{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.994334277620396,
  "eval_steps": 500,
  "global_step": 3174,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "grad_norm": 1.3795244693756104,
      "learning_rate": 2.358490566037736e-06,
      "loss": 2.6953,
      "step": 25
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2704118490219116,
      "learning_rate": 4.716981132075472e-06,
      "loss": 2.7089,
      "step": 50
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1683423519134521,
      "learning_rate": 7.0754716981132075e-06,
      "loss": 2.6511,
      "step": 75
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2082363367080688,
      "learning_rate": 9.433962264150944e-06,
      "loss": 2.5193,
      "step": 100
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1215194463729858,
      "learning_rate": 1.179245283018868e-05,
      "loss": 2.306,
      "step": 125
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6498327255249023,
      "learning_rate": 1.4150943396226415e-05,
      "loss": 2.1279,
      "step": 150
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5128926038742065,
      "learning_rate": 1.650943396226415e-05,
      "loss": 1.979,
      "step": 175
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4196425676345825,
      "learning_rate": 1.8867924528301888e-05,
      "loss": 1.857,
      "step": 200
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39977598190307617,
      "learning_rate": 2.122641509433962e-05,
      "loss": 1.7905,
      "step": 225
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3468642830848694,
      "learning_rate": 2.358490566037736e-05,
      "loss": 1.7396,
      "step": 250
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3703348934650421,
      "learning_rate": 2.5943396226415097e-05,
      "loss": 1.6826,
      "step": 275
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3471335470676422,
      "learning_rate": 2.830188679245283e-05,
      "loss": 1.6333,
      "step": 300
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30928292870521545,
      "learning_rate": 2.9926470588235295e-05,
      "loss": 1.5934,
      "step": 325
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.33998557925224304,
      "learning_rate": 2.966386554621849e-05,
      "loss": 1.5503,
      "step": 350
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3642776906490326,
      "learning_rate": 2.940126050420168e-05,
      "loss": 1.5243,
      "step": 375
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.31012433767318726,
      "learning_rate": 2.9138655462184876e-05,
      "loss": 1.4618,
      "step": 400
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4258916974067688,
      "learning_rate": 2.8876050420168067e-05,
      "loss": 1.4161,
      "step": 425
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3025980591773987,
      "learning_rate": 2.8613445378151262e-05,
      "loss": 1.419,
      "step": 450
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3354116678237915,
      "learning_rate": 2.8350840336134453e-05,
      "loss": 1.3576,
      "step": 475
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3400489091873169,
      "learning_rate": 2.8088235294117648e-05,
      "loss": 1.3323,
      "step": 500
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.315164715051651,
      "learning_rate": 2.7825630252100843e-05,
      "loss": 1.344,
      "step": 525
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.3593141734600067,
      "learning_rate": 2.7563025210084034e-05,
      "loss": 1.3023,
      "step": 550
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.4317137598991394,
      "learning_rate": 2.730042016806723e-05,
      "loss": 1.3028,
      "step": 575
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.3506380617618561,
      "learning_rate": 2.703781512605042e-05,
      "loss": 1.3026,
      "step": 600
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.33726122975349426,
      "learning_rate": 2.6775210084033615e-05,
      "loss": 1.319,
      "step": 625
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.37094271183013916,
      "learning_rate": 2.6512605042016806e-05,
      "loss": 1.2583,
      "step": 650
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.41374334692955017,
      "learning_rate": 2.625e-05,
      "loss": 1.2486,
      "step": 675
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.46000728011131287,
      "learning_rate": 2.5987394957983196e-05,
      "loss": 1.2463,
      "step": 700
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.3955087661743164,
      "learning_rate": 2.5724789915966387e-05,
      "loss": 1.2397,
      "step": 725
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.4096736013889313,
      "learning_rate": 2.546218487394958e-05,
      "loss": 1.229,
      "step": 750
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.3845139741897583,
      "learning_rate": 2.5199579831932773e-05,
      "loss": 1.2314,
      "step": 775
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.4077882170677185,
