 2.6997501792469515e-11,
      "loss": 0.286,
      "step": 965
    },
    {
      "epoch": 1.971014492753623,
      "grad_norm": 0.1963033676147461,
      "learning_rate": 0.0,
      "loss": 0.2669,
      "step": 966
    }
  ],
  "logging_steps": 1,
  "max_steps": 966,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 242,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.343597735784219e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}