{
  "best_metric": 1.8720966577529907,
  "best_model_checkpoint": "miner_id_24/checkpoint-450",
  "epoch": 0.29397354238118567,
  "eval_steps": 150,
  "global_step": 450,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006532745386248571,
      "grad_norm": 4.493376731872559,
      "learning_rate": 5e-06,
      "loss": 1.7869,
      "step": 1
    },
    {
      "epoch": 0.0006532745386248571,
      "eval_loss": 2.817268133163452,
      "eval_runtime": 49.2307,
      "eval_samples_per_second": 52.366,
      "eval_steps_per_second": 13.102,
      "step": 1
    },
    {
      "epoch": 0.0013065490772497142,
      "grad_norm": 4.3894362449646,
      "learning_rate": 1e-05,
      "loss": 1.6956,
      "step": 2
    },
    {
      "epoch": 0.0019598236158745713,
      "grad_norm": 3.021122932434082,
      "learning_rate": 1.5e-05,
      "loss": 2.0092,
      "step": 3
    },
    {
      "epoch": 0.0026130981544994283,
      "grad_norm": 1.7732617855072021,
      "learning_rate": 2e-05,
      "loss": 1.8001,
      "step": 4
    },
    {
      "epoch": 0.0032663726931242854,
      "grad_norm": 1.1165502071380615,
      "learning_rate": 2.5e-05,
      "loss": 1.7495,
      "step": 5
    },
    {
      "epoch": 0.0039196472317491425,
      "grad_norm": 1.0437995195388794,
      "learning_rate": 3e-05,
      "loss": 1.7746,
      "step": 6
    },
    {
      "epoch": 0.004572921770374,
      "grad_norm": 1.216927170753479,
      "learning_rate": 3.5e-05,
      "loss": 2.0735,
      "step": 7
    },
    {
      "epoch": 0.005226196308998857,
      "grad_norm": 1.1597315073013306,
      "learning_rate": 4e-05,
      "loss": 1.6368,
      "step": 8
    },
    {
      "epoch": 0.005879470847623714,
      "grad_norm": 1.0916095972061157,
      "learning_rate": 4.5e-05,
      "loss": 1.8234,
      "step": 9
    },
    {
      "epoch": 0.006532745386248571,
      "grad_norm": 1.0554530620574951,
      "learning_rate": 5e-05,
      "loss": 1.9011,
      "step": 10
    },
    {
      "epoch": 0.007186019924873428,
      "grad_norm": 1.123693823814392,
      "learning_rate": 5.500000000000001e-05,
      "loss": 1.9221,
      "step": 11
    },
    {
      "epoch": 0.007839294463498285,
      "grad_norm": 1.0953428745269775,
      "learning_rate": 6e-05,
      "loss": 1.936,
      "step": 12
    },
    {
      "epoch": 0.008492569002123142,
      "grad_norm": 1.094260811805725,
      "learning_rate": 6.500000000000001e-05,
      "loss": 1.8213,
      "step": 13
    },
    {
      "epoch": 0.009145843540748,
      "grad_norm": 1.2135497331619263,
      "learning_rate": 7e-05,
      "loss": 1.8456,
      "step": 14
    },
    {
      "epoch": 0.009799118079372856,
      "grad_norm": 1.2519779205322266,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.8708,
      "step": 15
    },
    {
      "epoch": 0.010452392617997713,
      "grad_norm": 1.2671680450439453,
      "learning_rate": 8e-05,
      "loss": 1.9304,
      "step": 16
    },
    {
      "epoch": 0.01110566715662257,
      "grad_norm": 1.2108627557754517,
      "learning_rate": 8.5e-05,
      "loss": 1.9269,
      "step": 17
    },
    {
      "epoch": 0.011758941695247428,
      "grad_norm": 1.2395938634872437,
      "learning_rate": 9e-05,
      "loss": 1.8435,
      "step": 18
    },
    {
      "epoch": 0.012412216233872285,
      "grad_norm": 1.314139723777771,
      "learning_rate": 9.5e-05,
      "loss": 1.875,
      "step": 19
    },
    {
      "epoch": 0.013065490772497142,
      "grad_norm": 1.375675082206726,
      "learning_rate": 0.0001,
      "loss": 1.8509,
      "step": 20
    },
    {
      "epoch": 0.013718765311121999,
      "grad_norm": 1.3801031112670898,
      "learning_rate": 9.999866555428618e-05,
      "loss": 1.9755,
      "step": 21
    },
    {
      "epoch": 0.014372039849746856,
      "grad_norm": 1.3882397413253784,
      "learning_rate": 9.999466228837451e-05,
      "loss": 1.8054,
      "step": 22
    },
    {
      "epoch": 0.015025314388371713,
      "grad_norm": 1.3915802240371704,
      "learning_rate": 9.998799041595064e-05,
      "loss": 1.882,
      "step": 23
    },
    {
      "epoch": 0.01567858892699657,
      "grad_norm": 1.4975143671035767,
      "learning_rate": 9.997865029314463e-05,
      "loss": 1.9664,
      "step": 24
    },
    {
      "epoch": 0.01633186346562143,
      "grad_norm": 1.480427861213684,
      "learning_rate": 9.996664241851197e-05,
      "loss": 1.7769,
      "step": 25
    },
    {
      "epoch": 0.016985138004246284,
      "grad_norm": 1.5946893692016602,
      "learning_rate": 9.995196743300692e-05,
      "loss": 1.8593,
      "step": 26
    },
    {
      "epoch": 0.017638412542871143,
      "grad_norm": 1.5638861656188965,
      "learning_rate": 9.993462611994832e-05,
      "loss": 1.8672,
      "step": 27
    },
    {
      "epoch": 0.018291687081496,
      "grad_norm": 1.680821180343628,
      "learning_rate": 9.991461940497786e-05,
      "loss": 1.7121,
      "step": 28
    },
    {
      "epoch": 0.018944961620120857,
      "grad_norm": 1.7322921752929688,
      "learning_rate": 9.989194835601048e-05,
      "loss": 1.7184,
      "step": 29
    },
    {
      "epoch": 0.019598236158745713,
      "grad_norm": 2.1110496520996094,
      "learning_rate": 9.986661418317759e-05,
      "loss": 2.0482,
      "step": 30
    },
    {
      "epoch": 0.02025151069737057,
      "grad_norm": 1.9458863735198975,
      "learning_rate": 9.983861823876231e-05,
      "loss": 1.7309,
      "step": 31
    },
    {
      "epoch": 0.020904785235995427,
      "grad_norm": 2.277191638946533,
      "learning_rate": 9.980796201712734e-05,
      "loss": 1.9828,
      "step": 32
    },
    {
      "epoch": 0.021558059774620286,
      "grad_norm": 2.6395316123962402,
      "learning_rate": 9.977464715463524e-05,
      "loss": 1.9904,
      "step": 33
    },
    {
      "epoch": 0.02221133431324514,
      "grad_norm": 2.5731561183929443,
      "learning_rate": 9.973867542956104e-05,
      "loss": 2.0996,
      "step": 34
    },
    {
      "epoch": 0.02286460885187,
      "grad_norm": 3.0783441066741943,
      "learning_rate": 9.97000487619973e-05,
      "loss": 2.4415,
      "step": 35
    },
    {
      "epoch": 0.023517883390494855,
      "grad_norm": 2.9236958026885986,
      "learning_rate": 9.965876921375165e-05,
      "loss": 2.0984,
      "step": 36
    },
    {
      "epoch": 0.024171157929119714,
      "grad_norm": 3.1245806217193604,
      "learning_rate": 9.961483898823678e-05,
      "loss": 2.1028,
      "step": 37
    },
    {
      "epoch": 0.02482443246774457,
      "grad_norm": 3.4645023345947266,
      "learning_rate": 9.956826043035268e-05,
      "loss": 2.1652,
      "step": 38
    },
    {
      "epoch": 0.025477707006369428,
      "grad_norm": 3.9145984649658203,
      "learning_rate": 9.951903602636166e-05,
      "loss": 2.159,
      "step": 39
    },
    {
      "epoch": 0.026130981544994283,
      "grad_norm": 3.910919189453125,
      "learning_rate": 9.946716840375551e-05,
      "loss": 2.2389,
      "step": 40
    },
    {
      "epoch": 0.026784256083619142,
      "grad_norm": 4.544792175292969,
      "learning_rate": 9.94126603311153e-05,
      "loss": 2.2694,
      "step": 41
    },
    {
      "epoch": 0.027437530622243998,
      "grad_norm": 4.551459789276123,
      "learning_rate": 9.935551471796358e-05,
      "loss": 2.1626,
      "step": 42
    },
    {
      "epoch": 0.028090805160868856,
      "grad_norm": 4.505721569061279,
      "learning_rate": 9.92957346146091e-05,
      "loss": 1.9922,
      "step": 43
    },
    {
      "epoch": 0.02874407969949371,
      "grad_norm": 4.990652561187744,
      "learning_rate": 9.923332321198395e-05,
      "loss": 2.3784,
      "step": 44
    },
    {
      "epoch": 0.02939735423811857,
      "grad_norm": 4.7291083335876465,
      "learning_rate": 9.916828384147331e-05,
      "loss": 2.2527,
      "step": 45
    },
    {
      "epoch": 0.030050628776743426,
      "grad_norm": 5.525903224945068,
      "learning_rate": 9.910061997473752e-05,
      "loss": 2.3289,
      "step": 46
    },
    {
      "epoch": 0.030703903315368285,
      "grad_norm": 6.138942718505859,
      "learning_rate": 9.903033522352687e-05,
      "loss": 2.5609,
      "step": 47
    },
    {
      "epoch": 0.03135717785399314,
      "grad_norm": 6.956801414489746,
      "learning_rate": 9.895743333948874e-05,
      "loss": 2.6955,
      "step": 48
    },
    {
      "epoch": 0.032010452392617995,
      "grad_norm": 7.849874496459961,
      "learning_rate": 9.888191821396744e-05,
      "loss": 2.8321,
      "step": 49
    },
    {
      "epoch": 0.03266372693124286,
      "grad_norm": 10.87897777557373,
      "learning_rate": 9.880379387779637e-05,
      "loss": 2.5853,
      "step": 50
    },
    {
      "epoch": 0.03331700146986771,
      "grad_norm": 4.779987812042236,
      "learning_rate": 9.872306450108292e-05,
      "loss": 1.8573,
      "step": 51
    },
    {
      "epoch": 0.03397027600849257,
      "grad_norm": 4.803715229034424,
      "learning_rate": 9.863973439298597e-05,
      "loss": 2.2457,
      "step": 52
    },
    {
      "epoch": 0.034623550547117424,
      "grad_norm": 3.4397003650665283,
      "learning_rate": 9.855380800148572e-05,
      "loss": 2.187,
      "step": 53
    },
    {
      "epoch": 0.035276825085742286,
      "grad_norm": 1.9696916341781616,
      "learning_rate": 9.846528991314639e-05,
      "loss": 1.9824,
      "step": 54
    },
    {
      "epoch": 0.03593009962436714,
      "grad_norm": 1.1352851390838623,
      "learning_rate": 9.837418485287127e-05,
      "loss": 1.8768,
      "step": 55
    },
    {
      "epoch": 0.036583374162992,
      "grad_norm": 1.01988685131073,
      "learning_rate": 9.828049768365068e-05,
      "loss": 1.8008,
      "step": 56
    },
    {
      "epoch": 0.03723664870161685,
