{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 1000,
  "global_step": 25000,
  "is_hyper_param_search": false,
  "is_local_process_zero": false,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004,
      "grad_norm": 0.6785533428192139,
      "learning_rate": 0.003999874817760761,
      "loss": 67.455439453125,
      "step": 100
    },
    {
      "epoch": 0.008,
      "grad_norm": 0.7726423144340515,
      "learning_rate": 0.00399943549159807,
      "loss": 60.8977392578125,
      "step": 200
    },
    {
      "epoch": 0.012,
      "grad_norm": 0.5963970422744751,
      "learning_rate": 0.003998680178657526,
      "loss": 58.8469140625,
      "step": 300
    },
    {
      "epoch": 0.016,
      "grad_norm": 0.5085615515708923,
      "learning_rate": 0.003997608998307273,
      "loss": 57.544306640625,
      "step": 400
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5642688870429993,
      "learning_rate": 0.003996222119834506,
      "loss": 56.976044921875,
      "step": 500
    },
    {
      "epoch": 0.024,
      "grad_norm": 0.47304418683052063,
      "learning_rate": 0.003994519762418718,
      "loss": 56.1641748046875,
      "step": 600
    },
    {
      "epoch": 0.028,
      "grad_norm": 0.45104825496673584,
      "learning_rate": 0.003992502195097064,
      "loss": 55.4853173828125,
      "step": 700
    },
    {
      "epoch": 0.032,
      "grad_norm": 0.5336799621582031,
      "learning_rate": 0.00399016973672184,
      "loss": 54.9941357421875,
      "step": 800
    },
    {
      "epoch": 0.036,
      "grad_norm": 0.5774248242378235,
      "learning_rate": 0.0039875227559100935,
      "loss": 54.54736328125,
      "step": 900
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7312104105949402,
      "learning_rate": 0.003984561670985366,
      "loss": 54.2763134765625,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "eval_accuracy": 0.1520635386119257,
      "eval_loss": 53.98247146606445,
      "eval_runtime": 23.3511,
      "eval_samples_per_second": 5.353,
      "eval_steps_per_second": 0.171,
      "step": 1000
    },
    {
      "epoch": 0.044,
      "grad_norm": 0.6063086986541748,
      "learning_rate": 0.003981286949911586,
      "loss": 53.98443359375,
      "step": 1100
    },
    {
      "epoch": 0.048,
      "grad_norm": 0.39948800206184387,
      "learning_rate": 0.003977699110219106,
      "loss": 53.585517578125,
      "step": 1200
    },
    {
      "epoch": 0.052,
      "grad_norm": 0.5279461145401001,
      "learning_rate": 0.0039737987189229235,
      "loss": 53.4054736328125,
      "step": 1300
    },
    {
      "epoch": 0.056,
      "grad_norm": 0.7522546648979187,
      "learning_rate": 0.00396958639243306,
      "loss": 53.0430322265625,
      "step": 1400
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7735885381698608,
      "learning_rate": 0.003965062796457152,
      "loss": 52.7428564453125,
      "step": 1500
    },
    {
      "epoch": 0.064,
      "grad_norm": 0.7539300322532654,
      "learning_rate": 0.0039602286458952415,
      "loss": 52.66658203125,
      "step": 1600
    },
    {
      "epoch": 0.068,
      "grad_norm": 0.49980393052101135,
      "learning_rate": 0.0039550847047267945,
      "loss": 52.561875,
      "step": 1700
    },
    {
      "epoch": 0.072,
      "grad_norm": 0.8233217000961304,
      "learning_rate": 0.0039496317858899645,
      "loss": 52.1762646484375,
      "step": 1800
    },
    {
      "epoch": 0.076,
      "grad_norm": 0.5626161098480225,
      "learning_rate": 0.003943870751153116,
      "loss": 51.989736328125,
      "step": 1900
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6408643126487732,
      "learning_rate": 0.003937802510978631,
      "loss": 51.8978759765625,
      "step": 2000
    },
    {
      "epoch": 0.08,
      "eval_accuracy": 0.15603128054740958,
      "eval_loss": 51.726192474365234,
      "eval_runtime": 7.2542,
      "eval_samples_per_second": 17.231,
      "eval_steps_per_second": 0.551,
      "step": 2000
    },
    {
      "epoch": 0.084,
      "grad_norm": 0.5317835807800293,
      "learning_rate": 0.0039314280243790255,
      "loss": 51.5766015625,
      "step": 2100
    },
    {
      "epoch": 0.088,
      "grad_norm": 0.7677087187767029,
      "learning_rate": 0.003924748298765388,
      "loss": 51.459482421875,
      "step": 2200
    },
    {
      "epoch": 0.092,
      "grad_norm": 0.8987520933151245,
      "learning_rate": 0.003917764389788164,
      "loss": 51.2251513671875,
      "step": 2300
    },
    {
      "epoch": 0.096,
      "grad_norm": 0.8509472608566284,
      "learning_rate": 0.003910477401170333,
      "loss": 51.086318359375,
      "step": 2400
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0043303966522217,
      "learning_rate": 0.003902888484532972,
      "loss": 51.013447265625,
      "step": 2500
    },
    {
      "epoch": 0.104,
      "grad_norm": 0.6577029228210449,
      "learning_rate": 0.0038949988392132555,
      "loss": 50.6555419921875,
      "step": 2600
    },
    {
      "epoch": 0.108,
      "grad_norm": 0.9883126020431519,
      "learning_rate": 0.0038868097120749187,
      "loss": 50.6927490234375,
      "step": 2700
    },
    {
      "epoch": 0.112,
      "grad_norm": 1.0359454154968262,
      "learning_rate": 0.003878322397311201,
      "loss": 50.8522021484375,
      "step": 2800
    },
    {
      "epoch": 0.116,
      "grad_norm": 0.8272144198417664,
      "learning_rate": 0.0038695382362403186,
      "loss": 50.349677734375,
      "step": 2900
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7374610304832458,
      "learning_rate": 0.0038604586170934807,
      "loss": 49.932939453125,
      "step": 3000
    },
    {
      "epoch": 0.12,
      "eval_accuracy": 0.16219354838709676,
      "eval_loss": 50.10387420654297,
      "eval_runtime": 7.1742,
      "eval_samples_per_second": 17.424,
      "eval_steps_per_second": 0.558,
      "step": 3000
    },
    {
      "epoch": 0.124,
      "grad_norm": 1.0479457378387451,
      "learning_rate": 0.0038510849747955015,
      "loss": 49.958212890625,
