{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.999697285661431,
  "eval_steps": 500,
  "global_step": 22296,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.9997567688496474e-05,
      "loss": 2.9132,
      "step": 100
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9990172715142793e-05,
      "loss": 0.0721,
      "step": 200
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.997781629993153e-05,
      "loss": 0.0757,
      "step": 300
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9960500896052476e-05,
      "loss": 0.0599,
      "step": 400
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9938477167054135e-05,
      "loss": 0.0353,
      "step": 500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9911304569050045e-05,
      "loss": 0.0256,
      "step": 600
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.987918618733232e-05,
      "loss": 0.0173,
      "step": 700
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.98421283985469e-05,
      "loss": 0.0276,
      "step": 800
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9800138559988466e-05,
      "loss": 0.0225,
      "step": 900
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.97532250081397e-05,
      "loss": 0.026,
      "step": 1000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.970139705701628e-05,
      "loss": 0.0271,
      "step": 1100
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.9644664996317616e-05,
      "loss": 0.0231,
      "step": 1200
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.958304008938407e-05,
      "loss": 0.0196,
      "step": 1300
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.951653457096072e-05,
      "loss": 0.0124,
      "step": 1400
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.944516164476834e-05,
      "loss": 0.0159,
      "step": 1500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.9368935480882034e-05,
      "loss": 0.0144,
      "step": 1600
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.9287871212917866e-05,
      "loss": 0.0182,
      "step": 1700
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.9201984935028426e-05,
      "loss": 0.018,
      "step": 1800
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.911129369870746e-05,
      "loss": 0.0156,
      "step": 1900
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.90158155094046e-05,
      "loss": 0.0146,
      "step": 2000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.8915569322950615e-05,
      "loss": 0.0196,
      "step": 2100
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.8810575041794e-05,
      "loss": 0.02,
      "step": 2200
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.8700853511049656e-05,
      "loss": 0.0143,
      "step": 2300
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.858642651436035e-05,
      "loss": 0.0111,
      "step": 2400
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.846731676957191e-05,
      "loss": 0.0179,
      "step": 2500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.834354792422293e-05,
      "loss": 0.0133,
      "step": 2600
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.821514455084985e-05,
      "loss": 0.0112,
      "step": 2700
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.8082132142108465e-05,
      "loss": 0.0131,
      "step": 2800
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.794453710571272e-05,
      "loss": 0.0134,
      "step": 2900
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.780238675919182e-05,
      "loss": 0.0084,
      "step": 3000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.765570932446672e-05,
      "loss": 0.0069,
      "step": 3100
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.75045339222471e-05,
      "loss": 0.0186,
      "step": 3200
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.734889056624983e-05,
      "loss": 0.0138,
      "step": 3300
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.718881015724017e-05,
      "loss": 0.012,
      "step": 3400
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.702432447689692e-05,
      "loss": 0.0121,
      "step": 3500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.6855466181502544e-05,
      "loss": 0.0104,
      "step": 3600
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.66822687954598e-05,
      "loss": 0.014,
      "step": 3700
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.65047667046359e-05,
      "loss": 0.0085,
      "step": 3800
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.632299514953571e-05,
      "loss": 0.0111,
      "step": 3900
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.613699021830524e-05,
      "loss": 0.0122,
      "step": 4000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.59467888395669e-05,
      "loss": 0.0082,
      "step": 4100
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.575242877508777e-05,
      "loss": 0.0087,
      "step": 4200
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.5553948612282607e-05,
      "loss": 0.0115,
      "step": 4300
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.5351387756552846e-05,
      "loss": 0.0069,
      "step": 4400
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.51447864234632e-05,
      "loss": 0.0099,
      "step": 4500
