{
  "best_metric": 0.4959826756763837,
  "best_model_checkpoint": "/fsx/anton/cosmopedia/edu_score/bert_snowflake_regression_4/checkpoint-27000",
  "epoch": 16.383495145631066,
  "eval_steps": 1000,
  "global_step": 27000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.06067961165048544,
      "grad_norm": 0.5638211965560913,
      "learning_rate": 0.0002990898058252427,
      "loss": 0.4753,
      "step": 100
    },
    {
      "epoch": 0.12135922330097088,
      "grad_norm": 0.47830212116241455,
      "learning_rate": 0.0002981796116504854,
      "loss": 0.357,
      "step": 200
    },
    {
      "epoch": 0.1820388349514563,
      "grad_norm": 0.6941384077072144,
      "learning_rate": 0.0002972694174757281,
      "loss": 0.3542,
      "step": 300
    },
    {
      "epoch": 0.24271844660194175,
      "grad_norm": 0.459163635969162,
      "learning_rate": 0.00029635922330097087,
      "loss": 0.3508,
      "step": 400
    },
    {
      "epoch": 0.30339805825242716,
      "grad_norm": 1.1585971117019653,
      "learning_rate": 0.0002954490291262136,
      "loss": 0.3407,
      "step": 500
    },
    {
      "epoch": 0.3640776699029126,
      "grad_norm": 0.6505594849586487,
      "learning_rate": 0.0002945388349514563,
      "loss": 0.3394,
      "step": 600
    },
    {
      "epoch": 0.42475728155339804,
      "grad_norm": 0.9804072976112366,
      "learning_rate": 0.000293628640776699,
      "loss": 0.3435,
      "step": 700
    },
    {
      "epoch": 0.4854368932038835,
      "grad_norm": 0.5816351175308228,
      "learning_rate": 0.0002927184466019417,
      "loss": 0.3323,
      "step": 800
    },
    {
      "epoch": 0.5461165048543689,
      "grad_norm": 0.6582027673721313,
      "learning_rate": 0.00029180825242718447,
      "loss": 0.3293,
      "step": 900
    },
    {
      "epoch": 0.6067961165048543,
      "grad_norm": 0.8432559370994568,
      "learning_rate": 0.0002908980582524271,
      "loss": 0.3337,
      "step": 1000
    },
    {
      "epoch": 0.6067961165048543,
      "eval_accuracy": 0.6516312117268014,
      "eval_f1_macro": 0.388617225018415,
      "eval_loss": 0.32428401708602905,
      "eval_precision": 0.5273989146868351,
      "eval_recall": 0.3731209274235363,
      "eval_runtime": 63.5791,
      "eval_samples_per_second": 737.144,
      "eval_steps_per_second": 5.772,
      "step": 1000
    },
    {
      "epoch": 0.6674757281553398,
      "grad_norm": 0.6387248039245605,
      "learning_rate": 0.0002899878640776699,
      "loss": 0.3292,
      "step": 1100
    },
    {
      "epoch": 0.7281553398058253,
      "grad_norm": 2.0111730098724365,
      "learning_rate": 0.0002890776699029126,
      "loss": 0.3283,
      "step": 1200
    },
    {
      "epoch": 0.7888349514563107,
      "grad_norm": 1.0319699048995972,
      "learning_rate": 0.0002881674757281553,
      "loss": 0.3236,
      "step": 1300
    },
    {
      "epoch": 0.8495145631067961,
      "grad_norm": 1.194286584854126,
      "learning_rate": 0.000287257281553398,
      "loss": 0.3178,
      "step": 1400
    },
    {
      "epoch": 0.9101941747572816,
      "grad_norm": 0.4329046308994293,
      "learning_rate": 0.00028634708737864073,
      "loss": 0.3234,
      "step": 1500
    },
    {
      "epoch": 0.970873786407767,
      "grad_norm": 0.4490291476249695,
      "learning_rate": 0.0002854368932038835,
      "loss": 0.3148,
      "step": 1600
    },
    {
      "epoch": 1.0315533980582525,
      "grad_norm": 1.9341398477554321,
      "learning_rate": 0.0002845266990291262,
      "loss": 0.3157,
      "step": 1700
    },
    {
      "epoch": 1.0922330097087378,
      "grad_norm": 0.6705629825592041,
      "learning_rate": 0.0002836165048543689,
      "loss": 0.3144,
      "step": 1800
    },
    {
      "epoch": 1.1529126213592233,
      "grad_norm": 0.3708420395851135,
      "learning_rate": 0.0002827063106796116,
      "loss": 0.3106,
      "step": 1900
    },
    {
      "epoch": 1.2135922330097086,
      "grad_norm": 0.4166070222854614,
      "learning_rate": 0.00028179611650485433,
      "loss": 0.3065,
      "step": 2000
    },
    {
      "epoch": 1.2135922330097086,
      "eval_accuracy": 0.6756993193505025,
      "eval_f1_macro": 0.4177625223934091,
      "eval_loss": 0.3020932972431183,
      "eval_precision": 0.5208365532058327,
      "eval_recall": 0.3908095186846785,
      "eval_runtime": 63.8686,
      "eval_samples_per_second": 733.803,
      "eval_steps_per_second": 5.746,
      "step": 2000
    },
    {
      "epoch": 1.2742718446601942,
      "grad_norm": 0.949810266494751,
      "learning_rate": 0.00028088592233009704,
      "loss": 0.309,
      "step": 2100
    },
    {
      "epoch": 1.3349514563106797,
      "grad_norm": 0.6933236718177795,
      "learning_rate": 0.00027997572815533975,
      "loss": 0.3039,
      "step": 2200
    },
    {
      "epoch": 1.395631067961165,
      "grad_norm": 0.4874693751335144,
      "learning_rate": 0.0002790655339805825,
      "loss": 0.3016,
      "step": 2300
    },
    {
      "epoch": 1.4563106796116505,
      "grad_norm": 0.5307803750038147,
      "learning_rate": 0.0002781553398058252,
      "loss": 0.295,
      "step": 2400
    },
    {
      "epoch": 1.516990291262136,
      "grad_norm": 0.7260825634002686,
      "learning_rate": 0.00027724514563106793,
      "loss": 0.298,
      "step": 2500
    },
    {
      "epoch": 1.5776699029126213,
      "grad_norm": 1.3546072244644165,
      "learning_rate": 0.00027633495145631064,
      "loss": 0.2937,
      "step": 2600
    },
    {
      "epoch": 1.6383495145631068,
      "grad_norm": 0.7695233821868896,
      "learning_rate": 0.00027542475728155335,
      "loss": 0.2939,
      "step": 2700
    },
    {
      "epoch": 1.6990291262135924,
      "grad_norm": 0.46857160329818726,
      "learning_rate": 0.0002745145631067961,
      "loss": 0.2911,
      "step": 2800
    },
    {
      "epoch": 1.7597087378640777,
      "grad_norm": 0.521542489528656,
      "learning_rate": 0.0002736043689320388,
      "loss": 0.2936,
      "step": 2900
    },
    {
      "epoch": 1.820388349514563,
      "grad_norm": 1.1797749996185303,
      "learning_rate": 0.00027269417475728154,
      "loss": 0.291,
      "step": 3000
    },
    {
      "epoch": 1.820388349514563,
      "eval_accuracy": 0.6800947361683061,
      "eval_f1_macro": 0.43005056999542096,
      "eval_loss": 0.2904761731624603,
      "eval_precision": 0.5283372179121356,
      "eval_recall": 0.40012210038254903,
      "eval_runtime": 63.5697,
      "eval_samples_per_second": 737.253,
      "eval_steps_per_second": 5.773,
      "step": 3000
    },
    {
      "epoch": 1.8810679611650487,
      "grad_norm": 0.9966709613800049,
      "learning_rate": 0.00027178398058252425,
      "loss": 0.2965,
      "step": 3100
    },
    {
      "epoch": 1.941747572815534,
      "grad_norm": 0.40996024012565613,
      "learning_rate": 0.00027087378640776696,
      "loss": 0.2941,
      "step": 3200
    },
    {
      "epoch": 2.0024271844660193,
      "grad_norm": 0.5450060367584229,
      "learning_rate": 0.00026996359223300967,
      "loss": 0.2912,
      "step": 3300
    },
    {
      "epoch": 2.063106796116505,
      "grad_norm": 0.5307539701461792,