      "step": 3100
    },
    {
      "epoch": 0.128,
      "grad_norm": 1.4167745113372803,
      "learning_rate": 0.0038414187907380216,
      "loss": 49.9526025390625,
      "step": 3200
    },
    {
      "epoch": 0.132,
      "grad_norm": 1.264487385749817,
      "learning_rate": 0.0038314615925453986,
      "loss": 49.983779296875,
      "step": 3300
    },
    {
      "epoch": 0.136,
      "grad_norm": 1.5543415546417236,
      "learning_rate": 0.003821214953833277,
      "loss": 49.699599609375,
      "step": 3400
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4869952201843262,
      "learning_rate": 0.0038106804939599037,
      "loss": 49.46021484375,
      "step": 3500
    },
    {
      "epoch": 0.144,
      "grad_norm": 1.2286678552627563,
      "learning_rate": 0.003799859877770204,
      "loss": 49.3774853515625,
      "step": 3600
    },
    {
      "epoch": 0.148,
      "grad_norm": 2.0684049129486084,
      "learning_rate": 0.003788754815332674,
      "loss": 48.850126953125,
      "step": 3700
    },
    {
      "epoch": 0.152,
      "grad_norm": 2.0739598274230957,
      "learning_rate": 0.003777367061669124,
      "loss": 48.7242919921875,
      "step": 3800
    },
    {
      "epoch": 0.156,
      "grad_norm": 2.031198740005493,
      "learning_rate": 0.0037656984164773206,
      "loss": 48.9626513671875,
      "step": 3900
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9467569589614868,
      "learning_rate": 0.003753750723846564,
      "loss": 48.71037109375,
      "step": 4000
    },
    {
      "epoch": 0.16,
      "eval_accuracy": 0.1748435972629521,
      "eval_loss": 48.56401824951172,
      "eval_runtime": 7.0573,
      "eval_samples_per_second": 17.712,
      "eval_steps_per_second": 0.567,
      "step": 4000
    },
    {
      "epoch": 0.164,
      "grad_norm": 2.0323381423950195,
      "learning_rate": 0.0037415258719662517,
      "loss": 48.445341796875,
      "step": 4100
    },
    {
      "epoch": 0.168,
      "grad_norm": 2.1837387084960938,
      "learning_rate": 0.003729025792827473,
      "loss": 48.2243603515625,
      "step": 4200
    },
    {
      "epoch": 0.172,
      "grad_norm": 3.023264169692993,
      "learning_rate": 0.0037162524619176844,
      "loss": 48.10587890625,
      "step": 4300
    },
    {
      "epoch": 0.176,
      "grad_norm": 2.3792617321014404,
      "learning_rate": 0.0037032078979085024,
      "loss": 48.301064453125,
      "step": 4400
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.003861665725708,
      "learning_rate": 0.0036898941623366784,
      "loss": 47.9244140625,
      "step": 4500
    },
    {
      "epoch": 0.184,
      "grad_norm": 2.263864517211914,
      "learning_rate": 0.003676313359278299,
      "loss": 47.8005126953125,
      "step": 4600
    },
    {
      "epoch": 0.188,
      "grad_norm": 2.5262391567230225,
      "learning_rate": 0.0036624676350162613,
      "loss": 47.87388671875,
      "step": 4700
    },
    {
      "epoch": 0.192,
      "grad_norm": 2.9222702980041504,
      "learning_rate": 0.0036483591777010794,
      "loss": 47.5597412109375,
      "step": 4800
    },
    {
      "epoch": 0.196,
      "grad_norm": 2.4583094120025635,
      "learning_rate": 0.0036339902170050703,
      "loss": 47.228359375,
      "step": 4900
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.630126953125,
      "learning_rate": 0.0036193630237699843,
      "loss": 47.68759765625,
      "step": 5000
    },
    {
      "epoch": 0.2,
      "eval_accuracy": 0.1841290322580645,
      "eval_loss": 47.29731750488281,
      "eval_runtime": 7.2208,
      "eval_samples_per_second": 17.311,
      "eval_steps_per_second": 0.554,
      "step": 5000
    },
    {
      "epoch": 0.204,
      "grad_norm": 2.726828098297119,
      "learning_rate": 0.003604479909648125,
      "loss": 47.3409326171875,
      "step": 5100
    },
    {
      "epoch": 0.208,
      "grad_norm": 2.4532039165496826,
      "learning_rate": 0.00358934322673702,
      "loss": 46.9724267578125,
      "step": 5200
    },
    {
      "epoch": 0.212,
      "grad_norm": 3.2446208000183105,
      "learning_rate": 0.003573955367207699,
      "loss": 47.0882275390625,
      "step": 5300
    },
    {
      "epoch": 0.216,
      "grad_norm": 3.021744966506958,
      "learning_rate": 0.003558318762926643,
      "loss": 46.8798291015625,
      "step": 5400
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.569321870803833,
      "learning_rate": 0.003542435885071452,
      "loss": 46.89576171875,
      "step": 5500
    },
    {
      "epoch": 0.224,
      "grad_norm": 2.7224721908569336,
      "learning_rate": 0.0035263092437403123,
      "loss": 46.96203125,
      "step": 5600
    },
    {
      "epoch": 0.228,
      "grad_norm": 2.9060356616973877,
      "learning_rate": 0.003509941387555298,
      "loss": 46.441123046875,
      "step": 5700
    },
    {
      "epoch": 0.232,
      "grad_norm": 3.4122211933135986,
      "learning_rate": 0.0034933349032595963,
      "loss": 46.44693359375,
      "step": 5800
    },
    {
      "epoch": 0.236,
      "grad_norm": 2.924081802368164,
      "learning_rate": 0.0034764924153087035,
      "loss": 46.6074658203125,
      "step": 5900
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.3687922954559326,
      "learning_rate": 0.003459416585455659,
      "loss": 46.527548828125,
      "step": 6000
    },
    {
      "epoch": 0.24,
      "eval_accuracy": 0.19418279569892474,
      "eval_loss": 46.334205627441406,
      "eval_runtime": 7.5021,
      "eval_samples_per_second": 16.662,
      "eval_steps_per_second": 0.533,
      "step": 6000
    },
    {
      "epoch": 0.244,
      "grad_norm": 3.177504062652588,
      "learning_rate": 0.0034421101123303897,
      "loss": 46.474375,
      "step": 6100
    },
    {
      "epoch": 0.248,
      "grad_norm": 3.0214405059814453,
      "learning_rate": 0.0034245757310132244,
      "loss": 46.3790576171875,
      "step": 6200
    },
    {
      "epoch": 0.252,
      "grad_norm": 2.933534860610962,
      "learning_rate": 0.003406816212602642,