      "grad_norm": 1.049869179725647,
      "learning_rate": 9.818423340630228e-05,
      "loss": 1.7975,
      "step": 57
    },
    {
      "epoch": 0.037889923240241714,
      "grad_norm": 1.0288408994674683,
      "learning_rate": 9.808539715920414e-05,
      "loss": 1.8469,
      "step": 58
    },
    {
      "epoch": 0.03854319777886657,
      "grad_norm": 0.9697356224060059,
      "learning_rate": 9.798399421802056e-05,
      "loss": 1.7539,
      "step": 59
    },
    {
      "epoch": 0.039196472317491425,
      "grad_norm": 1.0012383460998535,
      "learning_rate": 9.78800299954203e-05,
      "loss": 1.7325,
      "step": 60
    },
    {
      "epoch": 0.03984974685611628,
      "grad_norm": 1.0223060846328735,
      "learning_rate": 9.777351004078783e-05,
      "loss": 1.8353,
      "step": 61
    },
    {
      "epoch": 0.04050302139474114,
      "grad_norm": 1.005021572113037,
      "learning_rate": 9.766444003992703e-05,
      "loss": 1.8061,
      "step": 62
    },
    {
      "epoch": 0.041156295933366,
      "grad_norm": 1.061297059059143,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.8003,
      "step": 63
    },
    {
      "epoch": 0.04180957047199085,
      "grad_norm": 1.02419912815094,
      "learning_rate": 9.743867332300478e-05,
      "loss": 1.7351,
      "step": 64
    },
    {
      "epoch": 0.04246284501061571,
      "grad_norm": 1.0611597299575806,
      "learning_rate": 9.732198865788047e-05,
      "loss": 1.763,
      "step": 65
    },
    {
      "epoch": 0.04311611954924057,
      "grad_norm": 1.1353827714920044,
      "learning_rate": 9.72027780477588e-05,
      "loss": 1.6926,
      "step": 66
    },
    {
      "epoch": 0.043769394087865426,
      "grad_norm": 1.1362850666046143,
      "learning_rate": 9.708104785584323e-05,
      "loss": 1.6747,
      "step": 67
    },
    {
      "epoch": 0.04442266862649028,
      "grad_norm": 1.2011291980743408,
      "learning_rate": 9.695680457982713e-05,
      "loss": 1.7898,
      "step": 68
    },
    {
      "epoch": 0.04507594316511514,
      "grad_norm": 1.2961357831954956,
      "learning_rate": 9.683005485154677e-05,
      "loss": 1.9112,
      "step": 69
    },
    {
      "epoch": 0.04572921770374,
      "grad_norm": 1.229396939277649,
      "learning_rate": 9.67008054366274e-05,
      "loss": 2.0017,
      "step": 70
    },
    {
      "epoch": 0.046382492242364855,
      "grad_norm": 1.2769562005996704,
      "learning_rate": 9.656906323412217e-05,
      "loss": 1.7742,
      "step": 71
    },
    {
      "epoch": 0.04703576678098971,
      "grad_norm": 1.4329209327697754,
      "learning_rate": 9.643483527614372e-05,
      "loss": 2.0372,
      "step": 72
    },
    {
      "epoch": 0.047689041319614565,
      "grad_norm": 1.3083312511444092,
      "learning_rate": 9.629812872748901e-05,
      "loss": 1.8825,
      "step": 73
    },
    {
      "epoch": 0.04834231585823943,
      "grad_norm": 1.3528681993484497,
      "learning_rate": 9.615895088525677e-05,
      "loss": 1.7573,
      "step": 74
    },
    {
      "epoch": 0.04899559039686428,
      "grad_norm": 1.4028435945510864,
      "learning_rate": 9.601730917845797e-05,
      "loss": 1.8285,
      "step": 75
    },
    {
      "epoch": 0.04964886493548914,
      "grad_norm": 1.5294512510299683,
      "learning_rate": 9.587321116761938e-05,
      "loss": 1.7262,
      "step": 76
    },
    {
      "epoch": 0.050302139474113994,
      "grad_norm": 1.6173073053359985,
      "learning_rate": 9.57266645443799e-05,
      "loss": 1.7705,
      "step": 77
    },
    {
      "epoch": 0.050955414012738856,
      "grad_norm": 1.8100759983062744,
      "learning_rate": 9.557767713108009e-05,
      "loss": 1.9303,
      "step": 78
    },
    {
      "epoch": 0.05160868855136371,
      "grad_norm": 1.8436578512191772,
      "learning_rate": 9.542625688034449e-05,
      "loss": 2.0698,
      "step": 79
    },
    {
      "epoch": 0.05226196308998857,
      "grad_norm": 1.8567721843719482,
      "learning_rate": 9.527241187465734e-05,
      "loss": 1.9051,
      "step": 80
    },
    {
      "epoch": 0.05291523762861342,
      "grad_norm": 1.9776356220245361,
      "learning_rate": 9.511615032593096e-05,
      "loss": 1.665,
      "step": 81
    },
    {
      "epoch": 0.053568512167238284,
      "grad_norm": 2.2138235569000244,
      "learning_rate": 9.49574805750675e-05,
      "loss": 1.9731,
      "step": 82
    },
    {
      "epoch": 0.05422178670586314,
      "grad_norm": 2.2105331420898438,
      "learning_rate": 9.479641109151373e-05,
      "loss": 1.8617,
      "step": 83
    },
    {
      "epoch": 0.054875061244487995,
      "grad_norm": 2.3538990020751953,
      "learning_rate": 9.463295047280891e-05,
      "loss": 1.8796,
      "step": 84
    },
    {
      "epoch": 0.05552833578311285,
      "grad_norm": 2.7223575115203857,
      "learning_rate": 9.446710744412595e-05,
      "loss": 2.1008,
      "step": 85
    },
    {
      "epoch": 0.05618161032173771,
      "grad_norm": 2.535693645477295,
      "learning_rate": 9.429889085780557e-05,
      "loss": 1.9457,
      "step": 86
    },
    {
      "epoch": 0.05683488486036257,
      "grad_norm": 2.931142807006836,
      "learning_rate": 9.41283096928839e-05,
      "loss": 2.1196,
      "step": 87
    },
    {
      "epoch": 0.05748815939898742,
      "grad_norm": 3.281611204147339,
      "learning_rate": 9.395537305461311e-05,
      "loss": 2.27,
      "step": 88
    },
    {
      "epoch": 0.05814143393761228,
      "grad_norm": 3.470123529434204,
      "learning_rate": 9.378009017397542e-05,
      "loss": 2.2742,
      "step": 89
    },
    {
      "epoch": 0.05879470847623714,
      "grad_norm": 3.7174768447875977,
      "learning_rate": 9.360247040719039e-05,
      "loss": 1.9928,
      "step": 90
    },
    {
      "epoch": 0.059447983014861996,
      "grad_norm": 3.878044843673706,
      "learning_rate": 9.342252323521545e-05,
      "loss": 2.306,
      "step": 91
    },
    {
      "epoch": 0.06010125755348685,
      "grad_norm": 4.135204315185547,
      "learning_rate": 9.324025826323994e-05,
      "loss": 2.4951,
      "step": 92
    },
    {
      "epoch": 0.06075453209211171,
      "grad_norm": 4.10722541809082,
      "learning_rate": 9.305568522017227e-05,
      "loss": 2.0659,
      "step": 93
    },
    {
      "epoch": 0.06140780663073657,
      "grad_norm": 5.294311046600342,
      "learning_rate": 9.286881395812066e-05,
      "loss": 2.1877,
      "step": 94
    },
    {
      "epoch": 0.062061081169361425,
      "grad_norm": 4.588815689086914,
      "learning_rate": 9.267965445186733e-05,
      "loss": 2.6443,
      "step": 95
    },
    {
      "epoch": 0.06271435570798628,
      "grad_norm": 4.880858421325684,
      "learning_rate": 9.248821679833596e-05,
      "loss": 2.0848,
      "step": 96
    },
    {
      "epoch": 0.06336763024661114,
      "grad_norm": 4.187829494476318,
      "learning_rate": 9.229451121605279e-05,
      "loss": 1.4447,
      "step": 97
    },
    {
      "epoch": 0.06402090478523599,
      "grad_norm": 6.281601905822754,
      "learning_rate": 9.209854804460121e-05,
      "loss": 2.4293,
      "step": 98
    },
    {
      "epoch": 0.06467417932386085,
      "grad_norm": 7.709883689880371,
      "learning_rate": 9.190033774406977e-05,
      "loss": 2.5011,
      "step": 99
    },
    {
      "epoch": 0.06532745386248572,
      "grad_norm": 8.83784008026123,
      "learning_rate": 9.16998908944939e-05,
      "loss": 2.4641,
      "step": 100
    },
    {
      "epoch": 0.06598072840111056,
      "grad_norm": 4.936941146850586,
      "learning_rate": 9.149721819529119e-05,
      "loss": 2.1716,
      "step": 101
    },
    {
      "epoch": 0.06663400293973543,
      "grad_norm": 4.244714260101318,
      "learning_rate": 9.129233046469022e-05,
      "loss": 2.2284,
      "step": 102
    },
    {
      "epoch": 0.06728727747836027,
      "grad_norm": 3.146141767501831,
      "learning_rate": 9.108523863915314e-05,
      "loss": 2.0688,
      "step": 103
    },
    {
      "epoch": 0.06794055201698514,
      "grad_norm": 1.925838828086853,
      "learning_rate": 9.087595377279192e-05,
      "loss": 2.0631,
      "step": 104
    },
    {
      "epoch": 0.06859382655561,
      "grad_norm": 1.0202608108520508,
      "learning_rate": 9.066448703677828e-05,
      "loss": 1.7997,
      "step": 105
    },
    {
      "epoch": 0.06924710109423485,
      "grad_norm": 0.9885883927345276,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.83,
      "step": 106
    },
    {
      "epoch": 0.06990037563285971,
      "grad_norm": 0.8777309060096741,
      "learning_rate": 9.023505322219536e-05,
      "loss": 1.6259,
      "step": 107
    },
    {
      "epoch": 0.07055365017148457,
      "grad_norm": 0.9906489849090576,
      "learning_rate": 9.001710906587064e-05,
      "loss": 1.6581,
      "step": 108
    },
    {
      "epoch": 0.07120692471010942,
      "grad_norm": 1.1678723096847534,
      "learning_rate": 8.9797028883159e-05,
      "loss": 1.8921,
      "step": 109
    },
    {
      "epoch": 0.07186019924873428,
      "grad_norm": 1.0283783674240112,
      "learning_rate": 8.957482442146272e-05,
      "loss": 1.84,
      "step": 110
    },
    {
      "epoch": 0.07251347378735913,
      "grad_norm": 1.1051928997039795,
      "learning_rate": 8.935050754157344e-05,
      "loss": 1.8327,
      "step": 111
    },
    {
      "epoch": 0.073166748325984,
      "grad_norm": 1.1641846895217896,
      "learning_rate": 8.912409021703913e-05,
      "loss": 1.9533,
      "step": 112
    },
    {
      "epoch": 0.07382002286460886,
      "grad_norm": 1.0035812854766846,
      "learning_rate": 8.889558453352492e-05,
      "loss": 1.75,
      "step": 113
    },
    {
      "epoch": 0.0744732974032337,
      "grad_norm": 1.176912546157837,