      "learning_rate": 0.0002690533980582524,
      "loss": 0.2872,
      "step": 3400
    },
    {
      "epoch": 2.1237864077669903,
      "grad_norm": 0.5863193273544312,
      "learning_rate": 0.00026814320388349514,
      "loss": 0.2929,
      "step": 3500
    },
    {
      "epoch": 2.1844660194174756,
      "grad_norm": 0.584078311920166,
      "learning_rate": 0.00026723300970873785,
      "loss": 0.2879,
      "step": 3600
    },
    {
      "epoch": 2.2451456310679614,
      "grad_norm": 0.6381602883338928,
      "learning_rate": 0.00026632281553398056,
      "loss": 0.2892,
      "step": 3700
    },
    {
      "epoch": 2.3058252427184467,
      "grad_norm": 0.4760149121284485,
      "learning_rate": 0.00026541262135922327,
      "loss": 0.2863,
      "step": 3800
    },
    {
      "epoch": 2.366504854368932,
      "grad_norm": 0.4088296890258789,
      "learning_rate": 0.000264502427184466,
      "loss": 0.2913,
      "step": 3900
    },
    {
      "epoch": 2.4271844660194173,
      "grad_norm": 1.3476176261901855,
      "learning_rate": 0.00026359223300970874,
      "loss": 0.2845,
      "step": 4000
    },
    {
      "epoch": 2.4271844660194173,
      "eval_accuracy": 0.6929182580493738,
      "eval_f1_macro": 0.4442664055274559,
      "eval_loss": 0.280377060174942,
      "eval_precision": 0.5328225535144124,
      "eval_recall": 0.41160029395774395,
      "eval_runtime": 64.199,
      "eval_samples_per_second": 730.027,
      "eval_steps_per_second": 5.717,
      "step": 4000
    },
    {
      "epoch": 2.487864077669903,
      "grad_norm": 0.6318752765655518,
      "learning_rate": 0.0002626820388349514,
      "loss": 0.2812,
      "step": 4100
    },
    {
      "epoch": 2.5485436893203883,
      "grad_norm": 0.49435973167419434,
      "learning_rate": 0.00026177184466019416,
      "loss": 0.2803,
      "step": 4200
    },
    {
      "epoch": 2.6092233009708736,
      "grad_norm": 0.4300900101661682,
      "learning_rate": 0.00026086165048543687,
      "loss": 0.2853,
      "step": 4300
    },
    {
      "epoch": 2.6699029126213594,
      "grad_norm": 0.9545436501502991,
      "learning_rate": 0.0002599514563106796,
      "loss": 0.2813,
      "step": 4400
    },
    {
      "epoch": 2.7305825242718447,
      "grad_norm": 0.5803716778755188,
      "learning_rate": 0.0002590412621359223,
      "loss": 0.2838,
      "step": 4500
    },
    {
      "epoch": 2.79126213592233,
      "grad_norm": 1.4714713096618652,
      "learning_rate": 0.000258131067961165,
      "loss": 0.2814,
      "step": 4600
    },
    {
      "epoch": 2.8519417475728153,
      "grad_norm": 0.6767821311950684,
      "learning_rate": 0.00025722087378640777,
      "loss": 0.2741,
      "step": 4700
    },
    {
      "epoch": 2.912621359223301,
      "grad_norm": 0.4653462767601013,
      "learning_rate": 0.0002563106796116505,
      "loss": 0.2783,
      "step": 4800
    },
    {
      "epoch": 2.9733009708737863,
      "grad_norm": 1.3012775182724,
      "learning_rate": 0.0002554004854368932,
      "loss": 0.283,
      "step": 4900
    },
    {
      "epoch": 3.033980582524272,
      "grad_norm": 0.4733451306819916,
      "learning_rate": 0.0002544902912621359,
      "loss": 0.2767,
      "step": 5000
    },
    {
      "epoch": 3.033980582524272,
      "eval_accuracy": 0.6949666076343696,
      "eval_f1_macro": 0.4524166681181929,
      "eval_loss": 0.2772601842880249,
      "eval_precision": 0.52914110464024,
      "eval_recall": 0.42261704559523156,
      "eval_runtime": 63.4257,
      "eval_samples_per_second": 738.927,
      "eval_steps_per_second": 5.786,
      "step": 5000
    },
    {
      "epoch": 3.0946601941747574,
      "grad_norm": 0.4103662371635437,
      "learning_rate": 0.0002535800970873786,
      "loss": 0.2796,
      "step": 5100
    },
    {
      "epoch": 3.1553398058252426,
      "grad_norm": 0.4195462763309479,
      "learning_rate": 0.0002526699029126213,
      "loss": 0.2757,
      "step": 5200
    },
    {
      "epoch": 3.216019417475728,
      "grad_norm": 1.2391552925109863,
      "learning_rate": 0.0002517597087378641,
      "loss": 0.2783,
      "step": 5300
    },
    {
      "epoch": 3.2766990291262137,
      "grad_norm": 1.2029412984848022,
      "learning_rate": 0.0002508495145631068,
      "loss": 0.2772,
      "step": 5400
    },
    {
      "epoch": 3.337378640776699,
      "grad_norm": 0.5050978660583496,
      "learning_rate": 0.0002499393203883495,
      "loss": 0.2776,
      "step": 5500
    },
    {
      "epoch": 3.3980582524271843,
      "grad_norm": 1.0107412338256836,
      "learning_rate": 0.0002490291262135922,
      "loss": 0.2766,
      "step": 5600
    },
    {
      "epoch": 3.45873786407767,
      "grad_norm": 0.4374917149543762,
      "learning_rate": 0.0002481189320388349,
      "loss": 0.2719,
      "step": 5700
    },
    {
      "epoch": 3.5194174757281553,
      "grad_norm": 1.6768765449523926,
      "learning_rate": 0.0002472087378640777,
      "loss": 0.2803,
      "step": 5800
    },
    {
      "epoch": 3.5800970873786406,
      "grad_norm": 0.8120823502540588,
      "learning_rate": 0.0002462985436893204,
      "loss": 0.2723,
      "step": 5900
    },
    {
      "epoch": 3.6407766990291264,
      "grad_norm": 1.3967177867889404,
      "learning_rate": 0.0002453883495145631,
      "loss": 0.2796,
      "step": 6000
    },
    {
      "epoch": 3.6407766990291264,
      "eval_accuracy": 0.7001301555465466,
      "eval_f1_macro": 0.46180498265852127,
      "eval_loss": 0.272257536649704,
      "eval_precision": 0.5281578618931982,
      "eval_recall": 0.4315295129889904,
      "eval_runtime": 63.8213,
      "eval_samples_per_second": 734.347,
      "eval_steps_per_second": 5.75,
      "step": 6000
    },
    {
      "epoch": 3.7014563106796117,
      "grad_norm": 0.6093985438346863,
      "learning_rate": 0.0002444781553398058,
      "loss": 0.2744,
      "step": 6100
    },
    {
      "epoch": 3.762135922330097,
      "grad_norm": 0.7282202243804932,
      "learning_rate": 0.00024356796116504852,
      "loss": 0.2715,
      "step": 6200
    },
    {
      "epoch": 3.8228155339805827,
      "grad_norm": 1.1341967582702637,
      "learning_rate": 0.00024265776699029123,
      "loss": 0.2709,
      "step": 6300
    },
    {
      "epoch": 3.883495145631068,
      "grad_norm": 0.8576841354370117,
      "learning_rate": 0.00024174757281553394,
      "loss": 0.275,
      "step": 6400
    },
    {
      "epoch": 3.9441747572815533,
      "grad_norm": 0.5656840205192566,
      "learning_rate": 0.00024083737864077668,
      "loss": 0.2676,
      "step": 6500
    },
    {
      "epoch": 4.004854368932039,
      "grad_norm": 0.6544743180274963,
      "learning_rate": 0.0002399271844660194,
      "loss": 0.2734,
      "step": 6600
    },
    {
      "epoch": 4.065533980582524,
      "grad_norm": 1.5159205198287964,
      "learning_rate": 0.0002390169902912621,
      "loss": 0.2666,
      "step": 6700
    },
    {
      "epoch": 4.12621359223301,
      "grad_norm": 0.9112799763679504,
      "learning_rate": 0.00023810679611650483,
      "loss": 0.2646,
      "step": 6800
    },
    {
      "epoch": 4.186893203883495,
      "grad_norm": 0.6971092224121094,
      "learning_rate": 0.00023719660194174754,