      "loss": 46.0579736328125,
      "step": 6300
    },
    {
      "epoch": 0.256,
      "grad_norm": 2.5542778968811035,
      "learning_rate": 0.003388834363777341,
      "loss": 46.0082861328125,
      "step": 6400
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.1228818893432617,
      "learning_rate": 0.0033706330263526692,
      "loss": 46.218310546875,
      "step": 6500
    },
    {
      "epoch": 0.264,
      "grad_norm": 3.096647262573242,
      "learning_rate": 0.003352215076831515,
      "loss": 46.06939453125,
      "step": 6600
    },
    {
      "epoch": 0.268,
      "grad_norm": 3.2484421730041504,
      "learning_rate": 0.0033335834259497076,
      "loss": 46.158388671875,
      "step": 6700
    },
    {
      "epoch": 0.272,
      "grad_norm": 3.035205841064453,
      "learning_rate": 0.003314741018216011,
      "loss": 45.9556396484375,
      "step": 6800
    },
    {
      "epoch": 0.276,
      "grad_norm": 3.4140381813049316,
      "learning_rate": 0.0032956908314467803,
      "loss": 45.8059375,
      "step": 6900
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.5898549556732178,
      "learning_rate": 0.003276435876295352,
      "loss": 45.785087890625,
      "step": 7000
    },
    {
      "epoch": 0.28,
      "eval_accuracy": 0.2004613880742913,
      "eval_loss": 45.5724983215332,
      "eval_runtime": 7.1637,
      "eval_samples_per_second": 17.449,
      "eval_steps_per_second": 0.558,
      "step": 7000
    },
    {
      "epoch": 0.284,
      "grad_norm": 3.6645097732543945,
      "learning_rate": 0.0032569791957762473,
      "loss": 45.450126953125,
      "step": 7100
    },
    {
      "epoch": 0.288,
      "grad_norm": 3.449202060699463,
      "learning_rate": 0.003237323864784261,
      "loss": 45.6815869140625,
      "step": 7200
    },
    {
      "epoch": 0.292,
      "grad_norm": 3.7189524173736572,
      "learning_rate": 0.0032174729896085105,
      "loss": 45.68208984375,
      "step": 7300
    },
    {
      "epoch": 0.296,
      "grad_norm": 3.6789443492889404,
      "learning_rate": 0.0031974297074415237,
      "loss": 45.42337890625,
      "step": 7400
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.174764633178711,
      "learning_rate": 0.003177197185883444,
      "loss": 45.540263671875,
      "step": 7500
    },
    {
      "epoch": 0.304,
      "grad_norm": 3.5856988430023193,
      "learning_rate": 0.0031567786224414272,
      "loss": 45.4182177734375,
      "step": 7600
    },
    {
      "epoch": 0.308,
      "grad_norm": 3.8331663608551025,
      "learning_rate": 0.003136177244024319,
      "loss": 45.5009228515625,
      "step": 7700
    },
    {
      "epoch": 0.312,
      "grad_norm": 3.373854160308838,
      "learning_rate": 0.003115396306432675,
      "loss": 45.4340283203125,
      "step": 7800
    },
    {
      "epoch": 0.316,
      "grad_norm": 3.256587028503418,
      "learning_rate": 0.003094439093844223,
      "loss": 45.136494140625,
      "step": 7900
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.5598511695861816,
      "learning_rate": 0.0030733089182948376,
      "loss": 45.0095849609375,
      "step": 8000
    },
    {
      "epoch": 0.32,
      "eval_accuracy": 0.2007086999022483,
      "eval_loss": 45.0313720703125,
      "eval_runtime": 6.9806,
      "eval_samples_per_second": 17.907,
      "eval_steps_per_second": 0.573,
      "step": 8000
    },
    {
      "epoch": 0.324,
      "grad_norm": 3.4113783836364746,
      "learning_rate": 0.0030520091191551142,
      "loss": 45.00322265625,
      "step": 8100
    },
    {
      "epoch": 0.328,
      "grad_norm": 3.566800832748413,
      "learning_rate": 0.003030543062602622,
      "loss": 45.1212841796875,
      "step": 8200
    },
    {
      "epoch": 0.332,
      "grad_norm": 3.9237873554229736,
      "learning_rate": 0.003008914141089914,
      "loss": 45.1877978515625,
      "step": 8300
    },
    {
      "epoch": 0.336,
      "grad_norm": 4.103876113891602,
      "learning_rate": 0.0029871257728083995,
      "loss": 45.1942626953125,
      "step": 8400
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.259410858154297,
      "learning_rate": 0.0029651814011481333,
      "loss": 44.929365234375,
      "step": 8500
    },
    {
      "epoch": 0.344,
      "grad_norm": 3.518616199493408,
      "learning_rate": 0.002943084494153631,
      "loss": 44.5111328125,
      "step": 8600
    },
    {
      "epoch": 0.348,
      "grad_norm": 3.544551372528076,
      "learning_rate": 0.002920838543975789,
      "loss": 44.8691015625,
      "step": 8700
    },
    {
      "epoch": 0.352,
      "grad_norm": 3.5976264476776123,
      "learning_rate": 0.0028984470663199887,
      "loss": 44.924892578125,
      "step": 8800
    },
    {
      "epoch": 0.356,
      "grad_norm": 3.3831348419189453,
      "learning_rate": 0.0028759135998904796,
      "loss": 44.773828125,
      "step": 8900
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.3760013580322266,
      "learning_rate": 0.002853241705831138,
      "loss": 44.8477294921875,
      "step": 9000
    },
    {
      "epoch": 0.36,
      "eval_accuracy": 0.2051945259042033,
      "eval_loss": 44.57013702392578,
      "eval_runtime": 7.2683,
      "eval_samples_per_second": 17.198,
      "eval_steps_per_second": 0.55,
      "step": 9000
    },
    {
      "epoch": 0.364,
      "grad_norm": 2.7893764972686768,
      "learning_rate": 0.0028304349671626613,
      "loss": 44.7609765625,
      "step": 9100
    },
    {
      "epoch": 0.368,
      "grad_norm": 3.8668603897094727,
      "learning_rate": 0.002807496988216319,
      "loss": 44.6481201171875,
      "step": 9200
    },
    {
      "epoch": 0.372,
      "grad_norm": 3.4443302154541016,
      "learning_rate": 0.002784431394064333,
      "loss": 44.4382373046875,
      "step": 9300
    },
    {
      "epoch": 0.376,
      "grad_norm": 3.2611913681030273,
      "learning_rate": 0.0027612418299469746,
      "loss": 44.3794384765625,
      "step": 9400
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.5774736404418945,