      "learning_rate": 8.866500268816803e-05,
      "loss": 1.925,
      "step": 114
    },
    {
      "epoch": 0.07512657194185857,
      "grad_norm": 1.0900567770004272,
      "learning_rate": 8.84323569889266e-05,
      "loss": 1.7295,
      "step": 115
    },
    {
      "epoch": 0.07577984648048343,
      "grad_norm": 1.0858813524246216,
      "learning_rate": 8.819765985392296e-05,
      "loss": 1.8185,
      "step": 116
    },
    {
      "epoch": 0.07643312101910828,
      "grad_norm": 1.1561424732208252,
      "learning_rate": 8.79609238107805e-05,
      "loss": 1.7551,
      "step": 117
    },
    {
      "epoch": 0.07708639555773314,
      "grad_norm": 1.191009759902954,
      "learning_rate": 8.772216149595513e-05,
      "loss": 1.7119,
      "step": 118
    },
    {
      "epoch": 0.07773967009635799,
      "grad_norm": 1.2849817276000977,
      "learning_rate": 8.748138565406081e-05,
      "loss": 1.8629,
      "step": 119
    },
    {
      "epoch": 0.07839294463498285,
      "grad_norm": 1.2508069276809692,
      "learning_rate": 8.72386091371891e-05,
      "loss": 1.7054,
      "step": 120
    },
    {
      "epoch": 0.07904621917360771,
      "grad_norm": 1.4011170864105225,
      "learning_rate": 8.699384490422331e-05,
      "loss": 1.7732,
      "step": 121
    },
    {
      "epoch": 0.07969949371223256,
      "grad_norm": 1.7134177684783936,
      "learning_rate": 8.674710602014671e-05,
      "loss": 1.9894,
      "step": 122
    },
    {
      "epoch": 0.08035276825085742,
      "grad_norm": 1.4881147146224976,
      "learning_rate": 8.649840565534513e-05,
      "loss": 2.1368,
      "step": 123
    },
    {
      "epoch": 0.08100604278948229,
      "grad_norm": 1.4530311822891235,
      "learning_rate": 8.624775708490402e-05,
      "loss": 1.8706,
      "step": 124
    },
    {
      "epoch": 0.08165931732810713,
      "grad_norm": 1.6658281087875366,
      "learning_rate": 8.59951736878998e-05,
      "loss": 1.9829,
      "step": 125
    },
    {
      "epoch": 0.082312591866732,
      "grad_norm": 1.725263237953186,
      "learning_rate": 8.574066894668573e-05,
      "loss": 2.0811,
      "step": 126
    },
    {
      "epoch": 0.08296586640535684,
      "grad_norm": 1.6992791891098022,
      "learning_rate": 8.548425644617224e-05,
      "loss": 1.9159,
      "step": 127
    },
    {
      "epoch": 0.0836191409439817,
      "grad_norm": 1.7884573936462402,
      "learning_rate": 8.522594987310184e-05,
      "loss": 1.974,
      "step": 128
    },
    {
      "epoch": 0.08427241548260657,
      "grad_norm": 1.9292463064193726,
      "learning_rate": 8.49657630153185e-05,
      "loss": 1.9068,
      "step": 129
    },
    {
      "epoch": 0.08492569002123142,
      "grad_norm": 2.012984275817871,
      "learning_rate": 8.47037097610317e-05,
      "loss": 2.1171,
      "step": 130
    },
    {
      "epoch": 0.08557896455985628,
      "grad_norm": 2.215426445007324,
      "learning_rate": 8.443980409807512e-05,
      "loss": 2.0225,
      "step": 131
    },
    {
      "epoch": 0.08623223909848114,
      "grad_norm": 2.2149457931518555,
      "learning_rate": 8.417406011315998e-05,
      "loss": 1.7872,
      "step": 132
    },
    {
      "epoch": 0.08688551363710599,
      "grad_norm": 2.5462419986724854,
      "learning_rate": 8.390649199112315e-05,
      "loss": 1.9432,
      "step": 133
    },
    {
      "epoch": 0.08753878817573085,
      "grad_norm": 2.7230594158172607,
      "learning_rate": 8.363711401417e-05,
      "loss": 2.0397,
      "step": 134
    },
    {
      "epoch": 0.0881920627143557,
      "grad_norm": 2.6147336959838867,
      "learning_rate": 8.336594056111197e-05,
      "loss": 1.8311,
      "step": 135
    },
    {
      "epoch": 0.08884533725298056,
      "grad_norm": 3.2250797748565674,
      "learning_rate": 8.309298610659916e-05,
      "loss": 2.2002,
      "step": 136
    },
    {
      "epoch": 0.08949861179160543,
      "grad_norm": 3.181053876876831,
      "learning_rate": 8.281826522034764e-05,
      "loss": 1.9919,
      "step": 137
    },
    {
      "epoch": 0.09015188633023027,
      "grad_norm": 3.64414119720459,
      "learning_rate": 8.254179256636179e-05,
      "loss": 2.1375,
      "step": 138
    },
    {
      "epoch": 0.09080516086885514,
      "grad_norm": 3.2475857734680176,
      "learning_rate": 8.226358290215151e-05,
      "loss": 2.0877,
      "step": 139
    },
    {
      "epoch": 0.09145843540748,
      "grad_norm": 3.4159018993377686,
      "learning_rate": 8.198365107794457e-05,
      "loss": 1.7944,
      "step": 140
    },
    {
      "epoch": 0.09211170994610485,
      "grad_norm": 3.555100440979004,
      "learning_rate": 8.17020120358939e-05,
      "loss": 2.1612,
      "step": 141
    },
    {
      "epoch": 0.09276498448472971,
      "grad_norm": 4.028273105621338,
      "learning_rate": 8.141868080927996e-05,
      "loss": 2.1248,
      "step": 142
    },
    {
      "epoch": 0.09341825902335456,
      "grad_norm": 4.388622760772705,
      "learning_rate": 8.113367252170844e-05,
      "loss": 2.3343,
      "step": 143
    },
    {
      "epoch": 0.09407153356197942,
      "grad_norm": 4.28695821762085,
      "learning_rate": 8.084700238630283e-05,
      "loss": 2.2616,
      "step": 144
    },
    {
      "epoch": 0.09472480810060428,
      "grad_norm": 4.270381450653076,
      "learning_rate": 8.055868570489247e-05,
      "loss": 1.9253,
      "step": 145
    },
    {
      "epoch": 0.09537808263922913,
      "grad_norm": 4.905386447906494,
      "learning_rate": 8.026873786719573e-05,
      "loss": 2.1537,
      "step": 146
    },
    {
      "epoch": 0.096031357177854,
      "grad_norm": 4.994656085968018,
      "learning_rate": 7.997717434999861e-05,
      "loss": 2.0804,
      "step": 147
    },
    {
      "epoch": 0.09668463171647886,
      "grad_norm": 4.8317036628723145,
      "learning_rate": 7.968401071632855e-05,
      "loss": 2.1722,
      "step": 148
    },
    {
      "epoch": 0.0973379062551037,
      "grad_norm": 5.793895721435547,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.5444,
      "step": 149
    },
    {
      "epoch": 0.09799118079372857,
      "grad_norm": 7.024384021759033,
      "learning_rate": 7.909294577789766e-05,
      "loss": 2.6155,
      "step": 150
    },
    {
      "epoch": 0.09799118079372857,
      "eval_loss": 2.1448490619659424,
      "eval_runtime": 49.7716,
      "eval_samples_per_second": 51.797,
      "eval_steps_per_second": 12.959,
      "step": 150
    },
    {
      "epoch": 0.09864445533235343,
      "grad_norm": 2.752495050430298,
      "learning_rate": 7.879507602289979e-05,
      "loss": 1.7088,
      "step": 151
    },
    {
      "epoch": 0.09929772987097828,
      "grad_norm": 2.6285979747772217,
      "learning_rate": 7.849566924927082e-05,
      "loss": 2.0209,
      "step": 152
    },
    {
      "epoch": 0.09995100440960314,
      "grad_norm": 2.3913207054138184,
      "learning_rate": 7.819474143869414e-05,
      "loss": 2.0401,
      "step": 153
    },
    {
      "epoch": 0.10060427894822799,
      "grad_norm": 1.4358742237091064,
      "learning_rate": 7.789230865404287e-05,
      "loss": 1.7734,
      "step": 154
    },
    {
      "epoch": 0.10125755348685285,
      "grad_norm": 1.0302163362503052,
      "learning_rate": 7.75883870385223e-05,
      "loss": 1.8005,
      "step": 155
    },
    {
      "epoch": 0.10191082802547771,
      "grad_norm": 0.9365765452384949,
      "learning_rate": 7.728299281480833e-05,
      "loss": 1.775,
      "step": 156
    },
    {
      "epoch": 0.10256410256410256,
      "grad_norm": 0.8915041089057922,
      "learning_rate": 7.697614228418148e-05,
      "loss": 1.8217,
      "step": 157
    },
    {
      "epoch": 0.10321737710272742,
      "grad_norm": 0.9199675917625427,
      "learning_rate": 7.666785182565677e-05,
      "loss": 1.8245,
      "step": 158
    },
    {
      "epoch": 0.10387065164135229,
      "grad_norm": 0.9064541459083557,
      "learning_rate": 7.635813789510941e-05,
      "loss": 1.7981,
      "step": 159
    },
    {
      "epoch": 0.10452392617997713,
      "grad_norm": 0.9522148370742798,
      "learning_rate": 7.604701702439651e-05,
      "loss": 1.7425,
      "step": 160
    },
    {
      "epoch": 0.105177200718602,
      "grad_norm": 1.012233853340149,
      "learning_rate": 7.573450582047457e-05,
      "loss": 1.7662,
      "step": 161
    },
    {
      "epoch": 0.10583047525722684,
      "grad_norm": 1.1444512605667114,
      "learning_rate": 7.542062096451305e-05,
      "loss": 1.923,
      "step": 162
    },
    {
      "epoch": 0.1064837497958517,
      "grad_norm": 0.9591057300567627,
      "learning_rate": 7.510537921100398e-05,
      "loss": 1.6653,
      "step": 163
    },
    {
      "epoch": 0.10713702433447657,
      "grad_norm": 1.0296618938446045,
      "learning_rate": 7.47887973868676e-05,
      "loss": 1.7312,
      "step": 164
    },
    {
      "epoch": 0.10779029887310142,
      "grad_norm": 1.0364526510238647,
      "learning_rate": 7.447089239055428e-05,
      "loss": 1.7524,
      "step": 165
    },
    {
      "epoch": 0.10844357341172628,
      "grad_norm": 1.1254093647003174,
      "learning_rate": 7.41516811911424e-05,
      "loss": 1.933,
      "step": 166
    },
    {
      "epoch": 0.10909684795035114,
      "grad_norm": 1.1164376735687256,
      "learning_rate": 7.383118082743262e-05,
      "loss": 1.8894,
      "step": 167
    },
    {
      "epoch": 0.10975012248897599,
      "grad_norm": 1.215606451034546,
      "learning_rate": 7.350940840703842e-05,
      "loss": 1.9096,
      "step": 168
    },
    {
      "epoch": 0.11040339702760085,
      "grad_norm": 1.1947083473205566,
      "learning_rate": 7.318638110547288e-05,
      "loss": 1.9841,
      "step": 169
    },
    {
      "epoch": 0.1110566715662257,