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.4934185630757484e-05,
      "loss": 0.0087,
      "step": 4600
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.4719627190215064e-05,
      "loss": 0.0096,
      "step": 4700
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.450115369934976e-05,
      "loss": 0.0122,
      "step": 4800
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.427880853295274e-05,
      "loss": 0.0076,
      "step": 4900
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.4052635834481025e-05,
      "loss": 0.0071,
      "step": 5000
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.3822680507293455e-05,
      "loss": 0.0105,
      "step": 5100
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.358898820573581e-05,
      "loss": 0.0108,
      "step": 5200
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.3351605326076724e-05,
      "loss": 0.0063,
      "step": 5300
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.3110578997296416e-05,
      "loss": 0.0147,
      "step": 5400
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.286595707172986e-05,
      "loss": 0.0135,
      "step": 5500
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.261778811556646e-05,
      "loss": 0.0105,
      "step": 5600
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.236612139920786e-05,
      "loss": 0.0131,
      "step": 5700
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.2111006887486035e-05,
      "loss": 0.0085,
      "step": 5800
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.185249522974346e-05,
      "loss": 0.0052,
      "step": 5900
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.159063774977748e-05,
      "loss": 0.008,
      "step": 6000
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.1325486435650625e-05,
      "loss": 0.011,
      "step": 6100
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.105709392936914e-05,
      "loss": 0.0074,
      "step": 6200
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.0785513516431705e-05,
      "loss": 0.0147,
      "step": 6300
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.051079911525031e-05,
      "loss": 0.0049,
      "step": 6400
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.023300526644557e-05,
      "loss": 0.0075,
      "step": 6500
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.995501009115527e-05,
      "loss": 0.0114,
      "step": 6600
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.967125281105033e-05,
      "loss": 0.0089,
      "step": 6700
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.93845827632495e-05,
      "loss": 0.0066,
      "step": 6800
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.909505686199625e-05,
      "loss": 0.0079,
      "step": 6900
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.880273258852296e-05,
      "loss": 0.008,
      "step": 7000
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.850766797963886e-05,
      "loss": 0.0068,
      "step": 7100
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.8209921616207645e-05,
      "loss": 0.0069,
      "step": 7200
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.790955261151704e-05,
      "loss": 0.0088,
      "step": 7300
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.7606620599542756e-05,
      "loss": 0.0078,
      "step": 7400
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.730118572310899e-05,
      "loss": 0.008,
      "step": 7500
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.699330862194794e-05,
      "loss": 0.0057,
      "step": 7600
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.668305042066061e-05,
      "loss": 0.0046,
      "step": 7700
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.637047271658145e-05,
      "loss": 0.0059,
      "step": 7800
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.605563756754904e-05,
      "loss": 0.0046,
      "step": 7900
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.573860747958544e-05,
      "loss": 0.0085,
      "step": 8000
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.541944539448648e-05,
      "loss": 0.0064,
      "step": 8100
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.509821467732553e-05,
      "loss": 0.0064,
      "step": 8200
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.477822117325554e-05,
      "loss": 0.0061,
      "step": 8300
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.445306400520726e-05,
      "loss": 0.0087,
      "step": 8400
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.4126030066319e-05,
      "loss": 0.0042,
      "step": 8500
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.379718428450832e-05,
      "loss": 0.0071,
      "step": 8600
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.346659194740827e-05,
      "loss": 0.0043,
      "step": 8700
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.313431868940551e-05,
      "loss": 0.0062,
      "step": 8800
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.280043047860958e-05,
      "loss": 0.0096,
      "step": 8900
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.24649936037558e-05,
      "loss": 0.0049,
      "step": 9000
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.21280746610446e-05,
      "loss": 0.0043,
      "step": 9100