      "loss": 0.2681,
      "step": 6900
    },
    {
      "epoch": 4.247572815533981,
      "grad_norm": 0.5126680731773376,
      "learning_rate": 0.00023628640776699028,
      "loss": 0.2669,
      "step": 7000
    },
    {
      "epoch": 4.247572815533981,
      "eval_accuracy": 0.7024985597541981,
      "eval_f1_macro": 0.4678440617843855,
      "eval_loss": 0.26851192116737366,
      "eval_precision": 0.5330566032345198,
      "eval_recall": 0.43620949446199525,
      "eval_runtime": 63.9009,
      "eval_samples_per_second": 733.432,
      "eval_steps_per_second": 5.743,
      "step": 7000
    },
    {
      "epoch": 4.308252427184466,
      "grad_norm": 0.40426695346832275,
      "learning_rate": 0.000235376213592233,
      "loss": 0.2722,
      "step": 7100
    },
    {
      "epoch": 4.368932038834951,
      "grad_norm": 0.4849171042442322,
      "learning_rate": 0.0002344660194174757,
      "loss": 0.273,
      "step": 7200
    },
    {
      "epoch": 4.429611650485437,
      "grad_norm": 0.7887744307518005,
      "learning_rate": 0.00023355582524271844,
      "loss": 0.2711,
      "step": 7300
    },
    {
      "epoch": 4.490291262135923,
      "grad_norm": 1.282893419265747,
      "learning_rate": 0.00023264563106796115,
      "loss": 0.2633,
      "step": 7400
    },
    {
      "epoch": 4.550970873786408,
      "grad_norm": 0.4942808449268341,
      "learning_rate": 0.00023173543689320386,
      "loss": 0.2655,
      "step": 7500
    },
    {
      "epoch": 4.611650485436893,
      "grad_norm": 0.4098761975765228,
      "learning_rate": 0.0002308252427184466,
      "loss": 0.2719,
      "step": 7600
    },
    {
      "epoch": 4.672330097087379,
      "grad_norm": 1.1584999561309814,
      "learning_rate": 0.0002299150485436893,
      "loss": 0.2646,
      "step": 7700
    },
    {
      "epoch": 4.733009708737864,
      "grad_norm": 0.5956116318702698,
      "learning_rate": 0.00022900485436893204,
      "loss": 0.2681,
      "step": 7800
    },
    {
      "epoch": 4.793689320388349,
      "grad_norm": 0.7886734008789062,
      "learning_rate": 0.00022809466019417472,
      "loss": 0.2648,
      "step": 7900
    },
    {
      "epoch": 4.854368932038835,
      "grad_norm": 0.7086551785469055,
      "learning_rate": 0.00022718446601941746,
      "loss": 0.2667,
      "step": 8000
    },
    {
      "epoch": 4.854368932038835,
      "eval_accuracy": 0.7028826253013848,
      "eval_f1_macro": 0.4680422447785217,
      "eval_loss": 0.2663130462169647,
      "eval_precision": 0.5388642351614853,
      "eval_recall": 0.4370831407609595,
      "eval_runtime": 63.9774,
      "eval_samples_per_second": 732.556,
      "eval_steps_per_second": 5.736,
      "step": 8000
    },
    {
      "epoch": 4.915048543689321,
      "grad_norm": 0.6959311366081238,
      "learning_rate": 0.0002262742718446602,
      "loss": 0.2662,
      "step": 8100
    },
    {
      "epoch": 4.975728155339806,
      "grad_norm": 1.318164587020874,
      "learning_rate": 0.0002253640776699029,
      "loss": 0.2679,
      "step": 8200
    },
    {
      "epoch": 5.036407766990291,
      "grad_norm": 0.5080001354217529,
      "learning_rate": 0.00022445388349514561,
      "loss": 0.2662,
      "step": 8300
    },
    {
      "epoch": 5.097087378640777,
      "grad_norm": 0.40693196654319763,
      "learning_rate": 0.00022354368932038832,
      "loss": 0.2641,
      "step": 8400
    },
    {
      "epoch": 5.157766990291262,
      "grad_norm": 0.42492175102233887,
      "learning_rate": 0.00022263349514563106,
      "loss": 0.2669,
      "step": 8500
    },
    {
      "epoch": 5.218446601941747,
      "grad_norm": 0.9580904841423035,
      "learning_rate": 0.00022172330097087374,
      "loss": 0.2613,
      "step": 8600
    },
    {
      "epoch": 5.279126213592233,
      "grad_norm": 0.573049008846283,
      "learning_rate": 0.00022081310679611648,
      "loss": 0.26,
      "step": 8700
    },
    {
      "epoch": 5.339805825242719,
      "grad_norm": 0.7355623841285706,
      "learning_rate": 0.00021990291262135922,
      "loss": 0.2682,
      "step": 8800
    },
    {
      "epoch": 5.400485436893204,
      "grad_norm": 0.5867640972137451,
      "learning_rate": 0.00021899271844660193,
      "loss": 0.2677,
      "step": 8900
    },
    {
      "epoch": 5.461165048543689,
      "grad_norm": 1.283672571182251,
      "learning_rate": 0.00021808252427184464,
      "loss": 0.2582,
      "step": 9000
    },
    {
      "epoch": 5.461165048543689,
      "eval_accuracy": 0.7030319841252907,
      "eval_f1_macro": 0.476191462667983,
      "eval_loss": 0.2642744779586792,
      "eval_precision": 0.5389130008234221,
      "eval_recall": 0.4475692410266268,
      "eval_runtime": 63.527,
      "eval_samples_per_second": 737.749,
      "eval_steps_per_second": 5.777,
      "step": 9000
    },
    {
      "epoch": 5.521844660194175,
      "grad_norm": 0.4764248728752136,
      "learning_rate": 0.00021717233009708735,
      "loss": 0.2596,
      "step": 9100
    },
    {
      "epoch": 5.58252427184466,
      "grad_norm": 0.7470163106918335,
      "learning_rate": 0.00021626213592233008,
      "loss": 0.2632,
      "step": 9200
    },
    {
      "epoch": 5.643203883495145,
      "grad_norm": 0.7628265023231506,
      "learning_rate": 0.00021535194174757282,
      "loss": 0.2583,
      "step": 9300
    },
    {
      "epoch": 5.703883495145631,
      "grad_norm": 0.4257694184780121,
      "learning_rate": 0.0002144417475728155,
      "loss": 0.2627,
      "step": 9400
    },
    {
      "epoch": 5.764563106796117,
      "grad_norm": 0.5351702570915222,
      "learning_rate": 0.00021353155339805824,
      "loss": 0.2578,
      "step": 9500
    },
    {
      "epoch": 5.825242718446602,
      "grad_norm": 1.9470020532608032,
      "learning_rate": 0.00021262135922330095,
      "loss": 0.2599,
      "step": 9600
    },
    {
      "epoch": 5.885922330097087,
      "grad_norm": 0.41918280720710754,
      "learning_rate": 0.00021171116504854369,
      "loss": 0.2604,
      "step": 9700
    },
    {
      "epoch": 5.946601941747573,
      "grad_norm": 0.7866085767745972,
      "learning_rate": 0.00021080097087378637,
      "loss": 0.2645,
      "step": 9800
    },
    {
      "epoch": 6.007281553398058,
      "grad_norm": 0.773632287979126,
      "learning_rate": 0.0002098907766990291,
      "loss": 0.2594,
      "step": 9900
    },
    {
      "epoch": 6.067961165048544,
      "grad_norm": 0.905390202999115,
      "learning_rate": 0.00020898058252427184,
      "loss": 0.2587,
      "step": 10000
    },
    {
      "epoch": 6.067961165048544,
      "eval_accuracy": 0.7069153135468453,
      "eval_f1_macro": 0.47329615730761665,
      "eval_loss": 0.26157495379447937,
      "eval_precision": 0.5382757148939651,
      "eval_recall": 0.4417990410758896,
      "eval_runtime": 64.0102,
      "eval_samples_per_second": 732.181,
      "eval_steps_per_second": 5.733,
      "step": 10000
    },
    {
      "epoch": 6.128640776699029,
      "grad_norm": 0.7645794749259949,
      "learning_rate": 0.00020807038834951455,
      "loss": 0.2584,
      "step": 10100
    },
    {
      "epoch": 6.189320388349515,
      "grad_norm": 0.80213862657547,