      "grad_norm": 1.2013548612594604,
      "learning_rate": 7.286211616523193e-05,
      "loss": 1.8947,
      "step": 170
    },
    {
      "epoch": 0.11170994610485056,
      "grad_norm": 1.2469847202301025,
      "learning_rate": 7.253663089487395e-05,
      "loss": 2.0382,
      "step": 171
    },
    {
      "epoch": 0.11236322064347543,
      "grad_norm": 1.2501499652862549,
      "learning_rate": 7.220994266809591e-05,
      "loss": 1.8705,
      "step": 172
    },
    {
      "epoch": 0.11301649518210027,
      "grad_norm": 1.3564845323562622,
      "learning_rate": 7.188206892280594e-05,
      "loss": 1.8065,
      "step": 173
    },
    {
      "epoch": 0.11366976972072514,
      "grad_norm": 1.4328407049179077,
      "learning_rate": 7.155302716019263e-05,
      "loss": 1.998,
      "step": 174
    },
    {
      "epoch": 0.11432304425935,
      "grad_norm": 1.4313324689865112,
      "learning_rate": 7.122283494379076e-05,
      "loss": 1.9056,
      "step": 175
    },
    {
      "epoch": 0.11497631879797485,
      "grad_norm": 1.5474004745483398,
      "learning_rate": 7.089150989854385e-05,
      "loss": 1.7997,
      "step": 176
    },
    {
      "epoch": 0.11562959333659971,
      "grad_norm": 1.593336820602417,
      "learning_rate": 7.055906970986336e-05,
      "loss": 2.0353,
      "step": 177
    },
    {
      "epoch": 0.11628286787522456,
      "grad_norm": 1.4710558652877808,
      "learning_rate": 7.022553212268469e-05,
      "loss": 1.6336,
      "step": 178
    },
    {
      "epoch": 0.11693614241384942,
      "grad_norm": 1.683799386024475,
      "learning_rate": 6.989091494051998e-05,
      "loss": 1.7796,
      "step": 179
    },
    {
      "epoch": 0.11758941695247428,
      "grad_norm": 1.893215537071228,
      "learning_rate": 6.95552360245078e-05,
      "loss": 2.0098,
      "step": 180
    },
    {
      "epoch": 0.11824269149109913,
      "grad_norm": 1.9342585802078247,
      "learning_rate": 6.92185132924598e-05,
      "loss": 1.9887,
      "step": 181
    },
    {
      "epoch": 0.11889596602972399,
      "grad_norm": 2.1239845752716064,
      "learning_rate": 6.888076471790424e-05,
      "loss": 1.9182,
      "step": 182
    },
    {
      "epoch": 0.11954924056834886,
      "grad_norm": 2.2768619060516357,
      "learning_rate": 6.85420083291266e-05,
      "loss": 1.8638,
      "step": 183
    },
    {
      "epoch": 0.1202025151069737,
      "grad_norm": 2.6163580417633057,
      "learning_rate": 6.820226220820732e-05,
      "loss": 2.0315,
      "step": 184
    },
    {
      "epoch": 0.12085578964559857,
      "grad_norm": 2.6591720581054688,
      "learning_rate": 6.786154449005665e-05,
      "loss": 2.3064,
      "step": 185
    },
    {
      "epoch": 0.12150906418422341,
      "grad_norm": 2.8200557231903076,
      "learning_rate": 6.751987336144648e-05,
      "loss": 1.9273,
      "step": 186
    },
    {
      "epoch": 0.12216233872284828,
      "grad_norm": 2.890976905822754,
      "learning_rate": 6.717726706003974e-05,
      "loss": 1.6985,
      "step": 187
    },
    {
      "epoch": 0.12281561326147314,
      "grad_norm": 2.9265048503875732,
      "learning_rate": 6.683374387341687e-05,
      "loss": 1.8323,
      "step": 188
    },
    {
      "epoch": 0.12346888780009799,
      "grad_norm": 3.0600929260253906,
      "learning_rate": 6.648932213809962e-05,
      "loss": 1.8304,
      "step": 189
    },
    {
      "epoch": 0.12412216233872285,
      "grad_norm": 3.547772169113159,
      "learning_rate": 6.614402023857232e-05,
      "loss": 2.0603,
      "step": 190
    },
    {
      "epoch": 0.12477543687734771,
      "grad_norm": 4.024485111236572,
      "learning_rate": 6.579785660630056e-05,
      "loss": 1.9605,
      "step": 191
    },
    {
      "epoch": 0.12542871141597256,
      "grad_norm": 3.935825824737549,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.0046,
      "step": 192
    },
    {
      "epoch": 0.1260819859545974,
      "grad_norm": 3.59587025642395,
      "learning_rate": 6.510301809838689e-05,
      "loss": 1.9499,
      "step": 193
    },
    {
      "epoch": 0.12673526049322228,
      "grad_norm": 3.7973616123199463,
      "learning_rate": 6.475438031171574e-05,
      "loss": 1.6432,
      "step": 194
    },
    {
      "epoch": 0.12738853503184713,
      "grad_norm": 4.452199935913086,
      "learning_rate": 6.440495496826189e-05,
      "loss": 2.2546,
      "step": 195
    },
    {
      "epoch": 0.12804180957047198,
      "grad_norm": 4.514564037322998,
      "learning_rate": 6.405476071959143e-05,
      "loss": 2.1379,
      "step": 196
    },
    {
      "epoch": 0.12869508410909686,
      "grad_norm": 4.9529008865356445,
      "learning_rate": 6.370381625831292e-05,
      "loss": 2.4329,
      "step": 197
    },
    {
      "epoch": 0.1293483586477217,
      "grad_norm": 5.208448886871338,
      "learning_rate": 6.335214031707965e-05,
      "loss": 2.2603,
      "step": 198
    },
    {
      "epoch": 0.13000163318634655,
      "grad_norm": 5.450995922088623,
      "learning_rate": 6.299975166758971e-05,
      "loss": 2.3946,
      "step": 199
    },
    {
      "epoch": 0.13065490772497143,
      "grad_norm": 7.545385837554932,
      "learning_rate": 6.264666911958404e-05,
      "loss": 2.4191,
      "step": 200
    },
    {
      "epoch": 0.13130818226359628,
      "grad_norm": 1.9648524522781372,
      "learning_rate": 6.229291151984233e-05,
      "loss": 1.7811,
      "step": 201
    },
    {
      "epoch": 0.13196145680222113,
      "grad_norm": 2.3082010746002197,
      "learning_rate": 6.19384977511771e-05,
      "loss": 1.8742,
      "step": 202
    },
    {
      "epoch": 0.132614731340846,
      "grad_norm": 1.6148680448532104,
      "learning_rate": 6.158344673142573e-05,
      "loss": 1.7536,
      "step": 203
    },
    {
      "epoch": 0.13326800587947085,
      "grad_norm": 1.3659769296646118,
      "learning_rate": 6.122777741244067e-05,
      "loss": 1.8803,
      "step": 204
    },
    {
      "epoch": 0.1339212804180957,
      "grad_norm": 1.0062769651412964,
      "learning_rate": 6.0871508779077856e-05,
      "loss": 1.8137,
      "step": 205
    },
    {
      "epoch": 0.13457455495672055,
      "grad_norm": 0.9940685629844666,
      "learning_rate": 6.051465984818332e-05,
      "loss": 1.8704,
      "step": 206
    },
    {
      "epoch": 0.13522782949534543,
      "grad_norm": 0.8455464839935303,
      "learning_rate": 6.015724966757812e-05,
      "loss": 1.5295,
      "step": 207
    },
    {
      "epoch": 0.13588110403397027,
      "grad_norm": 0.9129886627197266,
      "learning_rate": 5.979929731504158e-05,
      "loss": 1.8757,
      "step": 208
    },
    {
      "epoch": 0.13653437857259512,
      "grad_norm": 0.8862000107765198,
      "learning_rate": 5.944082189729301e-05,
      "loss": 1.8128,
      "step": 209
    },
    {
      "epoch": 0.13718765311122,
      "grad_norm": 0.8938460946083069,
      "learning_rate": 5.908184254897182e-05,
      "loss": 1.7945,
      "step": 210
    },
    {
      "epoch": 0.13784092764984485,
      "grad_norm": 0.9378989338874817,
      "learning_rate": 5.872237843161612e-05,
      "loss": 1.6753,
      "step": 211
    },
    {
      "epoch": 0.1384942021884697,
      "grad_norm": 0.9450460076332092,
      "learning_rate": 5.8362448732639894e-05,
      "loss": 1.6474,
      "step": 212
    },
    {
      "epoch": 0.13914747672709457,
      "grad_norm": 1.0634076595306396,
      "learning_rate": 5.800207266430895e-05,
      "loss": 1.9188,
      "step": 213
    },
    {
      "epoch": 0.13980075126571942,
      "grad_norm": 1.0193463563919067,
      "learning_rate": 5.764126946271526e-05,
      "loss": 1.8544,
      "step": 214
    },
    {
      "epoch": 0.14045402580434427,
      "grad_norm": 1.0252861976623535,
      "learning_rate": 5.7280058386750255e-05,
      "loss": 1.7737,
      "step": 215
    },
    {
      "epoch": 0.14110730034296914,
      "grad_norm": 1.09612238407135,
      "learning_rate": 5.6918458717076815e-05,
      "loss": 1.8946,
      "step": 216
    },
    {
      "epoch": 0.141760574881594,
      "grad_norm": 1.0598078966140747,
      "learning_rate": 5.655648975510014e-05,
      "loss": 1.7915,
      "step": 217
    },
    {
      "epoch": 0.14241384942021884,
      "grad_norm": 1.1029284000396729,
      "learning_rate": 5.61941708219374e-05,
      "loss": 1.8305,
      "step": 218
    },
    {
      "epoch": 0.14306712395884372,
      "grad_norm": 1.1266640424728394,
      "learning_rate": 5.583152125738651e-05,
      "loss": 1.8656,
      "step": 219
    },
    {
      "epoch": 0.14372039849746857,
      "grad_norm": 1.3155699968338013,
      "learning_rate": 5.546856041889373e-05,
      "loss": 1.9242,
      "step": 220
    },
    {
      "epoch": 0.14437367303609341,
      "grad_norm": 1.251673936843872,
      "learning_rate": 5.510530768052047e-05,
      "loss": 1.8705,
      "step": 221
    },
    {
      "epoch": 0.14502694757471826,
      "grad_norm": 1.3764578104019165,
      "learning_rate": 5.4741782431909136e-05,
      "loss": 2.1015,
      "step": 222
    },
    {
      "epoch": 0.14568022211334314,
      "grad_norm": 1.3341801166534424,
      "learning_rate": 5.437800407724812e-05,
      "loss": 1.9499,
      "step": 223
    },
    {
      "epoch": 0.146333496651968,
      "grad_norm": 1.3375170230865479,
      "learning_rate": 5.401399203423606e-05,
      "loss": 2.0764,
      "step": 224
    },
    {
      "epoch": 0.14698677119059284,
      "grad_norm": 1.3811525106430054,
      "learning_rate": 5.364976573304538e-05,
      "loss": 1.7937,
      "step": 225
    },
    {
      "epoch": 0.1476400457292177,
      "grad_norm": 1.5332218408584595,
      "learning_rate": 5.328534461528515e-05,
      "loss": 2.0772,
      "step": 226
    },
    {
      "epoch": 0.14829332026784256,