      "learning_rate": 0.0027379319606964806,
      "loss": 44.4774462890625,
      "step": 9500
    },
    {
      "epoch": 0.384,
      "grad_norm": 3.2907772064208984,
      "learning_rate": 0.002714505470157871,
      "loss": 44.4849267578125,
      "step": 9600
    },
    {
      "epoch": 0.388,
      "grad_norm": 3.4475579261779785,
      "learning_rate": 0.0026909660606067583,
      "loss": 44.4125537109375,
      "step": 9700
    },
    {
      "epoch": 0.392,
      "grad_norm": 4.186856269836426,
      "learning_rate": 0.002667317452164251,
      "loss": 44.404677734375,
      "step": 9800
    },
    {
      "epoch": 0.396,
      "grad_norm": 3.1886963844299316,
      "learning_rate": 0.0026435633822090316,
      "loss": 44.2500146484375,
      "step": 9900
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.3009133338928223,
      "learning_rate": 0.002619707604786708,
      "loss": 44.2538232421875,
      "step": 10000
    },
    {
      "epoch": 0.4,
      "eval_accuracy": 0.20702737047898337,
      "eval_loss": 44.112098693847656,
      "eval_runtime": 7.3398,
      "eval_samples_per_second": 17.03,
      "eval_steps_per_second": 0.545,
      "step": 10000
    },
    {
      "epoch": 0.404,
      "grad_norm": 3.5040667057037354,
      "learning_rate": 0.0025957538900165333,
      "loss": 44.2118505859375,
      "step": 10100
    },
    {
      "epoch": 0.408,
      "grad_norm": 3.0154404640197754,
      "learning_rate": 0.0025717060234955805,
      "loss": 44.339365234375,
      "step": 10200
    },
    {
      "epoch": 0.412,
      "grad_norm": 3.0024242401123047,
      "learning_rate": 0.0025475678057004796,
      "loss": 44.2085009765625,
      "step": 10300
    },
    {
      "epoch": 0.416,
      "grad_norm": 2.868852138519287,
      "learning_rate": 0.002523343051386793,
      "loss": 44.1264453125,
      "step": 10400
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.098557472229004,
      "learning_rate": 0.0024990355889861417,
      "loss": 44.0060986328125,
      "step": 10500
    },
    {
      "epoch": 0.424,
      "grad_norm": 2.94388747215271,
      "learning_rate": 0.0024746492600011666,
      "loss": 44.090751953125,
      "step": 10600
    },
    {
      "epoch": 0.428,
      "grad_norm": 3.172335147857666,
      "learning_rate": 0.002450187918398426,
      "loss": 44.108330078125,
      "step": 10700
    },
    {
      "epoch": 0.432,
      "grad_norm": 3.123621702194214,
      "learning_rate": 0.0024256554299993214,
      "loss": 44.00998046875,
      "step": 10800
    },
    {
      "epoch": 0.436,
      "grad_norm": 2.8541388511657715,
      "learning_rate": 0.002401055671869152,
      "loss": 43.941904296875,
      "step": 10900
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.895085573196411,
      "learning_rate": 0.0023763925317043903,
      "loss": 43.792197265625,
      "step": 11000
    },
    {
      "epoch": 0.44,
      "eval_accuracy": 0.20995894428152492,
      "eval_loss": 43.750606536865234,
      "eval_runtime": 6.9444,
      "eval_samples_per_second": 18.0,
      "eval_steps_per_second": 0.576,
      "step": 11000
    },
    {
      "epoch": 0.444,
      "grad_norm": 2.346236228942871,
      "learning_rate": 0.0023516699072182786,
      "loss": 43.9089990234375,
      "step": 11100
    },
    {
      "epoch": 0.448,
      "grad_norm": 2.3568971157073975,
      "learning_rate": 0.002326891705524841,
      "loss": 43.76560546875,
      "step": 11200
    },
    {
      "epoch": 0.452,
      "grad_norm": 3.0081124305725098,
      "learning_rate": 0.0023020618425214144,
      "loss": 43.808974609375,
      "step": 11300
    },
    {
      "epoch": 0.456,
      "grad_norm": 3.7043468952178955,
      "learning_rate": 0.0022771842422697835,
      "loss": 43.849228515625,
      "step": 11400
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.2964367866516113,
      "learning_rate": 0.0022522628363760323,
      "loss": 43.599345703125,
      "step": 11500
    },
    {
      "epoch": 0.464,
      "grad_norm": 2.747478485107422,
      "learning_rate": 0.002227301563369202,
      "loss": 43.42494140625,
      "step": 11600
    },
    {
      "epoch": 0.468,
      "grad_norm": 2.878131866455078,
      "learning_rate": 0.0022023043680788512,
      "loss": 43.6928955078125,
      "step": 11700
    },
    {
      "epoch": 0.472,
      "grad_norm": 2.552713632583618,
      "learning_rate": 0.0021772752010116247,
      "loss": 43.6564453125,
      "step": 11800
    },
    {
      "epoch": 0.476,
      "grad_norm": 2.9993340969085693,
      "learning_rate": 0.002152218017726923,
      "loss": 43.607763671875,
      "step": 11900
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.911576747894287,
      "learning_rate": 0.002127136778211772,
      "loss": 43.8162890625,
      "step": 12000
    },
    {
      "epoch": 0.48,
      "eval_accuracy": 0.21305474095796675,
      "eval_loss": 43.4426155090332,
      "eval_runtime": 7.5722,
      "eval_samples_per_second": 16.508,
      "eval_steps_per_second": 0.528,
      "step": 12000
    },
    {
      "epoch": 0.484,
      "grad_norm": 3.012925148010254,
      "learning_rate": 0.0021020354462549986,
      "loss": 43.3377099609375,
      "step": 12100
    },
    {
      "epoch": 0.488,
      "grad_norm": 3.428558826446533,
      "learning_rate": 0.0020769179888207967,
      "loss": 43.5579638671875,
      "step": 12200
    },
    {
      "epoch": 0.492,
      "grad_norm": 3.1122782230377197,
      "learning_rate": 0.0020517883754218,
      "loss": 43.4773828125,
      "step": 12300
    },
    {
      "epoch": 0.496,
      "grad_norm": 2.5048506259918213,
      "learning_rate": 0.0020266505774917446,
      "loss": 43.3037744140625,
      "step": 12400
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.638044595718384,
      "learning_rate": 0.0020015085677578355,
      "loss": 43.562880859375,
      "step": 12500
    },
    {
      "epoch": 0.504,
      "grad_norm": 2.780961036682129,
      "learning_rate": 0.0019763663196129006,
      "loss": 43.3517138671875,
      "step": 12600
    },
    {