      "learning_rate": 0.00020716019417475726,
      "loss": 0.2576,
      "step": 10200
    },
    {
      "epoch": 6.25,
      "grad_norm": 0.49435296654701233,
      "learning_rate": 0.00020624999999999997,
      "loss": 0.2578,
      "step": 10300
    },
    {
      "epoch": 6.310679611650485,
      "grad_norm": 1.2024767398834229,
      "learning_rate": 0.0002053398058252427,
      "loss": 0.2628,
      "step": 10400
    },
    {
      "epoch": 6.371359223300971,
      "grad_norm": 0.43948036432266235,
      "learning_rate": 0.0002044296116504854,
      "loss": 0.2555,
      "step": 10500
    },
    {
      "epoch": 6.432038834951456,
      "grad_norm": 1.0257009267807007,
      "learning_rate": 0.00020351941747572813,
      "loss": 0.2582,
      "step": 10600
    },
    {
      "epoch": 6.492718446601942,
      "grad_norm": 1.2218210697174072,
      "learning_rate": 0.00020260922330097086,
      "loss": 0.2603,
      "step": 10700
    },
    {
      "epoch": 6.553398058252427,
      "grad_norm": 0.43541598320007324,
      "learning_rate": 0.00020169902912621357,
      "loss": 0.2529,
      "step": 10800
    },
    {
      "epoch": 6.614077669902913,
      "grad_norm": 0.4533527195453644,
      "learning_rate": 0.00020078883495145628,
      "loss": 0.255,
      "step": 10900
    },
    {
      "epoch": 6.674757281553398,
      "grad_norm": 0.714527428150177,
      "learning_rate": 0.000199878640776699,
      "loss": 0.2524,
      "step": 11000
    },
    {
      "epoch": 6.674757281553398,
      "eval_accuracy": 0.7088356412827789,
      "eval_f1_macro": 0.471743421411692,
      "eval_loss": 0.2603248357772827,
      "eval_precision": 0.5487634473468738,
      "eval_recall": 0.43766522368406974,
      "eval_runtime": 64.8962,
      "eval_samples_per_second": 722.184,
      "eval_steps_per_second": 5.655,
      "step": 11000
    },
    {
      "epoch": 6.735436893203883,
      "grad_norm": 0.6983916163444519,
      "learning_rate": 0.00019896844660194173,
      "loss": 0.2585,
      "step": 11100
    },
    {
      "epoch": 6.796116504854369,
      "grad_norm": 0.8627694845199585,
      "learning_rate": 0.00019805825242718447,
      "loss": 0.2557,
      "step": 11200
    },
    {
      "epoch": 6.856796116504855,
      "grad_norm": 0.40715476870536804,
      "learning_rate": 0.00019714805825242715,
      "loss": 0.2588,
      "step": 11300
    },
    {
      "epoch": 6.91747572815534,
      "grad_norm": 1.0038883686065674,
      "learning_rate": 0.0001962378640776699,
      "loss": 0.2542,
      "step": 11400
    },
    {
      "epoch": 6.978155339805825,
      "grad_norm": 0.941238522529602,
      "learning_rate": 0.0001953276699029126,
      "loss": 0.2597,
      "step": 11500
    },
    {
      "epoch": 7.038834951456311,
      "grad_norm": 0.6316859722137451,
      "learning_rate": 0.00019441747572815533,
      "loss": 0.2576,
      "step": 11600
    },
    {
      "epoch": 7.099514563106796,
      "grad_norm": 0.5387765169143677,
      "learning_rate": 0.00019350728155339804,
      "loss": 0.2525,
      "step": 11700
    },
    {
      "epoch": 7.160194174757281,
      "grad_norm": 0.9042656421661377,
      "learning_rate": 0.00019259708737864075,
      "loss": 0.2552,
      "step": 11800
    },
    {
      "epoch": 7.220873786407767,
      "grad_norm": 0.6641530990600586,
      "learning_rate": 0.0001916868932038835,
      "loss": 0.2495,
      "step": 11900
    },
    {
      "epoch": 7.281553398058253,
      "grad_norm": 1.050493836402893,
      "learning_rate": 0.0001907766990291262,
      "loss": 0.2538,
      "step": 12000
    },
    {
      "epoch": 7.281553398058253,
      "eval_accuracy": 0.7098171421255894,
      "eval_f1_macro": 0.4800779183645662,
      "eval_loss": 0.25889766216278076,
      "eval_precision": 0.5445483450890162,
      "eval_recall": 0.44845043483484776,
      "eval_runtime": 63.7915,
      "eval_samples_per_second": 734.69,
      "eval_steps_per_second": 5.753,
      "step": 12000
    },
    {
      "epoch": 7.342233009708738,
      "grad_norm": 0.7460100054740906,
      "learning_rate": 0.0001898665048543689,
      "loss": 0.2576,
      "step": 12100
    },
    {
      "epoch": 7.402912621359223,
      "grad_norm": 1.8915176391601562,
      "learning_rate": 0.00018895631067961165,
      "loss": 0.2563,
      "step": 12200
    },
    {
      "epoch": 7.463592233009709,
      "grad_norm": 0.6642426252365112,
      "learning_rate": 0.00018804611650485436,
      "loss": 0.2515,
      "step": 12300
    },
    {
      "epoch": 7.524271844660194,
      "grad_norm": 0.7037068009376526,
      "learning_rate": 0.00018713592233009707,
      "loss": 0.2525,
      "step": 12400
    },
    {
      "epoch": 7.584951456310679,
      "grad_norm": 0.44775909185409546,
      "learning_rate": 0.00018622572815533978,
      "loss": 0.2557,
      "step": 12500
    },
    {
      "epoch": 7.645631067961165,
      "grad_norm": 1.2254287004470825,
      "learning_rate": 0.0001853155339805825,
      "loss": 0.252,
      "step": 12600
    },
    {
      "epoch": 7.706310679611651,
      "grad_norm": 0.7916478514671326,
      "learning_rate": 0.00018440533980582525,
      "loss": 0.2503,
      "step": 12700
    },
    {
      "epoch": 7.766990291262136,
      "grad_norm": 0.5165425539016724,
      "learning_rate": 0.00018349514563106793,
      "loss": 0.259,
      "step": 12800
    },
    {
      "epoch": 7.827669902912621,
      "grad_norm": 0.45590564608573914,
      "learning_rate": 0.00018258495145631067,
      "loss": 0.2536,
      "step": 12900
    },
    {
      "epoch": 7.888349514563107,
      "grad_norm": 0.7460587620735168,
      "learning_rate": 0.00018167475728155338,
      "loss": 0.2536,
      "step": 13000
    },
    {
      "epoch": 7.888349514563107,
      "eval_accuracy": 0.7118014807860542,
      "eval_f1_macro": 0.48153346170307537,
      "eval_loss": 0.2587451934814453,
      "eval_precision": 0.5428621211876972,
      "eval_recall": 0.44971967808318786,
      "eval_runtime": 63.3555,
      "eval_samples_per_second": 739.747,
      "eval_steps_per_second": 5.793,
      "step": 13000
    },
    {
      "epoch": 7.949029126213592,
      "grad_norm": 0.5832739472389221,
      "learning_rate": 0.00018076456310679612,
      "loss": 0.2549,
      "step": 13100
    },
    {
      "epoch": 8.009708737864077,
      "grad_norm": 0.991656482219696,
      "learning_rate": 0.0001798543689320388,
      "loss": 0.2516,
      "step": 13200
    },
    {
      "epoch": 8.070388349514563,
      "grad_norm": 0.5339873433113098,
      "learning_rate": 0.00017894417475728154,
      "loss": 0.2492,
      "step": 13300
    },
    {
      "epoch": 8.131067961165048,
      "grad_norm": 0.5186127424240112,
      "learning_rate": 0.00017803398058252427,
      "loss": 0.2521,
      "step": 13400
    },
    {
      "epoch": 8.191747572815533,
      "grad_norm": 0.7012351751327515,
      "learning_rate": 0.00017712378640776698,
      "loss": 0.247,
      "step": 13500
    },
    {
      "epoch": 8.25242718446602,
      "grad_norm": 0.5204672813415527,
      "learning_rate": 0.0001762135922330097,
      "loss": 0.2528,
      "step": 13600
    },
    {
      "epoch": 8.313106796116505,
      "grad_norm": 0.46670639514923096,