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.1789740540919856e-05,
      "loss": 0.006,
      "step": 9200
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.145005841478868e-05,
      "loss": 0.0053,
      "step": 9300
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.1109095721685525e-05,
      "loss": 0.0075,
      "step": 9400
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.076692015488315e-05,
      "loss": 0.0048,
      "step": 9500
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.0423599648453037e-05,
      "loss": 0.0063,
      "step": 9600
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.007920236377807e-05,
      "loss": 0.0075,
      "step": 9700
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.9733796676020104e-05,
      "loss": 0.0082,
      "step": 9800
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.938745116054501e-05,
      "loss": 0.0054,
      "step": 9900
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.9040234579308025e-05,
      "loss": 0.0042,
      "step": 10000
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.8692215867202042e-05,
      "loss": 0.0112,
      "step": 10100
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.8343464118371603e-05,
      "loss": 0.0059,
      "step": 10200
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.7994048572495225e-05,
      "loss": 0.0034,
      "step": 10300
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.7644038601038813e-05,
      "loss": 0.0082,
      "step": 10400
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.7293503693483008e-05,
      "loss": 0.0047,
      "step": 10500
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6942513443527008e-05,
      "loss": 0.0024,
      "step": 10600
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.6591137535271742e-05,
      "loss": 0.0041,
      "step": 10700
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.6239445729385104e-05,
      "loss": 0.0041,
      "step": 10800
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5887507849251914e-05,
      "loss": 0.0059,
      "step": 10900
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.5535393767111533e-05,
      "loss": 0.0032,
      "step": 11000
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.5183173390185736e-05,
      "loss": 0.004,
      "step": 11100
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.483091664679959e-05,
      "loss": 0.0039,
      "step": 11200
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.44786934724982e-05,
      "loss": 0.004,
      "step": 11300
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4126573796162015e-05,
      "loss": 0.0033,
      "step": 11400
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.377462752612344e-05,
      "loss": 0.0091,
      "step": 11500
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.3422924536287548e-05,
      "loss": 0.0049,
      "step": 11600
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.307153465225958e-05,
      "loss": 0.0048,
      "step": 11700
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.272052763748209e-05,
      "loss": 0.0024,
      "step": 11800
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.2369973179384445e-05,
      "loss": 0.0058,
      "step": 11900
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.201994087554733e-05,
      "loss": 0.0098,
      "step": 12000
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.167050021988514e-05,
      "loss": 0.0063,
      "step": 12100
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.1321720588848995e-05,
      "loss": 0.0035,
      "step": 12200
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.097367122765301e-05,
      "loss": 0.0075,
      "step": 12300
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.0626421236526656e-05,
      "loss": 0.0048,
      "step": 12400
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.0280039556995902e-05,
      "loss": 0.0036,
      "step": 12500
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.9934594958195834e-05,
      "loss": 0.0039,
      "step": 12600
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.9590156023217553e-05,
      "loss": 0.0045,
      "step": 12700
    },
    {
      "epoch": 1.72,
      "learning_rate": 1.9250219243814994e-05,
      "loss": 0.0051,
      "step": 12800
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.8907984814652774e-05,
      "loss": 0.0033,
      "step": 12900
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.8566959868093257e-05,
      "loss": 0.0053,
      "step": 13000
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.8227212109768717e-05,
      "loss": 0.0018,
      "step": 13100
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.7888808991743948e-05,
      "loss": 0.0056,
      "step": 13200
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.7551817699124613e-05,
      "loss": 0.0054,
      "step": 13300
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.721630513671862e-05,
      "loss": 0.0087,
      "step": 13400
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.6882337915753093e-05,
      "loss": 0.0054,
      "step": 13500
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.65499823406497e-05,
      "loss": 0.0045,
      "step": 13600
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.621930439586083e-05,