      "epoch": 0.508,
      "grad_norm": 1.9473482370376587,
      "learning_rate": 0.0019512278064874483,
      "loss": 43.109091796875,
      "step": 12700
    },
    {
      "epoch": 0.512,
      "grad_norm": 3.174790143966675,
      "learning_rate": 0.0019260970012217105,
      "loss": 43.475107421875,
      "step": 12800
    },
    {
      "epoch": 0.516,
      "grad_norm": 2.280482769012451,
      "learning_rate": 0.001900977875437784,
      "loss": 43.4238671875,
      "step": 12900
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.096158266067505,
      "learning_rate": 0.0018758743989119647,
      "loss": 43.194453125,
      "step": 13000
    },
    {
      "epoch": 0.52,
      "eval_accuracy": 0.21564809384164224,
      "eval_loss": 43.14116668701172,
      "eval_runtime": 6.8523,
      "eval_samples_per_second": 18.242,
      "eval_steps_per_second": 0.584,
      "step": 13000
    },
    {
      "epoch": 0.524,
      "grad_norm": 2.7369933128356934,
      "learning_rate": 0.0018507905389473706,
      "loss": 43.233857421875,
      "step": 13100
    },
    {
      "epoch": 0.528,
      "grad_norm": 2.2296907901763916,
      "learning_rate": 0.001825730259746958,
      "loss": 43.15728515625,
      "step": 13200
    },
    {
      "epoch": 0.532,
      "grad_norm": 3.018181324005127,
      "learning_rate": 0.0018006975217870259,
      "loss": 43.116005859375,
      "step": 13300
    },
    {
      "epoch": 0.536,
      "grad_norm": 2.055466890335083,
      "learning_rate": 0.0017756962811913122,
      "loss": 43.1440869140625,
      "step": 13400
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.277271032333374,
      "learning_rate": 0.0017507304891057722,
      "loss": 43.4723193359375,
      "step": 13500
    },
    {
      "epoch": 0.544,
      "grad_norm": 2.3194973468780518,
      "learning_rate": 0.001725804091074152,
      "loss": 43.233828125,
      "step": 13600
    },
    {
      "epoch": 0.548,
      "grad_norm": 2.0250515937805176,
      "learning_rate": 0.0017009210264144388,
      "loss": 43.008115234375,
      "step": 13700
    },
    {
      "epoch": 0.552,
      "grad_norm": 2.4250712394714355,
      "learning_rate": 0.0016760852275963015,
      "loss": 42.6800048828125,
      "step": 13800
    },
    {
      "epoch": 0.556,
      "grad_norm": 1.7238541841506958,
      "learning_rate": 0.0016513006196196098,
      "loss": 42.93923828125,
      "step": 13900
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7167351245880127,
      "learning_rate": 0.0016265711193941361,
      "loss": 43.267294921875,
      "step": 14000
    },
    {
      "epoch": 0.56,
      "eval_accuracy": 0.2188279569892473,
      "eval_loss": 42.8851318359375,
      "eval_runtime": 7.0878,
      "eval_samples_per_second": 17.636,
      "eval_steps_per_second": 0.564,
      "step": 14000
    },
    {
      "epoch": 0.564,
      "grad_norm": 2.029331684112549,
      "learning_rate": 0.0016019006351205318,
      "loss": 43.2043701171875,
      "step": 14100
    },
    {
      "epoch": 0.568,
      "grad_norm": 2.470975399017334,
      "learning_rate": 0.0015772930656726884,
      "loss": 43.1994921875,
      "step": 14200
    },
    {
      "epoch": 0.572,
      "grad_norm": 1.9896589517593384,
      "learning_rate": 0.0015527522999815634,
      "loss": 42.8731396484375,
      "step": 14300
    },
    {
      "epoch": 0.576,
      "grad_norm": 2.9005286693573,
      "learning_rate": 0.001528282216420582,
      "loss": 42.8190380859375,
      "step": 14400
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.300215244293213,
      "learning_rate": 0.0015038866821927073,
      "loss": 42.9506982421875,
      "step": 14500
    },
    {
      "epoch": 0.584,
      "grad_norm": 1.927196979522705,
      "learning_rate": 0.0014795695527192762,
      "loss": 42.97125,
      "step": 14600
    },
    {
      "epoch": 0.588,
      "grad_norm": 1.9869160652160645,
      "learning_rate": 0.001455334671030695,
      "loss": 43.0368505859375,
      "step": 14700
    },
    {
      "epoch": 0.592,
      "grad_norm": 2.05895733833313,
      "learning_rate": 0.0014311858671590934,
      "loss": 42.886181640625,
      "step": 14800
    },
    {
      "epoch": 0.596,
      "grad_norm": 2.5150656700134277,
      "learning_rate": 0.0014071269575330373,
      "loss": 42.9019677734375,
      "step": 14900
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.3601467609405518,
      "learning_rate": 0.0013831617443743852,
      "loss": 42.6784130859375,
      "step": 15000
    },
    {
      "epoch": 0.6,
      "eval_accuracy": 0.22081524926686216,
      "eval_loss": 42.696502685546875,
      "eval_runtime": 7.2547,
      "eval_samples_per_second": 17.23,
      "eval_steps_per_second": 0.551,
      "step": 15000
    },
    {
      "epoch": 0.604,
      "grad_norm": 2.1606075763702393,
      "learning_rate": 0.0013592940150973943,
      "loss": 42.9130859375,
      "step": 15100
    },
    {
      "epoch": 0.608,
      "grad_norm": 2.021022319793701,
      "learning_rate": 0.0013355275417101637,
      "loss": 42.95048828125,
      "step": 15200
    },
    {
      "epoch": 0.612,
      "grad_norm": 2.6041364669799805,
      "learning_rate": 0.0013118660802185161,
      "loss": 42.6790380859375,
      "step": 15300
    },
    {
      "epoch": 0.616,
      "grad_norm": 2.121518135070801,
      "learning_rate": 0.0012883133700324026,
      "loss": 42.72169921875,
      "step": 15400
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0749258995056152,
      "learning_rate": 0.0012648731333749373,
      "loss": 42.6244970703125,
      "step": 15500
    },
    {
      "epoch": 0.624,
      "grad_norm": 2.3933210372924805,
      "learning_rate": 0.0012415490746941434,
      "loss": 42.629208984375,
      "step": 15600
    },
    {
      "epoch": 0.628,
      "grad_norm": 2.163004159927368,
      "learning_rate": 0.0012183448800775084,
      "loss": 42.9462548828125,
      "step": 15700
    },
    {
      "epoch": 0.632,
      "grad_norm": 1.9806509017944336,
      "learning_rate": 0.0011952642166694445,