      "learning_rate": 0.0001753033980582524,
      "loss": 0.2484,
      "step": 13700
    },
    {
      "epoch": 8.37378640776699,
      "grad_norm": 0.8041334748268127,
      "learning_rate": 0.00017439320388349514,
      "loss": 0.2534,
      "step": 13800
    },
    {
      "epoch": 8.434466019417476,
      "grad_norm": 0.482832133769989,
      "learning_rate": 0.00017348300970873787,
      "loss": 0.2463,
      "step": 13900
    },
    {
      "epoch": 8.495145631067961,
      "grad_norm": 0.9541172981262207,
      "learning_rate": 0.00017257281553398056,
      "loss": 0.2527,
      "step": 14000
    },
    {
      "epoch": 8.495145631067961,
      "eval_accuracy": 0.7112040454904304,
      "eval_f1_macro": 0.4703790685018183,
      "eval_loss": 0.2594839036464691,
      "eval_precision": 0.5469933734414426,
      "eval_recall": 0.43675078934194306,
      "eval_runtime": 64.5706,
      "eval_samples_per_second": 725.825,
      "eval_steps_per_second": 5.684,
      "step": 14000
    },
    {
      "epoch": 8.555825242718447,
      "grad_norm": 0.41729798913002014,
      "learning_rate": 0.0001716626213592233,
      "loss": 0.2566,
      "step": 14100
    },
    {
      "epoch": 8.616504854368932,
      "grad_norm": 1.1643282175064087,
      "learning_rate": 0.000170752427184466,
      "loss": 0.2492,
      "step": 14200
    },
    {
      "epoch": 8.677184466019417,
      "grad_norm": 0.5015943050384521,
      "learning_rate": 0.00016984223300970874,
      "loss": 0.2484,
      "step": 14300
    },
    {
      "epoch": 8.737864077669903,
      "grad_norm": 0.6088297963142395,
      "learning_rate": 0.00016893203883495142,
      "loss": 0.2481,
      "step": 14400
    },
    {
      "epoch": 8.798543689320388,
      "grad_norm": 0.6608941555023193,
      "learning_rate": 0.00016802184466019416,
      "loss": 0.2509,
      "step": 14500
    },
    {
      "epoch": 8.859223300970873,
      "grad_norm": 0.506313145160675,
      "learning_rate": 0.0001671116504854369,
      "loss": 0.2488,
      "step": 14600
    },
    {
      "epoch": 8.919902912621358,
      "grad_norm": 0.579694390296936,
      "learning_rate": 0.00016620145631067958,
      "loss": 0.252,
      "step": 14700
    },
    {
      "epoch": 8.980582524271846,
      "grad_norm": 1.048781156539917,
      "learning_rate": 0.00016529126213592232,
      "loss": 0.2569,
      "step": 14800
    },
    {
      "epoch": 9.04126213592233,
      "grad_norm": 0.4762475788593292,
      "learning_rate": 0.00016438106796116503,
      "loss": 0.247,
      "step": 14900
    },
    {
      "epoch": 9.101941747572816,
      "grad_norm": 0.5784612894058228,
      "learning_rate": 0.00016347087378640776,
      "loss": 0.249,
      "step": 15000
    },
    {
      "epoch": 9.101941747572816,
      "eval_accuracy": 0.7119295026351163,
      "eval_f1_macro": 0.48462198319681066,
      "eval_loss": 0.2577269971370697,
      "eval_precision": 0.6220453203620933,
      "eval_recall": 0.45259946630146386,
      "eval_runtime": 63.7117,
      "eval_samples_per_second": 735.611,
      "eval_steps_per_second": 5.76,
      "step": 15000
    },
    {
      "epoch": 9.162621359223301,
      "grad_norm": 1.0214685201644897,
      "learning_rate": 0.00016256067961165045,
      "loss": 0.2449,
      "step": 15100
    },
    {
      "epoch": 9.223300970873787,
      "grad_norm": 0.5309824347496033,
      "learning_rate": 0.00016165048543689318,
      "loss": 0.2456,
      "step": 15200
    },
    {
      "epoch": 9.283980582524272,
      "grad_norm": 2.5317320823669434,
      "learning_rate": 0.00016074029126213592,
      "loss": 0.2428,
      "step": 15300
    },
    {
      "epoch": 9.344660194174757,
      "grad_norm": 0.588573157787323,
      "learning_rate": 0.00015983009708737863,
      "loss": 0.2451,
      "step": 15400
    },
    {
      "epoch": 9.405339805825243,
      "grad_norm": 0.687719464302063,
      "learning_rate": 0.00015891990291262134,
      "loss": 0.2503,
      "step": 15500
    },
    {
      "epoch": 9.466019417475728,
      "grad_norm": 1.0443726778030396,
      "learning_rate": 0.00015800970873786405,
      "loss": 0.2485,
      "step": 15600
    },
    {
      "epoch": 9.526699029126213,
      "grad_norm": 1.1230727434158325,
      "learning_rate": 0.00015709951456310679,
      "loss": 0.2479,
      "step": 15700
    },
    {
      "epoch": 9.587378640776699,
      "grad_norm": 0.6641112565994263,
      "learning_rate": 0.00015618932038834952,
      "loss": 0.2489,
      "step": 15800
    },
    {
      "epoch": 9.648058252427184,
      "grad_norm": 0.4918789863586426,
      "learning_rate": 0.0001552791262135922,
      "loss": 0.2481,
      "step": 15900
    },
    {
      "epoch": 9.70873786407767,
      "grad_norm": 0.7862838506698608,
      "learning_rate": 0.00015436893203883494,
      "loss": 0.2481,
      "step": 16000
    },
    {
      "epoch": 9.70873786407767,
      "eval_accuracy": 0.713487101798707,
      "eval_f1_macro": 0.4839261840621368,
      "eval_loss": 0.2561655640602112,
      "eval_precision": 0.6345013597016592,
      "eval_recall": 0.4492915653199398,
      "eval_runtime": 63.9745,
      "eval_samples_per_second": 732.589,
      "eval_steps_per_second": 5.737,
      "step": 16000
    },
    {
      "epoch": 9.769417475728154,
      "grad_norm": 0.8769936561584473,
      "learning_rate": 0.00015345873786407765,
      "loss": 0.2479,
      "step": 16100
    },
    {
      "epoch": 9.830097087378642,
      "grad_norm": 0.9663941860198975,
      "learning_rate": 0.0001525485436893204,
      "loss": 0.2548,
      "step": 16200
    },
    {
      "epoch": 9.890776699029127,
      "grad_norm": 0.4475904405117035,
      "learning_rate": 0.00015163834951456307,
      "loss": 0.2504,
      "step": 16300
    },
    {
      "epoch": 9.951456310679612,
      "grad_norm": 0.9180653691291809,
      "learning_rate": 0.0001507281553398058,
      "loss": 0.246,
      "step": 16400
    },
    {
      "epoch": 10.012135922330097,
      "grad_norm": 1.1535005569458008,
      "learning_rate": 0.00014981796116504852,
      "loss": 0.2463,
      "step": 16500
    },
    {
      "epoch": 10.072815533980583,
      "grad_norm": 0.5621392130851746,
      "learning_rate": 0.00014890776699029125,
      "loss": 0.2432,
      "step": 16600
    },
    {
      "epoch": 10.133495145631068,
      "grad_norm": 0.5198004245758057,
      "learning_rate": 0.00014799757281553396,
      "loss": 0.2463,
      "step": 16700
    },
    {
      "epoch": 10.194174757281553,
      "grad_norm": 0.7625685930252075,
      "learning_rate": 0.00014708737864077667,
      "loss": 0.2474,
      "step": 16800
    },
    {
      "epoch": 10.254854368932039,
      "grad_norm": 0.4673856794834137,
      "learning_rate": 0.0001461771844660194,
      "loss": 0.244,
      "step": 16900
    },
    {
      "epoch": 10.315533980582524,
      "grad_norm": 0.6320896744728088,
      "learning_rate": 0.00014526699029126212,
      "loss": 0.241,
      "step": 17000
    },
    {
      "epoch": 10.315533980582524,
      "eval_accuracy": 0.7026265816032603,
      "eval_f1_macro": 0.4873663248833824,
      "eval_loss": 0.2613597810268402,
      "eval_precision": 0.5926410330413402,