      "learning_rate": 2.4936974789915968e-05,
      "loss": 1.2219,
      "step": 800
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.36021721363067627,
      "learning_rate": 2.467436974789916e-05,
      "loss": 1.234,
      "step": 825
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.36913222074508667,
      "learning_rate": 2.4411764705882354e-05,
      "loss": 1.1998,
      "step": 850
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.35471582412719727,
      "learning_rate": 2.414915966386555e-05,
      "loss": 1.1988,
      "step": 875
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.3558790683746338,
      "learning_rate": 2.3886554621848737e-05,
      "loss": 1.2106,
      "step": 900
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.36467084288597107,
      "learning_rate": 2.362394957983193e-05,
      "loss": 1.1717,
      "step": 925
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.381874680519104,
      "learning_rate": 2.3361344537815126e-05,
      "loss": 1.1896,
      "step": 950
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.3758748769760132,
      "learning_rate": 2.309873949579832e-05,
      "loss": 1.1712,
      "step": 975
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.35793235898017883,
      "learning_rate": 2.2836134453781513e-05,
      "loss": 1.1389,
      "step": 1000
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.44111478328704834,
      "learning_rate": 2.2573529411764707e-05,
      "loss": 1.1726,
      "step": 1025
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.3741939663887024,
      "learning_rate": 2.2310924369747902e-05,
      "loss": 1.1607,
      "step": 1050
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.3894720673561096,
      "learning_rate": 2.2048319327731093e-05,
      "loss": 1.2186,
      "step": 1075
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.3636987805366516,
      "learning_rate": 2.1785714285714285e-05,
      "loss": 1.1376,
      "step": 1100
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.42893752455711365,
      "learning_rate": 2.152310924369748e-05,
      "loss": 1.158,
      "step": 1125
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.3795158863067627,
      "learning_rate": 2.1260504201680674e-05,
      "loss": 1.1574,
      "step": 1150
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.36902275681495667,
      "learning_rate": 2.0997899159663866e-05,
      "loss": 1.1523,
      "step": 1175
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.431219220161438,
      "learning_rate": 2.073529411764706e-05,
      "loss": 1.1433,
      "step": 1200
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.4199659824371338,
      "learning_rate": 2.0472689075630252e-05,
      "loss": 1.1481,
      "step": 1225
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.6324878334999084,
      "learning_rate": 2.0210084033613447e-05,
      "loss": 1.1526,
      "step": 1250
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.523536205291748,
      "learning_rate": 1.9947478991596638e-05,
      "loss": 1.1216,
      "step": 1275
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.5140235424041748,
      "learning_rate": 1.9684873949579833e-05,
      "loss": 1.1539,
      "step": 1300
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.3695720136165619,
      "learning_rate": 1.9422268907563027e-05,
      "loss": 1.1666,
      "step": 1325
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.4080689251422882,
      "learning_rate": 1.915966386554622e-05,
      "loss": 1.1037,
      "step": 1350
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.35790908336639404,
      "learning_rate": 1.889705882352941e-05,
      "loss": 1.136,
      "step": 1375
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.42846861481666565,
      "learning_rate": 1.8634453781512605e-05,
      "loss": 1.1325,
      "step": 1400
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.37662366032600403,
      "learning_rate": 1.83718487394958e-05,
      "loss": 1.1439,
      "step": 1425
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.4963545501232147,
      "learning_rate": 1.810924369747899e-05,
      "loss": 1.1701,
      "step": 1450
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.4511197507381439,
      "learning_rate": 1.7846638655462186e-05,
      "loss": 1.1338,
      "step": 1475
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.44771987199783325,
      "learning_rate": 1.758403361344538e-05,
      "loss": 1.1021,
      "step": 1500
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.4158724248409271,
      "learning_rate": 1.7321428571428572e-05,
      "loss": 1.094,
      "step": 1525
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.43490564823150635,
      "learning_rate": 1.7058823529411763e-05,
      "loss": 1.1154,
      "step": 1550
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.4746383726596832,
      "learning_rate": 1.6796218487394958e-05,
      "loss": 1.1311,
      "step": 1575
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.4157463312149048,