      "loss": 0.0043,
      "step": 13700
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.5890369732769356e-05,
      "loss": 0.0031,
      "step": 13800
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.5563243656654454e-05,
      "loss": 0.0038,
      "step": 13900
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.5237991113726185e-05,
      "loss": 0.0018,
      "step": 14000
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.4914676678231354e-05,
      "loss": 0.0073,
      "step": 14100
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.4593364539633242e-05,
      "loss": 0.0049,
      "step": 14200
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.4274118489867638e-05,
      "loss": 0.0042,
      "step": 14300
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.3957001910677942e-05,
      "loss": 0.0047,
      "step": 14400
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.3642077761031574e-05,
      "loss": 0.0035,
      "step": 14500
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.3329408564620358e-05,
      "loss": 0.0054,
      "step": 14600
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.3019056397447415e-05,
      "loss": 0.0023,
      "step": 14700
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.2711082875502745e-05,
      "loss": 0.0057,
      "step": 14800
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.2405549142530348e-05,
      "loss": 0.0029,
      "step": 14900
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.2102515857888976e-05,
      "loss": 0.0027,
      "step": 15000
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.1802043184509044e-05,
      "loss": 0.0032,
      "step": 15100
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.1504190776948205e-05,
      "loss": 0.0019,
      "step": 15200
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.1209017769547758e-05,
      "loss": 0.0009,
      "step": 15300
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.0916582764692346e-05,
      "loss": 0.0009,
      "step": 15400
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.062694382117534e-05,
      "loss": 0.0009,
      "step": 15500
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.0340158442672049e-05,
      "loss": 0.0012,
      "step": 15600
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.005910772256064e-05,
      "loss": 0.0074,
      "step": 15700
    },
    {
      "epoch": 2.13,
      "learning_rate": 9.778169762000127e-06,
      "loss": 0.0012,
      "step": 15800
    },
    {
      "epoch": 2.14,
      "learning_rate": 9.500253878419744e-06,
      "loss": 0.0023,
      "step": 15900
    },
    {
      "epoch": 2.15,
      "learning_rate": 9.225415248048676e-06,
      "loss": 0.0013,
      "step": 16000
    },
    {
      "epoch": 2.17,
      "learning_rate": 8.956409824144358e-06,
      "loss": 0.0031,
      "step": 16100
    },
    {
      "epoch": 2.18,
      "learning_rate": 8.687856651695316e-06,
      "loss": 0.0029,
      "step": 16200
    },
    {
      "epoch": 2.19,
      "learning_rate": 8.422542022321477e-06,
      "loss": 0.0038,
      "step": 16300
    },
    {
      "epoch": 2.21,
      "learning_rate": 8.160518610453013e-06,
      "loss": 0.0032,
      "step": 16400
    },
    {
      "epoch": 2.22,
      "learning_rate": 7.901838437095879e-06,
      "loss": 0.0005,
      "step": 16500
    },
    {
      "epoch": 2.23,
      "learning_rate": 7.646552859503738e-06,
      "loss": 0.0021,
      "step": 16600
    },
    {
      "epoch": 2.25,
      "learning_rate": 7.3947125609818005e-06,
      "loss": 0.0007,
      "step": 16700
    },
    {
      "epoch": 2.26,
      "learning_rate": 7.146367540824281e-06,
      "loss": 0.002,
      "step": 16800
    },
    {
      "epoch": 2.27,
      "learning_rate": 6.9015671043878185e-06,
      "loss": 0.0028,
      "step": 16900
    },
    {
      "epoch": 2.29,
      "learning_rate": 6.660359853302589e-06,
      "loss": 0.0026,
      "step": 17000
    },
    {
      "epoch": 2.3,
      "learning_rate": 6.422793675823113e-06,
      "loss": 0.0004,
      "step": 17100
    },
    {
      "epoch": 2.31,
      "learning_rate": 6.188915737320777e-06,
      "loss": 0.0018,
      "step": 17200
    },
    {
      "epoch": 2.33,
      "learning_rate": 5.958772470919794e-06,
      "loss": 0.0043,
      "step": 17300
    },
    {
      "epoch": 2.34,
      "learning_rate": 5.732409568278555e-06,
      "loss": 0.0022,
      "step": 17400
    },
    {
      "epoch": 2.35,
      "learning_rate": 5.509871970518238e-06,
      "loss": 0.0002,
      "step": 17500
    },
    {
      "epoch": 2.37,
      "learning_rate": 5.291203859300376e-06,
      "loss": 0.0003,
      "step": 17600
    },
    {
      "epoch": 2.38,
      "learning_rate": 5.076448648055207e-06,
      "loss": 0.0014,
      "step": 17700
    },
    {
      "epoch": 2.39,
      "learning_rate": 4.86564897336258e-06,
      "loss": 0.0012,
      "step": 17800
    },
    {
      "epoch": 2.41,
      "learning_rate": 4.658846686487034e-06,
      "loss": 0.0035,
      "step": 17900
    },
    {
      "epoch": 2.42,
      "learning_rate": 4.4560828450688815e-06,
      "loss": 0.0014,
      "step": 18000
    },
    {
      "epoch": 2.44,
      "learning_rate": 4.2573977049727574e-06,
      "loss": 0.0012,
      "step": 18100