      "eval_recall": 0.46290412629796984,
      "eval_runtime": 63.8686,
      "eval_samples_per_second": 733.803,
      "eval_steps_per_second": 5.746,
      "step": 17000
    },
    {
      "epoch": 10.37621359223301,
      "grad_norm": 1.3767356872558594,
      "learning_rate": 0.00014435679611650483,
      "loss": 0.2431,
      "step": 17100
    },
    {
      "epoch": 10.436893203883495,
      "grad_norm": 0.6615713834762573,
      "learning_rate": 0.00014344660194174757,
      "loss": 0.2401,
      "step": 17200
    },
    {
      "epoch": 10.49757281553398,
      "grad_norm": 0.5104607343673706,
      "learning_rate": 0.00014253640776699028,
      "loss": 0.2467,
      "step": 17300
    },
    {
      "epoch": 10.558252427184467,
      "grad_norm": 1.2690117359161377,
      "learning_rate": 0.00014162621359223301,
      "loss": 0.2474,
      "step": 17400
    },
    {
      "epoch": 10.618932038834952,
      "grad_norm": 0.5738195180892944,
      "learning_rate": 0.00014071601941747572,
      "loss": 0.2508,
      "step": 17500
    },
    {
      "epoch": 10.679611650485437,
      "grad_norm": 0.9724490642547607,
      "learning_rate": 0.00013980582524271843,
      "loss": 0.2457,
      "step": 17600
    },
    {
      "epoch": 10.740291262135923,
      "grad_norm": 0.5149191617965698,
      "learning_rate": 0.00013889563106796114,
      "loss": 0.2453,
      "step": 17700
    },
    {
      "epoch": 10.800970873786408,
      "grad_norm": 0.4608076512813568,
      "learning_rate": 0.00013798543689320388,
      "loss": 0.2461,
      "step": 17800
    },
    {
      "epoch": 10.861650485436893,
      "grad_norm": 0.7955921292304993,
      "learning_rate": 0.0001370752427184466,
      "loss": 0.2427,
      "step": 17900
    },
    {
      "epoch": 10.922330097087379,
      "grad_norm": 0.75131756067276,
      "learning_rate": 0.00013616504854368933,
      "loss": 0.2441,
      "step": 18000
    },
    {
      "epoch": 10.922330097087379,
      "eval_accuracy": 0.712249557257772,
      "eval_f1_macro": 0.48099926256661046,
      "eval_loss": 0.2561965882778168,
      "eval_precision": 0.6261557755424023,
      "eval_recall": 0.44725199588180664,
      "eval_runtime": 63.857,
      "eval_samples_per_second": 733.937,
      "eval_steps_per_second": 5.747,
      "step": 18000
    },
    {
      "epoch": 10.983009708737864,
      "grad_norm": 0.791315495967865,
      "learning_rate": 0.00013525485436893204,
      "loss": 0.2486,
      "step": 18100
    },
    {
      "epoch": 11.04368932038835,
      "grad_norm": 0.7312906980514526,
      "learning_rate": 0.00013434466019417475,
      "loss": 0.2431,
      "step": 18200
    },
    {
      "epoch": 11.104368932038835,
      "grad_norm": 0.609377384185791,
      "learning_rate": 0.00013343446601941746,
      "loss": 0.2399,
      "step": 18300
    },
    {
      "epoch": 11.16504854368932,
      "grad_norm": 0.5242405533790588,
      "learning_rate": 0.0001325242718446602,
      "loss": 0.2424,
      "step": 18400
    },
    {
      "epoch": 11.225728155339805,
      "grad_norm": 0.6860948204994202,
      "learning_rate": 0.0001316140776699029,
      "loss": 0.2434,
      "step": 18500
    },
    {
      "epoch": 11.28640776699029,
      "grad_norm": 2.105470657348633,
      "learning_rate": 0.0001307038834951456,
      "loss": 0.2393,
      "step": 18600
    },
    {
      "epoch": 11.347087378640778,
      "grad_norm": 0.990110456943512,
      "learning_rate": 0.00012979368932038835,
      "loss": 0.2446,
      "step": 18700
    },
    {
      "epoch": 11.407766990291263,
      "grad_norm": 0.5341915488243103,
      "learning_rate": 0.00012888349514563106,
      "loss": 0.2449,
      "step": 18800
    },
    {
      "epoch": 11.468446601941748,
      "grad_norm": 1.2902977466583252,
      "learning_rate": 0.00012797330097087377,
      "loss": 0.245,
      "step": 18900
    },
    {
      "epoch": 11.529126213592233,
      "grad_norm": 0.5241170525550842,
      "learning_rate": 0.00012706310679611648,
      "loss": 0.2433,
      "step": 19000
    },
    {
      "epoch": 11.529126213592233,
      "eval_accuracy": 0.7136791345723004,
      "eval_f1_macro": 0.47974793179174946,
      "eval_loss": 0.25500378012657166,
      "eval_precision": 0.5467676042682321,
      "eval_recall": 0.44681795057266877,
      "eval_runtime": 63.4183,
      "eval_samples_per_second": 739.014,
      "eval_steps_per_second": 5.787,
      "step": 19000
    },
    {
      "epoch": 11.589805825242719,
      "grad_norm": 1.6583833694458008,
      "learning_rate": 0.00012615291262135922,
      "loss": 0.2444,
      "step": 19100
    },
    {
      "epoch": 11.650485436893204,
      "grad_norm": 0.5068759322166443,
      "learning_rate": 0.00012524271844660192,
      "loss": 0.2417,
      "step": 19200
    },
    {
      "epoch": 11.71116504854369,
      "grad_norm": 0.8810881972312927,
      "learning_rate": 0.00012433252427184466,
      "loss": 0.2443,
      "step": 19300
    },
    {
      "epoch": 11.771844660194175,
      "grad_norm": 1.347113847732544,
      "learning_rate": 0.00012342233009708737,
      "loss": 0.2408,
      "step": 19400
    },
    {
      "epoch": 11.83252427184466,
      "grad_norm": 0.5012452602386475,
      "learning_rate": 0.00012251213592233008,
      "loss": 0.2443,
      "step": 19500
    },
    {
      "epoch": 11.893203883495145,
      "grad_norm": 0.6998531222343445,
      "learning_rate": 0.0001216019417475728,
      "loss": 0.2449,
      "step": 19600
    },
    {
      "epoch": 11.95388349514563,
      "grad_norm": 0.9693962335586548,
      "learning_rate": 0.00012069174757281553,
      "loss": 0.2415,
      "step": 19700
    },
    {
      "epoch": 12.014563106796116,
      "grad_norm": 0.5657163858413696,
      "learning_rate": 0.00011978155339805824,
      "loss": 0.2416,
      "step": 19800
    },
    {
      "epoch": 12.075242718446601,
      "grad_norm": 0.527268648147583,
      "learning_rate": 0.00011887135922330096,
      "loss": 0.2377,
      "step": 19900
    },
    {
      "epoch": 12.135922330097088,
      "grad_norm": 0.4967799782752991,
      "learning_rate": 0.00011796116504854367,
      "loss": 0.2436,
      "step": 20000
    },
    {
      "epoch": 12.135922330097088,
      "eval_accuracy": 0.7124629270062091,
      "eval_f1_macro": 0.48898241670534826,
      "eval_loss": 0.2549245059490204,
      "eval_precision": 0.6280111982698415,
      "eval_recall": 0.45813565668329703,
      "eval_runtime": 63.6697,
      "eval_samples_per_second": 736.095,
      "eval_steps_per_second": 5.764,
      "step": 20000
    },
    {
      "epoch": 12.196601941747574,
      "grad_norm": 0.5225503444671631,
      "learning_rate": 0.00011705097087378641,
      "loss": 0.2405,
      "step": 20100
    },
    {
      "epoch": 12.257281553398059,
      "grad_norm": 0.7009778022766113,
      "learning_rate": 0.00011614077669902912,
      "loss": 0.2375,
      "step": 20200
    },
    {
      "epoch": 12.317961165048544,
      "grad_norm": 0.5107778310775757,
      "learning_rate": 0.00011523058252427184,
      "loss": 0.2396,
      "step": 20300
    },
    {
      "epoch": 12.37864077669903,
      "grad_norm": 0.5113606452941895,
      "learning_rate": 0.00011432038834951455,