      "loss": 42.9928662109375,
      "step": 15800
    },
    {
      "epoch": 0.636,
      "grad_norm": 2.2631726264953613,
      "learning_rate": 0.0011723107320917396,
      "loss": 42.7688671875,
      "step": 15900
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2971158027648926,
      "learning_rate": 0.0011494880538670915,
      "loss": 42.37572265625,
      "step": 16000
    },
    {
      "epoch": 0.64,
      "eval_accuracy": 0.2221368523949169,
      "eval_loss": 42.538299560546875,
      "eval_runtime": 7.2739,
      "eval_samples_per_second": 17.185,
      "eval_steps_per_second": 0.55,
      "step": 16000
    },
    {
      "epoch": 0.644,
      "grad_norm": 1.2762759923934937,
      "learning_rate": 0.001126799788845827,
      "loss": 42.4043359375,
      "step": 16100
    },
    {
      "epoch": 0.648,
      "grad_norm": 1.4422513246536255,
      "learning_rate": 0.0011042495226358784,
      "loss": 43.033349609375,
      "step": 16200
    },
    {
      "epoch": 0.652,
      "grad_norm": 1.816651463508606,
      "learning_rate": 0.0010818408190361227,
      "loss": 42.873017578125,
      "step": 16300
    },
    {
      "epoch": 0.656,
      "grad_norm": 2.664728879928589,
      "learning_rate": 0.0010595772194731657,
      "loss": 42.72642578125,
      "step": 16400
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.1422553062438965,
      "learning_rate": 0.0010374622424416619,
      "loss": 42.725224609375,
      "step": 16500
    },
    {
      "epoch": 0.664,
      "grad_norm": 2.044825315475464,
      "learning_rate": 0.0010154993829482593,
      "loss": 42.5387841796875,
      "step": 16600
    },
    {
      "epoch": 0.668,
      "grad_norm": 1.400777816772461,
      "learning_rate": 0.0009936921119592535,
      "loss": 42.594912109375,
      "step": 16700
    },
    {
      "epoch": 0.672,
      "grad_norm": 1.5564935207366943,
      "learning_rate": 0.0009720438758520462,
      "loss": 42.6428369140625,
      "step": 16800
    },
    {
      "epoch": 0.676,
      "grad_norm": 2.554816961288452,
      "learning_rate": 0.0009505580958704852,
      "loss": 42.628232421875,
      "step": 16900
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2824437618255615,
      "learning_rate": 0.0009292381675841768,
      "loss": 42.5581103515625,
      "step": 17000
    },
    {
      "epoch": 0.68,
      "eval_accuracy": 0.22342913000977518,
      "eval_loss": 42.39310836791992,
      "eval_runtime": 7.1782,
      "eval_samples_per_second": 17.414,
      "eval_steps_per_second": 0.557,
      "step": 17000
    },
    {
      "epoch": 0.684,
      "grad_norm": 1.755340576171875,
      "learning_rate": 0.0009080874603518585,
      "loss": 42.472861328125,
      "step": 17100
    },
    {
      "epoch": 0.688,
      "grad_norm": 1.1157244443893433,
      "learning_rate": 0.0008871093167889121,
      "loss": 42.5861279296875,
      "step": 17200
    },
    {
      "epoch": 0.692,
      "grad_norm": 1.2528804540634155,
      "learning_rate": 0.0008663070522391008,
      "loss": 42.4558740234375,
      "step": 17300
    },
    {
      "epoch": 0.696,
      "grad_norm": 1.7521710395812988,
      "learning_rate": 0.0008456839542506229,
      "loss": 42.712470703125,
      "step": 17400
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.830307960510254,
      "learning_rate": 0.0008252432820565525,
      "loss": 42.67123046875,
      "step": 17500
    },
    {
      "epoch": 0.704,
      "grad_norm": 1.6736066341400146,
      "learning_rate": 0.0008049882660597558,
      "loss": 42.42759765625,
      "step": 17600
    },
    {
      "epoch": 0.708,
      "grad_norm": 1.5861881971359253,
      "learning_rate": 0.0007849221073223665,
      "loss": 42.5102734375,
      "step": 17700
    },
    {
      "epoch": 0.712,
      "grad_norm": 1.659502387046814,
      "learning_rate": 0.0007650479770598948,
      "loss": 42.2067578125,
      "step": 17800
    },
    {
      "epoch": 0.716,
      "grad_norm": 1.2964304685592651,
      "learning_rate": 0.0007453690161400562,
      "loss": 42.520947265625,
      "step": 17900
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7349175214767456,
      "learning_rate": 0.000725888334586394,
      "loss": 42.7132568359375,
      "step": 18000
    },
    {
      "epoch": 0.72,
      "eval_accuracy": 0.224316715542522,
      "eval_loss": 42.29633331298828,
      "eval_runtime": 7.2524,
      "eval_samples_per_second": 17.236,
      "eval_steps_per_second": 0.552,
      "step": 18000
    },
    {
      "epoch": 0.724,
      "grad_norm": 1.7778606414794922,
      "learning_rate": 0.0007066090110867782,
      "loss": 42.758095703125,
      "step": 18100
    },
    {
      "epoch": 0.728,
      "grad_norm": 1.6367878913879395,
      "learning_rate": 0.0006875340925068554,
      "loss": 42.4748291015625,
      "step": 18200
    },
    {
      "epoch": 0.732,
      "grad_norm": 1.4886348247528076,
      "learning_rate": 0.0006686665934085281,
      "loss": 42.22041015625,
      "step": 18300
    },
    {
      "epoch": 0.736,
      "grad_norm": 1.0366780757904053,
      "learning_rate": 0.0006500094955735401,
      "loss": 42.1452880859375,
      "step": 18400
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.189237117767334,
      "learning_rate": 0.0006315657475322411,
      "loss": 42.422060546875,
      "step": 18500
    },
    {
      "epoch": 0.744,
      "grad_norm": 1.5588538646697998,
      "learning_rate": 0.0006133382640976061,
      "loss": 42.59294921875,
      "step": 18600
    },
    {
      "epoch": 0.748,
      "grad_norm": 1.2961610555648804,
      "learning_rate": 0.0005953299259045865,
      "loss": 42.5683740234375,
      "step": 18700
    },
    {
      "epoch": 0.752,
      "grad_norm": 1.513845682144165,
      "learning_rate": 0.000577543578954858,
      "loss": 42.31896484375,
      "step": 18800
    },
    {
      "epoch": 0.756,
      "grad_norm": 1.0572335720062256,
      "learning_rate": 0.0005599820341670454,
      "loss": 42.3233642578125,
      "step": 18900
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4302053451538086,