      "grad_norm": 1.4191771745681763,
      "learning_rate": 5.29207481329633e-05,
      "loss": 1.5856,
      "step": 227
    },
    {
      "epoch": 0.1489465948064674,
      "grad_norm": 1.5286246538162231,
      "learning_rate": 5.2555995747448364e-05,
      "loss": 1.8561,
      "step": 228
    },
    {
      "epoch": 0.14959986934509228,
      "grad_norm": 1.459039330482483,
      "learning_rate": 5.2191106928430644e-05,
      "loss": 1.6123,
      "step": 229
    },
    {
      "epoch": 0.15025314388371713,
      "grad_norm": 1.6728615760803223,
      "learning_rate": 5.182610115288295e-05,
      "loss": 1.8385,
      "step": 230
    },
    {
      "epoch": 0.15090641842234198,
      "grad_norm": 1.7702395915985107,
      "learning_rate": 5.1460997904021005e-05,
      "loss": 1.9649,
      "step": 231
    },
    {
      "epoch": 0.15155969296096686,
      "grad_norm": 2.0994086265563965,
      "learning_rate": 5.109581667026341e-05,
      "loss": 1.9379,
      "step": 232
    },
    {
      "epoch": 0.1522129674995917,
      "grad_norm": 2.1012067794799805,
      "learning_rate": 5.073057694419147e-05,
      "loss": 1.8879,
      "step": 233
    },
    {
      "epoch": 0.15286624203821655,
      "grad_norm": 2.374234676361084,
      "learning_rate": 5.036529822150865e-05,
      "loss": 2.2221,
      "step": 234
    },
    {
      "epoch": 0.15351951657684143,
      "grad_norm": 2.389725923538208,
      "learning_rate": 5e-05,
      "loss": 1.8453,
      "step": 235
    },
    {
      "epoch": 0.15417279111546628,
      "grad_norm": 2.585789203643799,
      "learning_rate": 4.963470177849135e-05,
      "loss": 1.902,
      "step": 236
    },
    {
      "epoch": 0.15482606565409113,
      "grad_norm": 2.6658775806427,
      "learning_rate": 4.9269423055808544e-05,
      "loss": 1.8788,
      "step": 237
    },
    {
      "epoch": 0.15547934019271598,
      "grad_norm": 3.1072347164154053,
      "learning_rate": 4.8904183329736596e-05,
      "loss": 2.1452,
      "step": 238
    },
    {
      "epoch": 0.15613261473134085,
      "grad_norm": 2.6957457065582275,
      "learning_rate": 4.853900209597903e-05,
      "loss": 1.5223,
      "step": 239
    },
    {
      "epoch": 0.1567858892699657,
      "grad_norm": 3.053032636642456,
      "learning_rate": 4.817389884711705e-05,
      "loss": 1.9055,
      "step": 240
    },
    {
      "epoch": 0.15743916380859055,
      "grad_norm": 3.4490959644317627,
      "learning_rate": 4.7808893071569374e-05,
      "loss": 1.8756,
      "step": 241
    },
    {
      "epoch": 0.15809243834721542,
      "grad_norm": 3.8312885761260986,
      "learning_rate": 4.744400425255165e-05,
      "loss": 2.4025,
      "step": 242
    },
    {
      "epoch": 0.15874571288584027,
      "grad_norm": 4.085025787353516,
      "learning_rate": 4.707925186703671e-05,
      "loss": 2.3357,
      "step": 243
    },
    {
      "epoch": 0.15939898742446512,
      "grad_norm": 3.8918635845184326,
      "learning_rate": 4.671465538471486e-05,
      "loss": 1.9563,
      "step": 244
    },
    {
      "epoch": 0.16005226196309,
      "grad_norm": 4.1230082511901855,
      "learning_rate": 4.6350234266954626e-05,
      "loss": 1.773,
      "step": 245
    },
    {
      "epoch": 0.16070553650171485,
      "grad_norm": 4.836731910705566,
      "learning_rate": 4.598600796576395e-05,
      "loss": 2.3734,
      "step": 246
    },
    {
      "epoch": 0.1613588110403397,
      "grad_norm": 4.500394821166992,
      "learning_rate": 4.562199592275188e-05,
      "loss": 2.0417,
      "step": 247
    },
    {
      "epoch": 0.16201208557896457,
      "grad_norm": 4.775685787200928,
      "learning_rate": 4.5258217568090876e-05,
      "loss": 1.8614,
      "step": 248
    },
    {
      "epoch": 0.16266536011758942,
      "grad_norm": 6.855151176452637,
      "learning_rate": 4.4894692319479544e-05,
      "loss": 2.6569,
      "step": 249
    },
    {
      "epoch": 0.16331863465621427,
      "grad_norm": 8.0609130859375,
      "learning_rate": 4.4531439581106295e-05,
      "loss": 3.1413,
      "step": 250
    },
    {
      "epoch": 0.16397190919483914,
      "grad_norm": 1.537583827972412,
      "learning_rate": 4.4168478742613506e-05,
      "loss": 1.6395,
      "step": 251
    },
    {
      "epoch": 0.164625183733464,
      "grad_norm": 1.7478508949279785,
      "learning_rate": 4.38058291780626e-05,
      "loss": 1.8051,
      "step": 252
    },
    {
      "epoch": 0.16527845827208884,
      "grad_norm": 1.6011650562286377,
      "learning_rate": 4.3443510244899864e-05,
      "loss": 1.846,
      "step": 253
    },
    {
      "epoch": 0.1659317328107137,
      "grad_norm": 1.3669625520706177,
      "learning_rate": 4.308154128292318e-05,
      "loss": 1.9526,
      "step": 254
    },
    {
      "epoch": 0.16658500734933857,
      "grad_norm": 1.0034922361373901,
      "learning_rate": 4.271994161324977e-05,
      "loss": 1.7868,
      "step": 255
    },
    {
      "epoch": 0.1672382818879634,
      "grad_norm": 0.8738464713096619,
      "learning_rate": 4.235873053728475e-05,
      "loss": 1.6867,
      "step": 256
    },
    {
      "epoch": 0.16789155642658826,
      "grad_norm": 0.8731593489646912,
      "learning_rate": 4.199792733569107e-05,
      "loss": 1.6642,
      "step": 257
    },
    {
      "epoch": 0.16854483096521314,
      "grad_norm": 0.8563171625137329,
      "learning_rate": 4.163755126736012e-05,
      "loss": 1.6139,
      "step": 258
    },
    {
      "epoch": 0.169198105503838,
      "grad_norm": 0.8093990683555603,
      "learning_rate": 4.127762156838389e-05,
      "loss": 1.5495,
      "step": 259
    },
    {
      "epoch": 0.16985138004246284,
      "grad_norm": 0.8655145168304443,
      "learning_rate": 4.0918157451028185e-05,
      "loss": 1.7698,
      "step": 260
    },
    {
      "epoch": 0.1705046545810877,
      "grad_norm": 0.8737233877182007,
      "learning_rate": 4.055917810270698e-05,
      "loss": 1.5935,
      "step": 261
    },
    {
      "epoch": 0.17115792911971256,
      "grad_norm": 0.9041045308113098,
      "learning_rate": 4.020070268495843e-05,
      "loss": 1.7678,
      "step": 262
    },
    {
      "epoch": 0.1718112036583374,
      "grad_norm": 0.8886668682098389,
      "learning_rate": 3.9842750332421896e-05,
      "loss": 1.6847,
      "step": 263
    },
    {
      "epoch": 0.17246447819696228,
      "grad_norm": 0.9836606979370117,
      "learning_rate": 3.94853401518167e-05,
      "loss": 1.8108,
      "step": 264
    },
    {
      "epoch": 0.17311775273558713,
      "grad_norm": 0.9882153272628784,
      "learning_rate": 3.9128491220922156e-05,
      "loss": 1.7874,
      "step": 265
    },
    {
      "epoch": 0.17377102727421198,
      "grad_norm": 1.035111427307129,
      "learning_rate": 3.877222258755935e-05,
      "loss": 1.897,
      "step": 266
    },
    {
      "epoch": 0.17442430181283686,
      "grad_norm": 0.9586417078971863,
      "learning_rate": 3.8416553268574285e-05,
      "loss": 1.5846,
      "step": 267
    },
    {
      "epoch": 0.1750775763514617,
      "grad_norm": 1.0587044954299927,
      "learning_rate": 3.80615022488229e-05,
      "loss": 1.8779,
      "step": 268
    },
    {
      "epoch": 0.17573085089008655,
      "grad_norm": 1.0741347074508667,
      "learning_rate": 3.770708848015768e-05,
      "loss": 1.7032,
      "step": 269
    },
    {
      "epoch": 0.1763841254287114,
      "grad_norm": 1.1013596057891846,
      "learning_rate": 3.735333088041596e-05,
      "loss": 1.7852,
      "step": 270
    },
    {
      "epoch": 0.17703739996733628,
      "grad_norm": 1.1117128133773804,
      "learning_rate": 3.7000248332410304e-05,
      "loss": 1.6237,
      "step": 271
    },
    {
      "epoch": 0.17769067450596113,
      "grad_norm": 1.2233641147613525,
      "learning_rate": 3.664785968292036e-05,
      "loss": 1.8614,
      "step": 272
    },
    {
      "epoch": 0.17834394904458598,
      "grad_norm": 1.2640923261642456,
      "learning_rate": 3.629618374168711e-05,
      "loss": 1.8498,
      "step": 273
    },
    {
      "epoch": 0.17899722358321085,
      "grad_norm": 1.2666478157043457,
      "learning_rate": 3.594523928040859e-05,
      "loss": 1.7807,
      "step": 274
    },
    {
      "epoch": 0.1796504981218357,
      "grad_norm": 1.3259072303771973,
      "learning_rate": 3.5595045031738125e-05,
      "loss": 1.8528,
      "step": 275
    },
    {
      "epoch": 0.18030377266046055,
      "grad_norm": 1.4057447910308838,
      "learning_rate": 3.5245619688284274e-05,
      "loss": 1.9501,
      "step": 276
    },
    {
      "epoch": 0.18095704719908542,
      "grad_norm": 1.2778233289718628,
      "learning_rate": 3.4896981901613104e-05,
      "loss": 1.5052,
      "step": 277
    },
    {
      "epoch": 0.18161032173771027,
      "grad_norm": 1.519966959953308,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.9984,
      "step": 278
    },
    {
      "epoch": 0.18226359627633512,
      "grad_norm": 1.6918063163757324,
      "learning_rate": 3.420214339369944e-05,
      "loss": 2.0946,
      "step": 279
    },
    {
      "epoch": 0.18291687081496,
      "grad_norm": 1.52606201171875,
      "learning_rate": 3.38559797614277e-05,
      "loss": 1.615,
      "step": 280
    },
    {
      "epoch": 0.18357014535358485,
      "grad_norm": 1.8061113357543945,
      "learning_rate": 3.351067786190038e-05,
      "loss": 1.9404,
      "step": 281
    },
    {
      "epoch": 0.1842234198922097,
      "grad_norm": 1.8564739227294922,
      "learning_rate": 3.316625612658315e-05,
      "loss": 1.7117,
      "step": 282
    },
    {
      "epoch": 0.18487669443083457,
      "grad_norm": 2.030742883682251,
      "learning_rate": 3.282273293996027e-05,
      "loss": 2.1687,
      "step": 283
    },
    {
      "epoch": 0.18552996896945942,