      "loss": 0.2379,
      "step": 20400
    },
    {
      "epoch": 12.439320388349515,
      "grad_norm": 0.6248862743377686,
      "learning_rate": 0.00011341019417475727,
      "loss": 0.2453,
      "step": 20500
    },
    {
      "epoch": 12.5,
      "grad_norm": 0.5648460984230042,
      "learning_rate": 0.0001125,
      "loss": 0.2391,
      "step": 20600
    },
    {
      "epoch": 12.560679611650485,
      "grad_norm": 0.7846024632453918,
      "learning_rate": 0.0001115898058252427,
      "loss": 0.2445,
      "step": 20700
    },
    {
      "epoch": 12.62135922330097,
      "grad_norm": 0.8849174976348877,
      "learning_rate": 0.00011067961165048543,
      "loss": 0.2392,
      "step": 20800
    },
    {
      "epoch": 12.682038834951456,
      "grad_norm": 0.5515842437744141,
      "learning_rate": 0.00010976941747572814,
      "loss": 0.2429,
      "step": 20900
    },
    {
      "epoch": 12.742718446601941,
      "grad_norm": 0.9129722118377686,
      "learning_rate": 0.00010885922330097086,
      "loss": 0.2459,
      "step": 21000
    },
    {
      "epoch": 12.742718446601941,
      "eval_accuracy": 0.7137218085219877,
      "eval_f1_macro": 0.49319871257798575,
      "eval_loss": 0.25367996096611023,
      "eval_precision": 0.6236252217705004,
      "eval_recall": 0.4640439902470517,
      "eval_runtime": 63.865,
      "eval_samples_per_second": 733.844,
      "eval_steps_per_second": 5.746,
      "step": 21000
    },
    {
      "epoch": 12.803398058252426,
      "grad_norm": 0.8533498048782349,
      "learning_rate": 0.00010794902912621357,
      "loss": 0.2384,
      "step": 21100
    },
    {
      "epoch": 12.864077669902912,
      "grad_norm": 0.5744712352752686,
      "learning_rate": 0.00010703883495145631,
      "loss": 0.241,
      "step": 21200
    },
    {
      "epoch": 12.924757281553399,
      "grad_norm": 0.9214051961898804,
      "learning_rate": 0.00010612864077669902,
      "loss": 0.2437,
      "step": 21300
    },
    {
      "epoch": 12.985436893203884,
      "grad_norm": 0.561357319355011,
      "learning_rate": 0.00010521844660194174,
      "loss": 0.2394,
      "step": 21400
    },
    {
      "epoch": 13.04611650485437,
      "grad_norm": 0.746048629283905,
      "learning_rate": 0.00010430825242718445,
      "loss": 0.2367,
      "step": 21500
    },
    {
      "epoch": 13.106796116504855,
      "grad_norm": 0.5449076294898987,
      "learning_rate": 0.00010339805825242718,
      "loss": 0.2392,
      "step": 21600
    },
    {
      "epoch": 13.16747572815534,
      "grad_norm": 0.7307832837104797,
      "learning_rate": 0.00010248786407766989,
      "loss": 0.2379,
      "step": 21700
    },
    {
      "epoch": 13.228155339805825,
      "grad_norm": 0.9420238137245178,
      "learning_rate": 0.00010157766990291262,
      "loss": 0.2348,
      "step": 21800
    },
    {
      "epoch": 13.28883495145631,
      "grad_norm": 0.43917521834373474,
      "learning_rate": 0.00010066747572815533,
      "loss": 0.2375,
      "step": 21900
    },
    {
      "epoch": 13.349514563106796,
      "grad_norm": 0.5659081339836121,
      "learning_rate": 9.975728155339806e-05,
      "loss": 0.24,
      "step": 22000
    },
    {
      "epoch": 13.349514563106796,
      "eval_accuracy": 0.7130176883521454,
      "eval_f1_macro": 0.4889124257466886,
      "eval_loss": 0.254099041223526,
      "eval_precision": 0.6244022587227488,
      "eval_recall": 0.4579878521374155,
      "eval_runtime": 63.8542,
      "eval_samples_per_second": 733.968,
      "eval_steps_per_second": 5.747,
      "step": 22000
    },
    {
      "epoch": 13.410194174757281,
      "grad_norm": 0.6600030660629272,
      "learning_rate": 9.884708737864077e-05,
      "loss": 0.2355,
      "step": 22100
    },
    {
      "epoch": 13.470873786407767,
      "grad_norm": 0.49261102080345154,
      "learning_rate": 9.793689320388349e-05,
      "loss": 0.2388,
      "step": 22200
    },
    {
      "epoch": 13.531553398058252,
      "grad_norm": 0.9407384395599365,
      "learning_rate": 9.70266990291262e-05,
      "loss": 0.2384,
      "step": 22300
    },
    {
      "epoch": 13.592233009708737,
      "grad_norm": 0.9749634265899658,
      "learning_rate": 9.611650485436893e-05,
      "loss": 0.2392,
      "step": 22400
    },
    {
      "epoch": 13.652912621359224,
      "grad_norm": 0.5800509452819824,
      "learning_rate": 9.520631067961164e-05,
      "loss": 0.2373,
      "step": 22500
    },
    {
      "epoch": 13.71359223300971,
      "grad_norm": 0.7015953063964844,
      "learning_rate": 9.429611650485437e-05,
      "loss": 0.2384,
      "step": 22600
    },
    {
      "epoch": 13.774271844660195,
      "grad_norm": 0.5446757078170776,
      "learning_rate": 9.338592233009708e-05,
      "loss": 0.2399,
      "step": 22700
    },
    {
      "epoch": 13.83495145631068,
      "grad_norm": 1.1201865673065186,
      "learning_rate": 9.247572815533979e-05,
      "loss": 0.2423,
      "step": 22800
    },
    {
      "epoch": 13.895631067961165,
      "grad_norm": 0.789397656917572,
      "learning_rate": 9.156553398058251e-05,
      "loss": 0.2401,
      "step": 22900
    },
    {
      "epoch": 13.95631067961165,
      "grad_norm": 0.9811580777168274,
      "learning_rate": 9.065533980582523e-05,
      "loss": 0.2429,
      "step": 23000
    },
    {
      "epoch": 13.95631067961165,
      "eval_accuracy": 0.7146819723899546,
      "eval_f1_macro": 0.4950660124078677,
      "eval_loss": 0.2540215849876404,
      "eval_precision": 0.6269585302337652,
      "eval_recall": 0.4640707919907739,
      "eval_runtime": 63.4303,
      "eval_samples_per_second": 738.874,
      "eval_steps_per_second": 5.786,
      "step": 23000
    },
    {
      "epoch": 14.016990291262136,
      "grad_norm": 0.6500277519226074,
      "learning_rate": 8.974514563106796e-05,
      "loss": 0.2399,
      "step": 23100
    },
    {
      "epoch": 14.077669902912621,
      "grad_norm": 0.7648112177848816,
      "learning_rate": 8.883495145631067e-05,
      "loss": 0.2355,
      "step": 23200
    },
    {
      "epoch": 14.138349514563107,
      "grad_norm": 0.6670911312103271,
      "learning_rate": 8.792475728155339e-05,
      "loss": 0.2371,
      "step": 23300
    },
    {
      "epoch": 14.199029126213592,
      "grad_norm": 0.8316827416419983,
      "learning_rate": 8.70145631067961e-05,
      "loss": 0.2355,
      "step": 23400
    },
    {
      "epoch": 14.259708737864077,
      "grad_norm": 0.4973256587982178,
      "learning_rate": 8.610436893203884e-05,
      "loss": 0.24,
      "step": 23500
    },
    {
      "epoch": 14.320388349514563,
      "grad_norm": 0.6986468434333801,
      "learning_rate": 8.519417475728155e-05,
      "loss": 0.2319,
      "step": 23600
    },
    {
      "epoch": 14.381067961165048,
      "grad_norm": 0.5316904783248901,
      "learning_rate": 8.428398058252427e-05,
      "loss": 0.2334,
      "step": 23700
    },
    {
      "epoch": 14.441747572815533,
      "grad_norm": 0.5285237431526184,
      "learning_rate": 8.337378640776698e-05,
      "loss": 0.2372,
      "step": 23800
    },
    {
      "epoch": 14.50242718446602,
      "grad_norm": 0.7617068290710449,