    },
    {
      "epoch": 2.45,
      "learning_rate": 4.06283071229539e-06,
      "loss": 0.0019,
      "step": 18200
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.872420495534187e-06,
      "loss": 0.0022,
      "step": 18300
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.686204857918046e-06,
      "loss": 0.0007,
      "step": 18400
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.5042207699021e-06,
      "loss": 0.0044,
      "step": 18500
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.3265043618277455e-06,
      "loss": 0.0018,
      "step": 18600
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.153090916749471e-06,
      "loss": 0.0003,
      "step": 18700
    },
    {
      "epoch": 2.53,
      "learning_rate": 2.9840148634299374e-06,
      "loss": 0.0007,
      "step": 18800
    },
    {
      "epoch": 2.54,
      "learning_rate": 2.8193097695046076e-06,
      "loss": 0.0017,
      "step": 18900
    },
    {
      "epoch": 2.56,
      "learning_rate": 2.6590083348174006e-06,
      "loss": 0.0008,
      "step": 19000
    },
    {
      "epoch": 2.57,
      "learning_rate": 2.503142384928589e-06,
      "loss": 0.0013,
      "step": 19100
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.351742864796275e-06,
      "loss": 0.0008,
      "step": 19200
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.204839832632749e-06,
      "loss": 0.0025,
      "step": 19300
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.062462453936828e-06,
      "loss": 0.0039,
      "step": 19400
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.9246389957034705e-06,
      "loss": 0.0008,
      "step": 19500
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.7913968208117833e-06,
      "loss": 0.0018,
      "step": 19600
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.664025834061389e-06,
      "loss": 0.0023,
      "step": 19700
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.53997821489342e-06,
      "loss": 0.0035,
      "step": 19800
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.4205882479710276e-06,
      "loss": 0.0034,
      "step": 19900
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.3058796364669029e-06,
      "loss": 0.0024,
      "step": 20000
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.1958751541374602e-06,
      "loss": 0.0012,
      "step": 20100
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.0905966408014807e-06,
      "loss": 0.0017,
      "step": 20200
    },
    {
      "epoch": 2.73,
      "learning_rate": 9.90064998004095e-07,
      "loss": 0.0034,
      "step": 20300
    },
    {
      "epoch": 2.74,
      "learning_rate": 8.943001848670724e-07,
      "loss": 0.0009,
      "step": 20400
    },
    {
      "epoch": 2.76,
      "learning_rate": 8.033212141262653e-07,
      "loss": 0.0019,
      "step": 20500
    },
    {
      "epoch": 2.77,
      "learning_rate": 7.171461483568537e-07,
      "loss": 0.0034,
      "step": 20600
    },
    {
      "epoch": 2.78,
      "learning_rate": 6.357920963873054e-07,
      "loss": 0.0037,
      "step": 20700
    },
    {
      "epoch": 2.8,
      "learning_rate": 5.592752099026599e-07,
      "loss": 0.0008,
      "step": 20800
    },
    {
      "epoch": 2.81,
      "learning_rate": 4.876106802378377e-07,
      "loss": 0.0005,
      "step": 20900
    },
    {
      "epoch": 2.83,
      "learning_rate": 4.2081273536161424e-07,
      "loss": 0.0017,
      "step": 21000
    },
    {
      "epoch": 2.84,
      "learning_rate": 3.5889463705187367e-07,
      "loss": 0.0016,
      "step": 21100
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.018686782626512e-07,
      "loss": 0.0013,
      "step": 21200
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.5024309873995513e-07,
      "loss": 0.0017,
      "step": 21300
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.0298522443275014e-07,
      "loss": 0.001,
      "step": 21400
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.6065044323308464e-07,
      "loss": 0.0014,
      "step": 21500
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.2324716010710114e-07,
      "loss": 0.002,
      "step": 21600
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.078280094243219e-08,
      "loss": 0.0013,
      "step": 21700
    },
    {
      "epoch": 2.93,
      "learning_rate": 6.326381107389646e-08,
      "loss": 0.0013,
      "step": 21800
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.0695654003883313e-08,
      "loss": 0.0028,
      "step": 21900
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.308281031763726e-08,
      "loss": 0.0023,
      "step": 22000
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.0428776793702798e-08,
      "loss": 0.0016,
      "step": 22100
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.7360657097047492e-09,
      "loss": 0.0042,
      "step": 22200
    },
    {
      "epoch": 3.0,
      "step": 22296,
      "total_flos": 9.011276227990733e+18,
      "train_loss": 0.020508435523056583,
      "train_runtime": 294350.325,
      "train_samples_per_second": 1.212,
      "train_steps_per_second": 0.076
    }
  ],
  "logging_steps": 100,
  "max_steps": 22296,
  "num_train_epochs": 3,
  "save_steps": 3000,
  "total_flos": 9.011276227990733e+18,
  "trial_name": null,
  "trial_params": null
}