      "grad_norm": 2.187425374984741,
      "learning_rate": 3.248012663855353e-05,
      "loss": 1.8413,
      "step": 284
    },
    {
      "epoch": 0.18618324350808427,
      "grad_norm": 2.310781240463257,
      "learning_rate": 3.2138455509943366e-05,
      "loss": 1.806,
      "step": 285
    },
    {
      "epoch": 0.18683651804670912,
      "grad_norm": 2.575011730194092,
      "learning_rate": 3.179773779179267e-05,
      "loss": 2.0005,
      "step": 286
    },
    {
      "epoch": 0.187489792585334,
      "grad_norm": 2.6773452758789062,
      "learning_rate": 3.145799167087342e-05,
      "loss": 2.1075,
      "step": 287
    },
    {
      "epoch": 0.18814306712395884,
      "grad_norm": 2.9845211505889893,
      "learning_rate": 3.111923528209577e-05,
      "loss": 2.0038,
      "step": 288
    },
    {
      "epoch": 0.1887963416625837,
      "grad_norm": 3.069061756134033,
      "learning_rate": 3.078148670754022e-05,
      "loss": 1.9766,
      "step": 289
    },
    {
      "epoch": 0.18944961620120856,
      "grad_norm": 3.2706315517425537,
      "learning_rate": 3.0444763975492208e-05,
      "loss": 1.9039,
      "step": 290
    },
    {
      "epoch": 0.1901028907398334,
      "grad_norm": 3.069990634918213,
      "learning_rate": 3.0109085059480017e-05,
      "loss": 1.827,
      "step": 291
    },
    {
      "epoch": 0.19075616527845826,
      "grad_norm": 3.9291090965270996,
      "learning_rate": 2.977446787731532e-05,
      "loss": 2.0023,
      "step": 292
    },
    {
      "epoch": 0.19140943981708314,
      "grad_norm": 3.550333023071289,
      "learning_rate": 2.944093029013664e-05,
      "loss": 1.8593,
      "step": 293
    },
    {
      "epoch": 0.192062714355708,
      "grad_norm": 3.8922524452209473,
      "learning_rate": 2.910849010145617e-05,
      "loss": 2.4151,
      "step": 294
    },
    {
      "epoch": 0.19271598889433283,
      "grad_norm": 4.204081058502197,
      "learning_rate": 2.8777165056209256e-05,
      "loss": 1.8811,
      "step": 295
    },
    {
      "epoch": 0.1933692634329577,
      "grad_norm": 3.8980872631073,
      "learning_rate": 2.8446972839807384e-05,
      "loss": 2.0131,
      "step": 296
    },
    {
      "epoch": 0.19402253797158256,
      "grad_norm": 4.564809322357178,
      "learning_rate": 2.8117931077194065e-05,
      "loss": 2.2069,
      "step": 297
    },
    {
      "epoch": 0.1946758125102074,
      "grad_norm": 4.025120735168457,
      "learning_rate": 2.7790057331904117e-05,
      "loss": 1.8485,
      "step": 298
    },
    {
      "epoch": 0.19532908704883228,
      "grad_norm": 6.8745903968811035,
      "learning_rate": 2.746336910512606e-05,
      "loss": 2.2955,
      "step": 299
    },
    {
      "epoch": 0.19598236158745713,
      "grad_norm": 7.931851863861084,
      "learning_rate": 2.7137883834768073e-05,
      "loss": 2.4004,
      "step": 300
    },
    {
      "epoch": 0.19598236158745713,
      "eval_loss": 1.901733160018921,
      "eval_runtime": 49.6565,
      "eval_samples_per_second": 51.917,
      "eval_steps_per_second": 12.989,
      "step": 300
    },
    {
      "epoch": 0.19663563612608198,
      "grad_norm": 0.6562613844871521,
      "learning_rate": 2.6813618894527138e-05,
      "loss": 1.4429,
      "step": 301
    },
    {
      "epoch": 0.19728891066470686,
      "grad_norm": 0.7796823978424072,
      "learning_rate": 2.6490591592961578e-05,
      "loss": 1.6504,
      "step": 302
    },
    {
      "epoch": 0.1979421852033317,
      "grad_norm": 0.7866513729095459,
      "learning_rate": 2.6168819172567392e-05,
      "loss": 1.7214,
      "step": 303
    },
    {
      "epoch": 0.19859545974195655,
      "grad_norm": 0.7742399573326111,
      "learning_rate": 2.5848318808857606e-05,
      "loss": 1.708,
      "step": 304
    },
    {
      "epoch": 0.1992487342805814,
      "grad_norm": 0.8145656585693359,
      "learning_rate": 2.5529107609445733e-05,
      "loss": 1.8016,
      "step": 305
    },
    {
      "epoch": 0.19990200881920628,
      "grad_norm": 0.7754919528961182,
      "learning_rate": 2.521120261313241e-05,
      "loss": 1.3985,
      "step": 306
    },
    {
      "epoch": 0.20055528335783113,
      "grad_norm": 0.8018137812614441,
      "learning_rate": 2.4894620788996037e-05,
      "loss": 1.7272,
      "step": 307
    },
    {
      "epoch": 0.20120855789645598,
      "grad_norm": 0.8388513922691345,
      "learning_rate": 2.457937903548695e-05,
      "loss": 1.7512,
      "step": 308
    },
    {
      "epoch": 0.20186183243508085,
      "grad_norm": 0.8507857918739319,
      "learning_rate": 2.426549417952542e-05,
      "loss": 1.7692,
      "step": 309
    },
    {
      "epoch": 0.2025151069737057,
      "grad_norm": 0.852091372013092,
      "learning_rate": 2.3952982975603496e-05,
      "loss": 1.7928,
      "step": 310
    },
    {
      "epoch": 0.20316838151233055,
      "grad_norm": 0.8825722336769104,
      "learning_rate": 2.3641862104890595e-05,
      "loss": 1.7296,
      "step": 311
    },
    {
      "epoch": 0.20382165605095542,
      "grad_norm": 0.8713802099227905,
      "learning_rate": 2.3332148174343254e-05,
      "loss": 1.6381,
      "step": 312
    },
    {
      "epoch": 0.20447493058958027,
      "grad_norm": 0.9311487078666687,
      "learning_rate": 2.3023857715818532e-05,
      "loss": 1.6957,
      "step": 313
    },
    {
      "epoch": 0.20512820512820512,
      "grad_norm": 0.932761013507843,
      "learning_rate": 2.2717007185191674e-05,
      "loss": 1.5765,
      "step": 314
    },
    {
      "epoch": 0.20578147966683,
      "grad_norm": 0.9953693151473999,
      "learning_rate": 2.24116129614777e-05,
      "loss": 1.9627,
      "step": 315
    },
    {
      "epoch": 0.20643475420545485,
      "grad_norm": 1.0859706401824951,
      "learning_rate": 2.2107691345957133e-05,
      "loss": 1.9455,
      "step": 316
    },
    {
      "epoch": 0.2070880287440797,
      "grad_norm": 1.123296856880188,
      "learning_rate": 2.1805258561305862e-05,
      "loss": 1.9259,
      "step": 317
    },
    {
      "epoch": 0.20774130328270457,
      "grad_norm": 1.0700737237930298,
      "learning_rate": 2.1504330750729186e-05,
      "loss": 1.6377,
      "step": 318
    },
    {
      "epoch": 0.20839457782132942,
      "grad_norm": 1.1277694702148438,
      "learning_rate": 2.120492397710022e-05,
      "loss": 1.7249,
      "step": 319
    },
    {
      "epoch": 0.20904785235995427,
      "grad_norm": 1.2123768329620361,
      "learning_rate": 2.090705422210237e-05,
      "loss": 1.9393,
      "step": 320
    },
    {
      "epoch": 0.20970112689857912,
      "grad_norm": 1.2336665391921997,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.7775,
      "step": 321
    },
    {
      "epoch": 0.210354401437204,
      "grad_norm": 1.2337082624435425,
      "learning_rate": 2.0315989283671473e-05,
      "loss": 1.8453,
      "step": 322
    },
    {
      "epoch": 0.21100767597582884,
      "grad_norm": 1.2970980405807495,
      "learning_rate": 2.0022825650001387e-05,
      "loss": 1.8926,
      "step": 323
    },
    {
      "epoch": 0.2116609505144537,
      "grad_norm": 1.3687418699264526,
      "learning_rate": 1.9731262132804274e-05,
      "loss": 2.0265,
      "step": 324
    },
    {
      "epoch": 0.21231422505307856,
      "grad_norm": 1.4175206422805786,
      "learning_rate": 1.9441314295107537e-05,
      "loss": 1.9937,
      "step": 325
    },
    {
      "epoch": 0.2129674995917034,
      "grad_norm": 1.4179284572601318,
      "learning_rate": 1.9152997613697183e-05,
      "loss": 1.7641,
      "step": 326
    },
    {
      "epoch": 0.21362077413032826,
      "grad_norm": 1.5280499458312988,
      "learning_rate": 1.8866327478291546e-05,
      "loss": 1.9075,
      "step": 327
    },
    {
      "epoch": 0.21427404866895314,
      "grad_norm": 1.6011749505996704,
      "learning_rate": 1.8581319190720035e-05,
      "loss": 1.9116,
      "step": 328
    },
    {
      "epoch": 0.21492732320757799,
      "grad_norm": 1.6942389011383057,
      "learning_rate": 1.8297987964106115e-05,
      "loss": 1.9134,
      "step": 329
    },
    {
      "epoch": 0.21558059774620283,
      "grad_norm": 1.9133878946304321,
      "learning_rate": 1.801634892205545e-05,
      "loss": 1.9993,
      "step": 330
    },
    {
      "epoch": 0.2162338722848277,
      "grad_norm": 1.8108885288238525,
      "learning_rate": 1.7736417097848506e-05,
      "loss": 1.6007,
      "step": 331
    },
    {
      "epoch": 0.21688714682345256,
      "grad_norm": 2.2170968055725098,
      "learning_rate": 1.7458207433638223e-05,
      "loss": 2.0009,
      "step": 332
    },
    {
      "epoch": 0.2175404213620774,
      "grad_norm": 2.21116042137146,
      "learning_rate": 1.718173477965236e-05,
      "loss": 1.7992,
      "step": 333
    },
    {
      "epoch": 0.21819369590070228,
      "grad_norm": 2.4326541423797607,
      "learning_rate": 1.6907013893400837e-05,
      "loss": 2.0342,
      "step": 334
    },
    {
      "epoch": 0.21884697043932713,
      "grad_norm": 2.476644515991211,
      "learning_rate": 1.6634059438888033e-05,
      "loss": 1.6655,
      "step": 335
    },
    {
      "epoch": 0.21950024497795198,
      "grad_norm": 2.754042387008667,
      "learning_rate": 1.636288598583e-05,
      "loss": 2.1341,
      "step": 336
    },
    {
      "epoch": 0.22015351951657683,
      "grad_norm": 2.8820323944091797,
      "learning_rate": 1.6093508008876857e-05,
      "loss": 1.7264,
      "step": 337
    },
    {
      "epoch": 0.2208067940552017,
      "grad_norm": 2.952387571334839,
      "learning_rate": 1.5825939886840037e-05,
      "loss": 1.9281,
      "step": 338
    },
    {
      "epoch": 0.22146006859382655,
      "grad_norm": 2.917340040206909,
      "learning_rate": 1.5560195901924894e-05,
      "loss": 1.9721,