      "learning_rate": 0.0005426480669324906,
      "loss": 42.2752880859375,
      "step": 19000
    },
    {
      "epoch": 0.76,
      "eval_accuracy": 0.22522482893450635,
      "eval_loss": 42.213226318359375,
      "eval_runtime": 7.0911,
      "eval_samples_per_second": 17.628,
      "eval_steps_per_second": 0.564,
      "step": 19000
    },
    {
      "epoch": 0.764,
      "grad_norm": 1.4170910120010376,
      "learning_rate": 0.0005255444166766354,
      "loss": 42.50333984375,
      "step": 19100
    },
    {
      "epoch": 0.768,
      "grad_norm": 1.3892245292663574,
      "learning_rate": 0.0005086737864260862,
      "loss": 42.474443359375,
      "step": 19200
    },
    {
      "epoch": 0.772,
      "grad_norm": 1.5735739469528198,
      "learning_rate": 0.0004920388423814364,
      "loss": 42.25490234375,
      "step": 19300
    },
    {
      "epoch": 0.776,
      "grad_norm": 1.2609901428222656,
      "learning_rate": 0.0004756422134959031,
      "loss": 42.3439208984375,
      "step": 19400
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1414227485656738,
      "learning_rate": 0.00045948649105985397,
      "loss": 42.31787109375,
      "step": 19500
    },
    {
      "epoch": 0.784,
      "grad_norm": 1.301095962524414,
      "learning_rate": 0.0004435742282912836,
      "loss": 42.43177734375,
      "step": 19600
    },
    {
      "epoch": 0.788,
      "grad_norm": 1.1665655374526978,
      "learning_rate": 0.0004279079399323087,
      "loss": 42.386767578125,
      "step": 19700
    },
    {
      "epoch": 0.792,
      "grad_norm": 1.5682238340377808,
      "learning_rate": 0.0004124901018517433,
      "loss": 42.2175439453125,
      "step": 19800
    },
    {
      "epoch": 0.796,
      "grad_norm": 1.259811520576477,
      "learning_rate": 0.00039732315065381754,
      "loss": 42.252421875,
      "step": 19900
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8927878141403198,
      "learning_rate": 0.00038240948329310156,
      "loss": 42.2313232421875,
      "step": 20000
    },
    {
      "epoch": 0.8,
      "eval_accuracy": 0.2259481915933529,
      "eval_loss": 42.14704895019531,
      "eval_runtime": 6.8564,
      "eval_samples_per_second": 18.231,
      "eval_steps_per_second": 0.583,
      "step": 20000
    },
    {
      "epoch": 0.804,
      "grad_norm": 1.055684208869934,
      "learning_rate": 0.00036775145669569497,
      "loss": 42.2956591796875,
      "step": 20100
    },
    {
      "epoch": 0.808,
      "grad_norm": 1.0924086570739746,
      "learning_rate": 0.0003533513873867442,
      "loss": 42.5350732421875,
      "step": 20200
    },
    {
      "epoch": 0.812,
      "grad_norm": 1.0316708087921143,
      "learning_rate": 0.0003392115511243421,
      "loss": 42.4560107421875,
      "step": 20300
    },
    {
      "epoch": 0.816,
      "grad_norm": 1.1133615970611572,
      "learning_rate": 0.00032533418253987323,
      "loss": 42.3101513671875,
      "step": 20400
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.103115200996399,
      "learning_rate": 0.00031172147478485514,
      "loss": 42.077275390625,
      "step": 20500
    },
    {
      "epoch": 0.824,
      "grad_norm": 1.1964024305343628,
      "learning_rate": 0.00029837557918433946,
      "loss": 42.031953125,
      "step": 20600
    },
    {
      "epoch": 0.828,
      "grad_norm": 1.4636120796203613,
      "learning_rate": 0.00028529860489691903,
      "loss": 42.072587890625,
      "step": 20700
    },
    {
      "epoch": 0.832,
      "grad_norm": 1.1224356889724731,
      "learning_rate": 0.00027249261858140186,
      "loss": 42.538427734375,
      "step": 20800
    },
    {
      "epoch": 0.836,
      "grad_norm": 0.7813518643379211,
      "learning_rate": 0.00025995964407019923,
      "loss": 42.4304736328125,
      "step": 20900
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8456737399101257,
      "learning_rate": 0.0002477016620494856,
      "loss": 42.2731982421875,
      "step": 21000
    },
    {
      "epoch": 0.84,
      "eval_accuracy": 0.22656598240469208,
      "eval_loss": 42.09526824951172,
      "eval_runtime": 7.2222,
      "eval_samples_per_second": 17.308,
      "eval_steps_per_second": 0.554,
      "step": 21000
    },
    {
      "epoch": 0.844,
      "grad_norm": 1.4627004861831665,
      "learning_rate": 0.00023572060974617082,
      "loss": 42.2644970703125,
      "step": 21100
    },
    {
      "epoch": 0.848,
      "grad_norm": 1.2561938762664795,
      "learning_rate": 0.0002240183806217493,
      "loss": 42.0703955078125,
      "step": 21200
    },
    {
      "epoch": 0.852,
      "grad_norm": 1.2216907739639282,
      "learning_rate": 0.00021259682407305847,
      "loss": 42.2020556640625,
      "step": 21300
    },
    {
      "epoch": 0.856,
      "grad_norm": 0.8914040923118591,
      "learning_rate": 0.00020145774514000438,
      "loss": 42.357060546875,
      "step": 21400
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1476701498031616,
      "learning_rate": 0.00019060290422029657,
      "loss": 42.13251953125,
      "step": 21500
    },
    {
      "epoch": 0.864,
      "grad_norm": 1.119214653968811,
      "learning_rate": 0.00018003401679123887,
      "loss": 42.2200732421875,
      "step": 21600
    },
    {
      "epoch": 0.868,
      "grad_norm": 0.9182925820350647,
      "learning_rate": 0.0001697527531386196,
      "loss": 42.22140625,
      "step": 21700
    },
    {
      "epoch": 0.872,
      "grad_norm": 1.2138001918792725,
      "learning_rate": 0.00015976073809273973,
      "loss": 42.254619140625,
      "step": 21800
    },
    {
      "epoch": 0.876,
      "grad_norm": 1.321160078048706,
      "learning_rate": 0.00015005955077163137,
      "loss": 42.439716796875,
      "step": 21900
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8333455920219421,
      "learning_rate": 0.00014065072433149607,
      "loss": 42.4203271484375,
      "step": 22000
    },
    {
      "epoch": 0.88,
      "eval_accuracy": 0.2271329423264907,
      "eval_loss": 42.06480407714844,