      "learning_rate": 1.6533613445378153e-05,
      "loss": 1.1202,
      "step": 1600
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.38272300362586975,
      "learning_rate": 1.6271008403361344e-05,
      "loss": 1.1173,
      "step": 1625
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.5032052397727966,
      "learning_rate": 1.600840336134454e-05,
      "loss": 1.1313,
      "step": 1650
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.3842039704322815,
      "learning_rate": 1.5745798319327734e-05,
      "loss": 1.0984,
      "step": 1675
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.43160513043403625,
      "learning_rate": 1.5483193277310925e-05,
      "loss": 1.1108,
      "step": 1700
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.420173704624176,
      "learning_rate": 1.5220588235294118e-05,
      "loss": 1.144,
      "step": 1725
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.43490853905677795,
      "learning_rate": 1.4957983193277311e-05,
      "loss": 1.0752,
      "step": 1750
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.45708540081977844,
      "learning_rate": 1.4695378151260504e-05,
      "loss": 1.1447,
      "step": 1775
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.417322039604187,
      "learning_rate": 1.4432773109243699e-05,
      "loss": 1.102,
      "step": 1800
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.4371644854545593,
      "learning_rate": 1.417016806722689e-05,
      "loss": 1.1473,
      "step": 1825
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.4273310899734497,
      "learning_rate": 1.3907563025210085e-05,
      "loss": 1.0967,
      "step": 1850
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.5089781880378723,
      "learning_rate": 1.3644957983193278e-05,
      "loss": 1.1297,
      "step": 1875
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.48617228865623474,
      "learning_rate": 1.3382352941176471e-05,
      "loss": 1.0955,
      "step": 1900
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.4370473623275757,
      "learning_rate": 1.3119747899159664e-05,
      "loss": 1.0791,
      "step": 1925
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.4495941400527954,
      "learning_rate": 1.2857142857142857e-05,
      "loss": 1.0648,
      "step": 1950
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.4138700067996979,
      "learning_rate": 1.259453781512605e-05,
      "loss": 1.0948,
      "step": 1975
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.4161551296710968,
      "learning_rate": 1.2331932773109243e-05,
      "loss": 1.0947,
      "step": 2000
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.3938988745212555,
      "learning_rate": 1.2069327731092438e-05,
      "loss": 1.0863,
      "step": 2025
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.44733569025993347,
      "learning_rate": 1.180672268907563e-05,
      "loss": 1.1015,
      "step": 2050
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.4151917099952698,
      "learning_rate": 1.1544117647058824e-05,
      "loss": 1.0817,
      "step": 2075
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.45207536220550537,
      "learning_rate": 1.1281512605042017e-05,
      "loss": 1.0935,
      "step": 2100
    },
    {
      "epoch": 4.01,
      "grad_norm": 0.43334582448005676,
      "learning_rate": 1.1018907563025212e-05,
      "loss": 1.0843,
      "step": 2125
    },
    {
      "epoch": 4.06,
      "grad_norm": 0.44301116466522217,
      "learning_rate": 1.0756302521008403e-05,
      "loss": 1.0617,
      "step": 2150
    },
    {
      "epoch": 4.11,
      "grad_norm": 0.42584851384162903,
      "learning_rate": 1.0493697478991596e-05,
      "loss": 1.102,
      "step": 2175
    },
    {
      "epoch": 4.15,
      "grad_norm": 0.46070751547813416,
      "learning_rate": 1.0231092436974791e-05,
      "loss": 1.0943,
      "step": 2200
    },
    {
      "epoch": 4.2,
      "grad_norm": 0.43757393956184387,
      "learning_rate": 9.968487394957983e-06,
      "loss": 1.082,
      "step": 2225
    },
    {
      "epoch": 4.25,
      "grad_norm": 0.43552663922309875,
      "learning_rate": 9.705882352941177e-06,
      "loss": 1.1033,
      "step": 2250
    },
    {
      "epoch": 4.3,
      "grad_norm": 0.44868725538253784,
      "learning_rate": 9.44327731092437e-06,
      "loss": 1.0912,
      "step": 2275
    },
    {
      "epoch": 4.34,
      "grad_norm": 0.43542513251304626,
      "learning_rate": 9.180672268907563e-06,
      "loss": 1.1113,
      "step": 2300
    },
    {
      "epoch": 4.39,
      "grad_norm": 0.47481635212898254,
      "learning_rate": 8.918067226890756e-06,
      "loss": 1.0455,
      "step": 2325
    },
    {
      "epoch": 4.44,
      "grad_norm": 0.46137455105781555,
      "learning_rate": 8.65546218487395e-06,
      "loss": 1.0898,
      "step": 2350
    },
    {
      "epoch": 4.49,
      "grad_norm": 0.4473894536495209,
      "learning_rate": 8.392857142857143e-06,
      "loss": 1.0836,
      "step": 2375
    },
    {
      "epoch": 4.53,
      "grad_norm": 0.39784467220306396,