      "step": 339
    },
    {
      "epoch": 0.2221133431324514,
      "grad_norm": 3.519662380218506,
      "learning_rate": 1.5296290238968303e-05,
      "loss": 2.2708,
      "step": 340
    },
    {
      "epoch": 0.22276661767107628,
      "grad_norm": 3.6640732288360596,
      "learning_rate": 1.50342369846815e-05,
      "loss": 1.9861,
      "step": 341
    },
    {
      "epoch": 0.22341989220970113,
      "grad_norm": 3.531583547592163,
      "learning_rate": 1.4774050126898164e-05,
      "loss": 2.135,
      "step": 342
    },
    {
      "epoch": 0.22407316674832597,
      "grad_norm": 4.102410793304443,
      "learning_rate": 1.451574355382776e-05,
      "loss": 2.0721,
      "step": 343
    },
    {
      "epoch": 0.22472644128695085,
      "grad_norm": 3.8423666954040527,
      "learning_rate": 1.425933105331429e-05,
      "loss": 2.0958,
      "step": 344
    },
    {
      "epoch": 0.2253797158255757,
      "grad_norm": 3.899322986602783,
      "learning_rate": 1.4004826312100216e-05,
      "loss": 2.1085,
      "step": 345
    },
    {
      "epoch": 0.22603299036420055,
      "grad_norm": 4.240949630737305,
      "learning_rate": 1.3752242915095992e-05,
      "loss": 2.0961,
      "step": 346
    },
    {
      "epoch": 0.22668626490282542,
      "grad_norm": 4.414612293243408,
      "learning_rate": 1.3501594344654884e-05,
      "loss": 1.8973,
      "step": 347
    },
    {
      "epoch": 0.22733953944145027,
      "grad_norm": 5.2035746574401855,
      "learning_rate": 1.3252893979853304e-05,
      "loss": 2.3987,
      "step": 348
    },
    {
      "epoch": 0.22799281398007512,
      "grad_norm": 6.975047588348389,
      "learning_rate": 1.3006155095776707e-05,
      "loss": 2.3935,
      "step": 349
    },
    {
      "epoch": 0.2286460885187,
      "grad_norm": 9.769250869750977,
      "learning_rate": 1.2761390862810907e-05,
      "loss": 3.1776,
      "step": 350
    },
    {
      "epoch": 0.22929936305732485,
      "grad_norm": 0.6493518948554993,
      "learning_rate": 1.2518614345939212e-05,
      "loss": 1.517,
      "step": 351
    },
    {
      "epoch": 0.2299526375959497,
      "grad_norm": 0.7682214379310608,
      "learning_rate": 1.227783850404487e-05,
      "loss": 1.7512,
      "step": 352
    },
    {
      "epoch": 0.23060591213457454,
      "grad_norm": 0.7854646444320679,
      "learning_rate": 1.2039076189219517e-05,
      "loss": 1.7238,
      "step": 353
    },
    {
      "epoch": 0.23125918667319942,
      "grad_norm": 0.7929527163505554,
      "learning_rate": 1.1802340146077045e-05,
      "loss": 1.8564,
      "step": 354
    },
    {
      "epoch": 0.23191246121182427,
      "grad_norm": 0.8244699835777283,
      "learning_rate": 1.1567643011073392e-05,
      "loss": 1.6421,
      "step": 355
    },
    {
      "epoch": 0.23256573575044912,
      "grad_norm": 0.8072249889373779,
      "learning_rate": 1.1334997311832002e-05,
      "loss": 1.6762,
      "step": 356
    },
    {
      "epoch": 0.233219010289074,
      "grad_norm": 0.8119028806686401,
      "learning_rate": 1.1104415466475087e-05,
      "loss": 1.6528,
      "step": 357
    },
    {
      "epoch": 0.23387228482769884,
      "grad_norm": 0.8597144484519958,
      "learning_rate": 1.0875909782960886e-05,
      "loss": 1.8121,
      "step": 358
    },
    {
      "epoch": 0.2345255593663237,
      "grad_norm": 0.8278293609619141,
      "learning_rate": 1.0649492458426564e-05,
      "loss": 1.6059,
      "step": 359
    },
    {
      "epoch": 0.23517883390494856,
      "grad_norm": 0.8635007739067078,
      "learning_rate": 1.0425175578537299e-05,
      "loss": 1.7619,
      "step": 360
    },
    {
      "epoch": 0.2358321084435734,
      "grad_norm": 0.873079776763916,
      "learning_rate": 1.020297111684101e-05,
      "loss": 1.7476,
      "step": 361
    },
    {
      "epoch": 0.23648538298219826,
      "grad_norm": 0.8993417024612427,
      "learning_rate": 9.98289093412938e-06,
      "loss": 1.7883,
      "step": 362
    },
    {
      "epoch": 0.23713865752082314,
      "grad_norm": 0.8890747427940369,
      "learning_rate": 9.764946777804646e-06,
      "loss": 1.6988,
      "step": 363
    },
    {
      "epoch": 0.23779193205944799,
      "grad_norm": 0.926846981048584,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.618,
      "step": 364
    },
    {
      "epoch": 0.23844520659807283,
      "grad_norm": 0.9523332118988037,
      "learning_rate": 9.335512963221732e-06,
      "loss": 1.6626,
      "step": 365
    },
    {
      "epoch": 0.2390984811366977,
      "grad_norm": 0.9900181293487549,
      "learning_rate": 9.124046227208082e-06,
      "loss": 1.646,
      "step": 366
    },
    {
      "epoch": 0.23975175567532256,
      "grad_norm": 1.0082789659500122,
      "learning_rate": 8.914761360846869e-06,
      "loss": 1.7179,
      "step": 367
    },
    {
      "epoch": 0.2404050302139474,
      "grad_norm": 1.045430302619934,
      "learning_rate": 8.707669535309793e-06,
      "loss": 1.6883,
      "step": 368
    },
    {
      "epoch": 0.24105830475257226,
      "grad_norm": 1.0794562101364136,
      "learning_rate": 8.502781804708826e-06,
      "loss": 1.8601,
      "step": 369
    },
    {
      "epoch": 0.24171157929119713,
      "grad_norm": 1.1484591960906982,
      "learning_rate": 8.30010910550611e-06,
      "loss": 1.7968,
      "step": 370
    },
    {
      "epoch": 0.24236485382982198,
      "grad_norm": 1.1732699871063232,
      "learning_rate": 8.09966225593024e-06,
      "loss": 1.6156,
      "step": 371
    },
    {
      "epoch": 0.24301812836844683,
      "grad_norm": 1.1297372579574585,
      "learning_rate": 7.901451955398792e-06,
      "loss": 1.5771,
      "step": 372
    },
    {
      "epoch": 0.2436714029070717,
      "grad_norm": 1.2523436546325684,
      "learning_rate": 7.705488783947202e-06,
      "loss": 2.0035,
      "step": 373
    },
    {
      "epoch": 0.24432467744569655,
      "grad_norm": 1.260083794593811,
      "learning_rate": 7.511783201664052e-06,
      "loss": 1.8555,
      "step": 374
    },
    {
      "epoch": 0.2449779519843214,
      "grad_norm": 1.3140181303024292,
      "learning_rate": 7.320345548132679e-06,
      "loss": 1.8955,
      "step": 375
    },
    {
      "epoch": 0.24563122652294628,
      "grad_norm": 1.4021167755126953,
      "learning_rate": 7.131186041879357e-06,
      "loss": 1.7796,
      "step": 376
    },
    {
      "epoch": 0.24628450106157113,
      "grad_norm": 1.448630690574646,
      "learning_rate": 6.944314779827749e-06,
      "loss": 1.9388,
      "step": 377
    },
    {
      "epoch": 0.24693777560019597,
      "grad_norm": 1.4907878637313843,
      "learning_rate": 6.759741736760061e-06,
      "loss": 1.7724,
      "step": 378
    },
    {
      "epoch": 0.24759105013882085,
      "grad_norm": 1.641314148902893,
      "learning_rate": 6.577476764784546e-06,
      "loss": 1.8177,
      "step": 379
    },
    {
      "epoch": 0.2482443246774457,
      "grad_norm": 1.7289384603500366,
      "learning_rate": 6.397529592809614e-06,
      "loss": 2.0931,
      "step": 380
    },
    {
      "epoch": 0.24889759921607055,
      "grad_norm": 1.696372389793396,
      "learning_rate": 6.219909826024589e-06,
      "loss": 1.7295,
      "step": 381
    },
    {
      "epoch": 0.24955087375469542,
      "grad_norm": 2.057739496231079,
      "learning_rate": 6.0446269453868945e-06,
      "loss": 1.9874,
      "step": 382
    },
    {
      "epoch": 0.25020414829332027,
      "grad_norm": 2.0013246536254883,
      "learning_rate": 5.871690307116107e-06,
      "loss": 1.8563,
      "step": 383
    },
    {
      "epoch": 0.2508574228319451,
      "grad_norm": 2.1932458877563477,
      "learning_rate": 5.701109142194422e-06,
      "loss": 2.0006,
      "step": 384
    },
    {
      "epoch": 0.25151069737056997,
      "grad_norm": 2.1517393589019775,
      "learning_rate": 5.532892555874059e-06,
      "loss": 1.9219,
      "step": 385
    },
    {
      "epoch": 0.2521639719091948,
      "grad_norm": 2.4914627075195312,
      "learning_rate": 5.3670495271910925e-06,
      "loss": 2.1114,
      "step": 386
    },
    {
      "epoch": 0.2528172464478197,
      "grad_norm": 2.510403633117676,
      "learning_rate": 5.203588908486279e-06,
      "loss": 1.6452,
      "step": 387
    },
    {
      "epoch": 0.25347052098644457,
      "grad_norm": 3.1688222885131836,
      "learning_rate": 5.042519424932513e-06,
      "loss": 2.1944,
      "step": 388
    },
    {
      "epoch": 0.2541237955250694,
      "grad_norm": 3.033691167831421,
      "learning_rate": 4.883849674069058e-06,
      "loss": 1.8288,
      "step": 389
    },
    {
      "epoch": 0.25477707006369427,
      "grad_norm": 3.0903337001800537,
      "learning_rate": 4.727588125342669e-06,
      "loss": 1.8231,
      "step": 390
    },
    {
      "epoch": 0.2554303446023191,
      "grad_norm": 3.200044631958008,
      "learning_rate": 4.573743119655516e-06,
      "loss": 2.1962,
      "step": 391
    },
    {
      "epoch": 0.25608361914094396,
      "grad_norm": 3.5135629177093506,
      "learning_rate": 4.422322868919937e-06,
      "loss": 1.8037,
      "step": 392
    },
    {
      "epoch": 0.2567368936795688,
      "grad_norm": 3.3575761318206787,
      "learning_rate": 4.273335455620097e-06,
      "loss": 1.5867,
      "step": 393
    },
    {
      "epoch": 0.2573901682181937,
      "grad_norm": 4.191507816314697,
      "learning_rate": 4.126788832380629e-06,
      "loss": 2.2939,
      "step": 394
    },
    {
      "epoch": 0.25804344275681856,
      "grad_norm": 3.7249908447265625,
      "learning_rate": 3.982690821542035e-06,
      "loss": 1.874,
      "step": 395
    },
    {
      "epoch": 0.2586967172954434,
      "grad_norm": 3.9365010261535645,
      "learning_rate": 3.8410491147432395e-06,
      "loss": 1.6268,