      "eval_runtime": 6.8562,
      "eval_samples_per_second": 18.232,
      "eval_steps_per_second": 0.583,
      "step": 22000
    },
    {
      "epoch": 0.884,
      "grad_norm": 1.0895456075668335,
      "learning_rate": 0.0001315357457244073,
      "loss": 41.9967822265625,
      "step": 22100
    },
    {
      "epoch": 0.888,
      "grad_norm": 0.9787936210632324,
      "learning_rate": 0.000122716055463316,
      "loss": 41.919248046875,
      "step": 22200
    },
    {
      "epoch": 0.892,
      "grad_norm": 0.8595688343048096,
      "learning_rate": 0.00011419304739439529,
      "loss": 42.098642578125,
      "step": 22300
    },
    {
      "epoch": 0.896,
      "grad_norm": 0.8901764154434204,
      "learning_rate": 0.00010596806847675766,
      "loss": 42.3766357421875,
      "step": 22400
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9945716261863708,
      "learning_rate": 9.80424185695874e-05,
      "loss": 42.459091796875,
      "step": 22500
    },
    {
      "epoch": 0.904,
      "grad_norm": 0.9749753475189209,
      "learning_rate": 9.041735022671139e-05,
      "loss": 42.227529296875,
      "step": 22600
    },
    {
      "epoch": 0.908,
      "grad_norm": 0.8896678686141968,
      "learning_rate": 8.309406849864854e-05,
      "loss": 41.9632080078125,
      "step": 22700
    },
    {
      "epoch": 0.912,
      "grad_norm": 0.9081159830093384,
      "learning_rate": 7.607373074216573e-05,
      "loss": 41.939306640625,
      "step": 22800
    },
    {
      "epoch": 0.916,
      "grad_norm": 0.8322077989578247,
      "learning_rate": 6.935744643737252e-05,
      "loss": 42.21763671875,
      "step": 22900
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8767455816268921,
      "learning_rate": 6.294627701237876e-05,
      "loss": 42.1103955078125,
      "step": 23000
    },
    {
      "epoch": 0.92,
      "eval_accuracy": 0.22741544477028347,
      "eval_loss": 42.044464111328125,
      "eval_runtime": 7.5593,
      "eval_samples_per_second": 16.536,
      "eval_steps_per_second": 0.529,
      "step": 23000
    },
    {
      "epoch": 0.924,
      "grad_norm": 0.8490807414054871,
      "learning_rate": 5.6841235675552106e-05,
      "loss": 42.408046875,
      "step": 23100
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.8971438407897949,
      "learning_rate": 5.1043287255389206e-05,
      "loss": 42.293935546875,
      "step": 23200
    },
    {
      "epoch": 0.932,
      "grad_norm": 0.8031958341598511,
      "learning_rate": 4.555334804803857e-05,
      "loss": 42.037109375,
      "step": 23300
    },
    {
      "epoch": 0.936,
      "grad_norm": 0.9145307540893555,
      "learning_rate": 4.03722856724893e-05,
      "loss": 42.266640625,
      "step": 23400
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.105343222618103,
      "learning_rate": 3.5500918933456085e-05,
      "loss": 41.9810498046875,
      "step": 23500
    },
    {
      "epoch": 0.944,
      "grad_norm": 1.2218998670578003,
      "learning_rate": 3.094001769197474e-05,
      "loss": 42.188876953125,
      "step": 23600
    },
    {
      "epoch": 0.948,
      "grad_norm": 0.8797289729118347,
      "learning_rate": 2.669030274373663e-05,
      "loss": 42.359853515625,
      "step": 23700
    },
    {
      "epoch": 0.952,
      "grad_norm": 0.8956351280212402,
      "learning_rate": 2.2752445705175097e-05,
      "loss": 42.1323095703125,
      "step": 23800
    },
    {
      "epoch": 0.956,
      "grad_norm": 0.8726316690444946,
      "learning_rate": 1.9127068907324184e-05,
      "loss": 42.337939453125,
      "step": 23900
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1114789247512817,
      "learning_rate": 1.581474529746729e-05,
      "loss": 42.27337890625,
      "step": 24000
    },
    {
      "epoch": 0.96,
      "eval_accuracy": 0.22743890518084067,
      "eval_loss": 42.037322998046875,
      "eval_runtime": 7.0814,
      "eval_samples_per_second": 17.652,
      "eval_steps_per_second": 0.565,
      "step": 24000
    },
    {
      "epoch": 0.964,
      "grad_norm": 1.4101399183273315,
      "learning_rate": 1.281599834858893e-05,
      "loss": 42.2501513671875,
      "step": 24100
    },
    {
      "epoch": 0.968,
      "grad_norm": 1.015312910079956,
      "learning_rate": 1.0131301976646912e-05,
      "loss": 42.1179931640625,
      "step": 24200
    },
    {
      "epoch": 0.972,
      "grad_norm": 0.8839344382286072,
      "learning_rate": 7.761080465675363e-06,
      "loss": 42.1094140625,
      "step": 24300
    },
    {
      "epoch": 0.976,
      "grad_norm": 1.0040006637573242,
      "learning_rate": 5.705708400731702e-06,
      "loss": 42.0864111328125,
      "step": 24400
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9537904262542725,
      "learning_rate": 3.965510608697098e-06,
      "loss": 42.1570068359375,
      "step": 24500
    },
    {
      "epoch": 0.984,
      "grad_norm": 0.9448596239089966,
      "learning_rate": 2.5407621069435395e-06,
      "loss": 42.424609375,
      "step": 24600
    },
    {
      "epoch": 0.988,
      "grad_norm": 1.1038907766342163,
      "learning_rate": 1.4316880598685967e-06,
      "loss": 42.390634765625,
      "step": 24700
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.9914430379867554,
      "learning_rate": 6.3846374331189e-07,
      "loss": 42.1272607421875,
      "step": 24800
    },
    {
      "epoch": 0.996,
      "grad_norm": 0.954636812210083,
      "learning_rate": 1.6121451685435774e-07,
      "loss": 42.1366357421875,
      "step": 24900
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.122141718864441,
      "learning_rate": 1.5804007658104523e-11,
      "loss": 42.0115185546875,
      "step": 25000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.22748093841642228,
      "eval_loss": 42.03590393066406,
      "eval_runtime": 6.9933,
      "eval_samples_per_second": 17.874,
      "eval_steps_per_second": 0.572,
      "step": 25000
    }
  ],
  "logging_steps": 100,
  "max_steps": 25000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}