      "learning_rate": 8.24635922330097e-05,
      "loss": 0.2434,
      "step": 23900
    },
    {
      "epoch": 14.563106796116505,
      "grad_norm": 0.4870689809322357,
      "learning_rate": 8.155339805825241e-05,
      "loss": 0.2388,
      "step": 24000
    },
    {
      "epoch": 14.563106796116505,
      "eval_accuracy": 0.7128896665030832,
      "eval_f1_macro": 0.48631927316876294,
      "eval_loss": 0.2543109357357025,
      "eval_precision": 0.6364589670810381,
      "eval_recall": 0.451184525833245,
      "eval_runtime": 63.5544,
      "eval_samples_per_second": 737.432,
      "eval_steps_per_second": 5.775,
      "step": 24000
    },
    {
      "epoch": 14.62378640776699,
      "grad_norm": 0.48867112398147583,
      "learning_rate": 8.064320388349515e-05,
      "loss": 0.2424,
      "step": 24100
    },
    {
      "epoch": 14.684466019417476,
      "grad_norm": 1.3717137575149536,
      "learning_rate": 7.973300970873786e-05,
      "loss": 0.2373,
      "step": 24200
    },
    {
      "epoch": 14.745145631067961,
      "grad_norm": 0.5244446396827698,
      "learning_rate": 7.882281553398058e-05,
      "loss": 0.2406,
      "step": 24300
    },
    {
      "epoch": 14.805825242718447,
      "grad_norm": 0.49141696095466614,
      "learning_rate": 7.791262135922329e-05,
      "loss": 0.2345,
      "step": 24400
    },
    {
      "epoch": 14.866504854368932,
      "grad_norm": 0.5758472681045532,
      "learning_rate": 7.700242718446602e-05,
      "loss": 0.2372,
      "step": 24500
    },
    {
      "epoch": 14.927184466019417,
      "grad_norm": 0.4706755578517914,
      "learning_rate": 7.609223300970873e-05,
      "loss": 0.2363,
      "step": 24600
    },
    {
      "epoch": 14.987864077669903,
      "grad_norm": 1.1661343574523926,
      "learning_rate": 7.518203883495146e-05,
      "loss": 0.2363,
      "step": 24700
    },
    {
      "epoch": 15.048543689320388,
      "grad_norm": 0.6530236005783081,
      "learning_rate": 7.427184466019417e-05,
      "loss": 0.2371,
      "step": 24800
    },
    {
      "epoch": 15.109223300970873,
      "grad_norm": 0.6235638856887817,
      "learning_rate": 7.336165048543688e-05,
      "loss": 0.2365,
      "step": 24900
    },
    {
      "epoch": 15.169902912621358,
      "grad_norm": 1.0099172592163086,
      "learning_rate": 7.24514563106796e-05,
      "loss": 0.2333,
      "step": 25000
    },
    {
      "epoch": 15.169902912621358,
      "eval_accuracy": 0.7117374698615231,
      "eval_f1_macro": 0.4930065435769275,
      "eval_loss": 0.25366273522377014,
      "eval_precision": 0.6254110142417604,
      "eval_recall": 0.4636190323809964,
      "eval_runtime": 63.6634,
      "eval_samples_per_second": 736.168,
      "eval_steps_per_second": 5.765,
      "step": 25000
    },
    {
      "epoch": 15.230582524271846,
      "grad_norm": 0.4479919672012329,
      "learning_rate": 7.154126213592233e-05,
      "loss": 0.2363,
      "step": 25100
    },
    {
      "epoch": 15.29126213592233,
      "grad_norm": 0.5574977993965149,
      "learning_rate": 7.063106796116504e-05,
      "loss": 0.2323,
      "step": 25200
    },
    {
      "epoch": 15.351941747572816,
      "grad_norm": 0.815531849861145,
      "learning_rate": 6.972087378640776e-05,
      "loss": 0.2378,
      "step": 25300
    },
    {
      "epoch": 15.412621359223301,
      "grad_norm": 0.47219938039779663,
      "learning_rate": 6.881067961165048e-05,
      "loss": 0.2363,
      "step": 25400
    },
    {
      "epoch": 15.473300970873787,
      "grad_norm": 1.4584532976150513,
      "learning_rate": 6.79004854368932e-05,
      "loss": 0.2323,
      "step": 25500
    },
    {
      "epoch": 15.533980582524272,
      "grad_norm": 0.5099394917488098,
      "learning_rate": 6.699029126213592e-05,
      "loss": 0.2343,
      "step": 25600
    },
    {
      "epoch": 15.594660194174757,
      "grad_norm": 0.624839186668396,
      "learning_rate": 6.608009708737864e-05,
      "loss": 0.2357,
      "step": 25700
    },
    {
      "epoch": 15.655339805825243,
      "grad_norm": 0.9318602085113525,
      "learning_rate": 6.516990291262135e-05,
      "loss": 0.2331,
      "step": 25800
    },
    {
      "epoch": 15.716019417475728,
      "grad_norm": 0.6715418100357056,
      "learning_rate": 6.425970873786407e-05,
      "loss": 0.2365,
      "step": 25900
    },
    {
      "epoch": 15.776699029126213,
      "grad_norm": 0.48603758215904236,
      "learning_rate": 6.334951456310678e-05,
      "loss": 0.2392,
      "step": 26000
    },
    {
      "epoch": 15.776699029126213,
      "eval_accuracy": 0.7145326135660486,
      "eval_f1_macro": 0.49345940490874235,
      "eval_loss": 0.2531893253326416,
      "eval_precision": 0.6264577154052439,
      "eval_recall": 0.4631683363045007,
      "eval_runtime": 63.9314,
      "eval_samples_per_second": 733.082,
      "eval_steps_per_second": 5.741,
      "step": 26000
    },
    {
      "epoch": 15.837378640776699,
      "grad_norm": 1.03753662109375,
      "learning_rate": 6.243932038834951e-05,
      "loss": 0.234,
      "step": 26100
    },
    {
      "epoch": 15.898058252427184,
      "grad_norm": 0.7353742718696594,
      "learning_rate": 6.152912621359223e-05,
      "loss": 0.2381,
      "step": 26200
    },
    {
      "epoch": 15.95873786407767,
      "grad_norm": 0.8260138034820557,
      "learning_rate": 6.061893203883495e-05,
      "loss": 0.2394,
      "step": 26300
    },
    {
      "epoch": 16.019417475728154,
      "grad_norm": 0.7960408926010132,
      "learning_rate": 5.9708737864077663e-05,
      "loss": 0.2377,
      "step": 26400
    },
    {
      "epoch": 16.08009708737864,
      "grad_norm": 0.5640716552734375,
      "learning_rate": 5.879854368932038e-05,
      "loss": 0.2374,
      "step": 26500
    },
    {
      "epoch": 16.140776699029125,
      "grad_norm": 0.8281972408294678,
      "learning_rate": 5.78883495145631e-05,
      "loss": 0.2369,
      "step": 26600
    },
    {
      "epoch": 16.20145631067961,
      "grad_norm": 0.49466079473495483,
      "learning_rate": 5.697815533980582e-05,
      "loss": 0.2352,
      "step": 26700
    },
    {
      "epoch": 16.262135922330096,
      "grad_norm": 0.5278394818305969,
      "learning_rate": 5.6067961165048536e-05,
      "loss": 0.2321,
      "step": 26800
    },
    {
      "epoch": 16.32281553398058,
      "grad_norm": 1.5943635702133179,
      "learning_rate": 5.515776699029126e-05,
      "loss": 0.2368,
      "step": 26900
    },
    {
      "epoch": 16.383495145631066,
      "grad_norm": 0.7204076647758484,
      "learning_rate": 5.4247572815533976e-05,
      "loss": 0.2344,
      "step": 27000
    },
    {
      "epoch": 16.383495145631066,
      "eval_accuracy": 0.7135511127232381,
      "eval_f1_macro": 0.4959826756763837,
      "eval_loss": 0.2531072497367859,
      "eval_precision": 0.5967354989597543,
      "eval_recall": 0.4670103239766295,
      "eval_runtime": 63.8776,
      "eval_samples_per_second": 733.7,
      "eval_steps_per_second": 5.745,
      "step": 27000
    }
  ],
  "logging_steps": 100,
  "max_steps": 32960,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 1000,
  "total_flos": 1.8182410386600591e+18,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}