      "learning_rate": 8.130252100840336e-06,
      "loss": 1.0629,
      "step": 2400
    },
    {
      "epoch": 4.58,
      "grad_norm": 0.48481184244155884,
      "learning_rate": 7.86764705882353e-06,
      "loss": 1.1173,
      "step": 2425
    },
    {
      "epoch": 4.63,
      "grad_norm": 0.485196590423584,
      "learning_rate": 7.605042016806723e-06,
      "loss": 1.0673,
      "step": 2450
    },
    {
      "epoch": 4.67,
      "grad_norm": 0.5114961266517639,
      "learning_rate": 7.342436974789916e-06,
      "loss": 1.0877,
      "step": 2475
    },
    {
      "epoch": 4.72,
      "grad_norm": 0.4506637752056122,
      "learning_rate": 7.07983193277311e-06,
      "loss": 1.0995,
      "step": 2500
    },
    {
      "epoch": 4.77,
      "grad_norm": 0.45109784603118896,
      "learning_rate": 6.817226890756303e-06,
      "loss": 1.0819,
      "step": 2525
    },
    {
      "epoch": 4.82,
      "grad_norm": 0.4272564947605133,
      "learning_rate": 6.554621848739496e-06,
      "loss": 1.1109,
      "step": 2550
    },
    {
      "epoch": 4.86,
      "grad_norm": 0.4301404058933258,
      "learning_rate": 6.29201680672269e-06,
      "loss": 1.0738,
      "step": 2575
    },
    {
      "epoch": 4.91,
      "grad_norm": 0.49940961599349976,
      "learning_rate": 6.029411764705883e-06,
      "loss": 1.0865,
      "step": 2600
    },
    {
      "epoch": 4.96,
      "grad_norm": 0.41319113969802856,
      "learning_rate": 5.7773109243697485e-06,
      "loss": 1.0535,
      "step": 2625
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.4326096773147583,
      "learning_rate": 5.5147058823529415e-06,
      "loss": 1.0745,
      "step": 2650
    },
    {
      "epoch": 5.05,
      "grad_norm": 0.4360290765762329,
      "learning_rate": 5.252100840336135e-06,
      "loss": 1.0745,
      "step": 2675
    },
    {
      "epoch": 5.1,
      "grad_norm": 0.42354682087898254,
      "learning_rate": 4.989495798319328e-06,
      "loss": 1.0685,
      "step": 2700
    },
    {
      "epoch": 5.15,
      "grad_norm": 0.49250248074531555,
      "learning_rate": 4.726890756302521e-06,
      "loss": 1.0841,
      "step": 2725
    },
    {
      "epoch": 5.19,
      "grad_norm": 0.4505230784416199,
      "learning_rate": 4.464285714285715e-06,
      "loss": 1.0935,
      "step": 2750
    },
    {
      "epoch": 5.24,
      "grad_norm": 0.41872066259384155,
      "learning_rate": 4.201680672268908e-06,
      "loss": 1.0827,
      "step": 2775
    },
    {
      "epoch": 5.29,
      "grad_norm": 0.45635831356048584,
      "learning_rate": 3.939075630252101e-06,
      "loss": 1.0973,
      "step": 2800
    },
    {
      "epoch": 5.34,
      "grad_norm": 0.49893826246261597,
      "learning_rate": 3.6764705882352942e-06,
      "loss": 1.0859,
      "step": 2825
    },
    {
      "epoch": 5.38,
      "grad_norm": 0.5377572774887085,
      "learning_rate": 3.4138655462184873e-06,
      "loss": 1.088,
      "step": 2850
    },
    {
      "epoch": 5.43,
      "grad_norm": 0.45102909207344055,
      "learning_rate": 3.1512605042016808e-06,
      "loss": 1.0875,
      "step": 2875
    },
    {
      "epoch": 5.48,
      "grad_norm": 0.3922051191329956,
      "learning_rate": 2.8886554621848742e-06,
      "loss": 1.0708,
      "step": 2900
    },
    {
      "epoch": 5.52,
      "grad_norm": 0.4416084289550781,
      "learning_rate": 2.6260504201680673e-06,
      "loss": 1.0816,
      "step": 2925
    },
    {
      "epoch": 5.57,
      "grad_norm": 0.5171985626220703,
      "learning_rate": 2.3634453781512604e-06,
      "loss": 1.0859,
      "step": 2950
    },
    {
      "epoch": 5.62,
      "grad_norm": 0.4239521920681,
      "learning_rate": 2.100840336134454e-06,
      "loss": 1.0387,
      "step": 2975
    },
    {
      "epoch": 5.67,
      "grad_norm": 0.5627429485321045,
      "learning_rate": 1.8382352941176471e-06,
      "loss": 1.0818,
      "step": 3000
    },
    {
      "epoch": 5.71,
      "grad_norm": 0.4605351686477661,
      "learning_rate": 1.5756302521008404e-06,
      "loss": 1.0637,
      "step": 3025
    },
    {
      "epoch": 5.76,
      "grad_norm": 0.40121838450431824,
      "learning_rate": 1.3130252100840336e-06,
      "loss": 1.039,
      "step": 3050
    },
    {
      "epoch": 5.81,
      "grad_norm": 0.45940887928009033,
      "learning_rate": 1.050420168067227e-06,
      "loss": 1.0434,
      "step": 3075
    },
    {
      "epoch": 5.85,
      "grad_norm": 0.4496408998966217,
      "learning_rate": 7.878151260504202e-07,
      "loss": 1.1024,
      "step": 3100
    },
    {
      "epoch": 5.9,
      "grad_norm": 0.4458378553390503,
      "learning_rate": 5.252100840336135e-07,
      "loss": 1.0948,
      "step": 3125
    },
    {
      "epoch": 5.95,
      "grad_norm": 0.49208617210388184,
      "learning_rate": 2.6260504201680673e-07,
      "loss": 1.0673,
      "step": 3150
    }
  ],
  "logging_steps": 25,
  "max_steps": 3174,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 500,
  "total_flos": 1.3634839262527488e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}