      "step": 396
    },
    {
      "epoch": 0.25934999183406826,
      "grad_norm": 4.847893238067627,
      "learning_rate": 3.7018712725109926e-06,
      "loss": 2.2378,
      "step": 397
    },
    {
      "epoch": 0.2600032663726931,
      "grad_norm": 4.60577392578125,
      "learning_rate": 3.5651647238562904e-06,
      "loss": 2.1083,
      "step": 398
    },
    {
      "epoch": 0.26065654091131796,
      "grad_norm": 5.157360553741455,
      "learning_rate": 3.430936765877857e-06,
      "loss": 2.6833,
      "step": 399
    },
    {
      "epoch": 0.26130981544994286,
      "grad_norm": 7.585251331329346,
      "learning_rate": 3.299194563372604e-06,
      "loss": 2.6346,
      "step": 400
    },
    {
      "epoch": 0.2619630899885677,
      "grad_norm": 0.5975827574729919,
      "learning_rate": 3.1699451484532463e-06,
      "loss": 1.4453,
      "step": 401
    },
    {
      "epoch": 0.26261636452719256,
      "grad_norm": 0.7209427356719971,
      "learning_rate": 3.0431954201728784e-06,
      "loss": 1.6149,
      "step": 402
    },
    {
      "epoch": 0.2632696390658174,
      "grad_norm": 0.7632143497467041,
      "learning_rate": 2.9189521441567726e-06,
      "loss": 1.8857,
      "step": 403
    },
    {
      "epoch": 0.26392291360444226,
      "grad_norm": 0.7980583310127258,
      "learning_rate": 2.797221952241219e-06,
      "loss": 1.7687,
      "step": 404
    },
    {
      "epoch": 0.2645761881430671,
      "grad_norm": 0.7555380463600159,
      "learning_rate": 2.6780113421195298e-06,
      "loss": 1.7941,
      "step": 405
    },
    {
      "epoch": 0.265229462681692,
      "grad_norm": 0.7890255451202393,
      "learning_rate": 2.561326676995218e-06,
      "loss": 1.6515,
      "step": 406
    },
    {
      "epoch": 0.26588273722031686,
      "grad_norm": 0.7849552631378174,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.6997,
      "step": 407
    },
    {
      "epoch": 0.2665360117589417,
      "grad_norm": 0.8335103392601013,
      "learning_rate": 2.3355599600729915e-06,
      "loss": 1.759,
      "step": 408
    },
    {
      "epoch": 0.26718928629756655,
      "grad_norm": 0.8428650498390198,
      "learning_rate": 2.2264899592121744e-06,
      "loss": 1.6588,
      "step": 409
    },
    {
      "epoch": 0.2678425608361914,
      "grad_norm": 0.8650283217430115,
      "learning_rate": 2.1199700045797077e-06,
      "loss": 1.7422,
      "step": 410
    },
    {
      "epoch": 0.26849583537481625,
      "grad_norm": 0.8699631690979004,
      "learning_rate": 2.0160057819794466e-06,
      "loss": 1.7059,
      "step": 411
    },
    {
      "epoch": 0.2691491099134411,
      "grad_norm": 1.0831049680709839,
      "learning_rate": 1.9146028407958484e-06,
      "loss": 1.8546,
      "step": 412
    },
    {
      "epoch": 0.269802384452066,
      "grad_norm": 0.9282965660095215,
      "learning_rate": 1.8157665936977263e-06,
      "loss": 1.7158,
      "step": 413
    },
    {
      "epoch": 0.27045565899069085,
      "grad_norm": 0.9040890336036682,
      "learning_rate": 1.7195023163493252e-06,
      "loss": 1.7635,
      "step": 414
    },
    {
      "epoch": 0.2711089335293157,
      "grad_norm": 0.9390372633934021,
      "learning_rate": 1.6258151471287396e-06,
      "loss": 1.5842,
      "step": 415
    },
    {
      "epoch": 0.27176220806794055,
      "grad_norm": 1.0083568096160889,
      "learning_rate": 1.5347100868536246e-06,
      "loss": 1.6398,
      "step": 416
    },
    {
      "epoch": 0.2724154826065654,
      "grad_norm": 1.0582306385040283,
      "learning_rate": 1.4461919985142735e-06,
      "loss": 1.9054,
      "step": 417
    },
    {
      "epoch": 0.27306875714519024,
      "grad_norm": 1.031751275062561,
      "learning_rate": 1.3602656070140275e-06,
      "loss": 1.6964,
      "step": 418
    },
    {
      "epoch": 0.27372203168381515,
      "grad_norm": 1.1092511415481567,
      "learning_rate": 1.27693549891707e-06,
      "loss": 1.7938,
      "step": 419
    },
    {
      "epoch": 0.27437530622244,
      "grad_norm": 1.1369683742523193,
      "learning_rate": 1.196206122203647e-06,
      "loss": 1.9818,
      "step": 420
    },
    {
      "epoch": 0.27502858076106484,
      "grad_norm": 1.1983885765075684,
      "learning_rate": 1.1180817860325599e-06,
      "loss": 1.8541,
      "step": 421
    },
    {
      "epoch": 0.2756818552996897,
      "grad_norm": 1.272538185119629,
      "learning_rate": 1.0425666605112517e-06,
      "loss": 1.8536,
      "step": 422
    },
    {
      "epoch": 0.27633512983831454,
      "grad_norm": 1.3353904485702515,
      "learning_rate": 9.696647764731337e-07,
      "loss": 1.9665,
      "step": 423
    },
    {
      "epoch": 0.2769884043769394,
      "grad_norm": 1.3927992582321167,
      "learning_rate": 8.993800252624862e-07,
      "loss": 1.8969,
      "step": 424
    },
    {
      "epoch": 0.27764167891556424,
      "grad_norm": 1.3316028118133545,
      "learning_rate": 8.317161585266964e-07,
      "loss": 1.6964,
      "step": 425
    },
    {
      "epoch": 0.27829495345418914,
      "grad_norm": 1.4543392658233643,
      "learning_rate": 7.666767880160464e-07,
      "loss": 1.8206,
      "step": 426
    },
    {
      "epoch": 0.278948227992814,
      "grad_norm": 1.5308529138565063,
      "learning_rate": 7.042653853909064e-07,
      "loss": 1.8566,
      "step": 427
    },
    {
      "epoch": 0.27960150253143884,
      "grad_norm": 1.5561844110488892,
      "learning_rate": 6.444852820364222e-07,
      "loss": 1.9246,
      "step": 428
    },
    {
      "epoch": 0.2802547770700637,
      "grad_norm": 1.674401879310608,
      "learning_rate": 5.87339668884701e-07,
      "loss": 1.8902,
      "step": 429
    },
    {
      "epoch": 0.28090805160868854,
      "grad_norm": 1.7931857109069824,
      "learning_rate": 5.328315962444874e-07,
      "loss": 1.8674,
      "step": 430
    },
    {
      "epoch": 0.2815613261473134,
      "grad_norm": 1.801674485206604,
      "learning_rate": 4.809639736383431e-07,
      "loss": 1.8449,
      "step": 431
    },
    {
      "epoch": 0.2822146006859383,
      "grad_norm": 1.9189867973327637,
      "learning_rate": 4.317395696473214e-07,
      "loss": 1.8103,
      "step": 432
    },
    {
      "epoch": 0.28286787522456314,
      "grad_norm": 2.1448371410369873,
      "learning_rate": 3.851610117632354e-07,
      "loss": 1.9176,
      "step": 433
    },
    {
      "epoch": 0.283521149763188,
      "grad_norm": 2.214698314666748,
      "learning_rate": 3.4123078624834216e-07,
      "loss": 1.8616,
      "step": 434
    },
    {
      "epoch": 0.28417442430181283,
      "grad_norm": 2.553487539291382,
      "learning_rate": 2.9995123800270476e-07,
      "loss": 2.1907,
      "step": 435
    },
    {
      "epoch": 0.2848276988404377,
      "grad_norm": 2.483630657196045,
      "learning_rate": 2.613245704389644e-07,
      "loss": 1.8343,
      "step": 436
    },
    {
      "epoch": 0.28548097337906253,
      "grad_norm": 2.82547926902771,
      "learning_rate": 2.2535284536476242e-07,
      "loss": 1.8856,
      "step": 437
    },
    {
      "epoch": 0.28613424791768743,
      "grad_norm": 2.9452755451202393,
      "learning_rate": 1.920379828726726e-07,
      "loss": 1.6969,
      "step": 438
    },
    {
      "epoch": 0.2867875224563123,
      "grad_norm": 3.092703342437744,
      "learning_rate": 1.6138176123770554e-07,
      "loss": 1.9073,
      "step": 439
    },
    {
      "epoch": 0.28744079699493713,
      "grad_norm": 3.6752867698669434,
      "learning_rate": 1.333858168224178e-07,
      "loss": 2.4456,
      "step": 440
    },
    {
      "epoch": 0.288094071533562,
      "grad_norm": 3.8016953468322754,
      "learning_rate": 1.0805164398952072e-07,
      "loss": 2.0624,
      "step": 441
    },
    {
      "epoch": 0.28874734607218683,
      "grad_norm": 3.2749228477478027,
      "learning_rate": 8.53805950221498e-08,
      "loss": 1.6564,
      "step": 442
    },
    {
      "epoch": 0.2894006206108117,
      "grad_norm": 3.4237606525421143,
      "learning_rate": 6.537388005167233e-08,
      "loss": 1.6582,
      "step": 443
    },
    {
      "epoch": 0.2900538951494365,
      "grad_norm": 3.6927602291107178,
      "learning_rate": 4.8032566993089225e-08,
      "loss": 1.9234,
      "step": 444
    },
    {
      "epoch": 0.29070716968806143,
      "grad_norm": 4.420969486236572,
      "learning_rate": 3.3357581488030475e-08,
      "loss": 2.1773,
      "step": 445
    },
    {
      "epoch": 0.2913604442266863,
      "grad_norm": 5.129368782043457,
      "learning_rate": 2.134970685536697e-08,
      "loss": 2.3766,
      "step": 446
    },
    {
      "epoch": 0.2920137187653111,
      "grad_norm": 4.547641754150391,
      "learning_rate": 1.200958404936059e-08,
      "loss": 2.0766,
      "step": 447
    },
    {
      "epoch": 0.292666993303936,
      "grad_norm": 4.891174793243408,
      "learning_rate": 5.337711625497121e-09,
      "loss": 2.1816,
      "step": 448
    },
    {
      "epoch": 0.2933202678425608,
      "grad_norm": 6.047499656677246,
      "learning_rate": 1.3344457138297906e-09,
      "loss": 2.7319,
      "step": 449
    },
    {
      "epoch": 0.29397354238118567,
      "grad_norm": 8.506380081176758,
      "learning_rate": 0.0,
      "loss": 2.4519,
      "step": 450
    },
    {
      "epoch": 0.29397354238118567,
      "eval_loss": 1.8720966577529907,
      "eval_runtime": 49.6963,
      "eval_samples_per_second": 51.875,
      "eval_steps_per_second": 12.979,
      "step": 450
    }
  ],
  "logging_steps": 1,
  "max_steps": 450,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 150,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 2,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9.230704871630438e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}