{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.4926270156655391,
  "eval_steps": 99999,
  "global_step": 15000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.9800000000000004e-05,
      "loss": 3.0349,
      "step": 100
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.94e-05,
      "loss": 3.2422,
      "step": 200
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.92e-05,
      "loss": 3.2539,
      "step": 300
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.920000000000001e-05,
      "loss": 3.111,
      "step": 400
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.92e-05,
      "loss": 3.0833,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.967945507362518e-05,
      "loss": 3.0845,
      "step": 600
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.934555410865138e-05,
      "loss": 3.0598,
      "step": 700
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.901165314367759e-05,
      "loss": 3.0556,
      "step": 800
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.86777521787038e-05,
      "loss": 3.0666,
      "step": 900
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.834385121373001e-05,
      "loss": 3.0681,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.800995024875622e-05,
      "loss": 3.0349,
      "step": 1100
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.767604928378244e-05,
      "loss": 3.0354,
      "step": 1200
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.734214831880864e-05,
      "loss": 3.1831,
      "step": 1300
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.700824735383486e-05,
      "loss": 3.0632,
      "step": 1400
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.667434638886107e-05,
      "loss": 3.1388,
      "step": 1500
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.634044542388728e-05,
      "loss": 3.0572,
      "step": 1600
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.600654445891349e-05,
      "loss": 3.0525,
      "step": 1700
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.567264349393971e-05,
      "loss": 3.4111,
      "step": 1800
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.533874252896591e-05,
      "loss": 3.1258,
      "step": 1900
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.500484156399212e-05,
      "loss": 3.2196,
      "step": 2000
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.467094059901834e-05,
      "loss": 3.0257,
      "step": 2100
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.433703963404454e-05,
      "loss": 3.0797,
      "step": 2200
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.400313866907076e-05,
      "loss": 3.4158,
      "step": 2300
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.366923770409697e-05,
      "loss": 3.0948,
      "step": 2400
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.333533673912319e-05,
      "loss": 3.0128,
      "step": 2500
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.300143577414939e-05,
      "loss": 3.0169,
      "step": 2600
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.266753480917561e-05,
      "loss": 3.0737,
      "step": 2700
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.233363384420181e-05,
      "loss": 3.0862,
      "step": 2800
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.199973287922803e-05,
      "loss": 3.0661,
      "step": 2900
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.166583191425424e-05,
      "loss": 3.0505,
      "step": 3000
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.133193094928046e-05,
      "loss": 3.0437,
      "step": 3100
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.099802998430665e-05,
      "loss": 3.0424,
      "step": 3200
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.066412901933287e-05,
      "loss": 3.0934,
      "step": 3300
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.033356706400882e-05,
      "loss": 3.2011,
      "step": 3400
    },
    {
      "epoch": 0.11,
      "learning_rate": 8.999966609903502e-05,
      "loss": 3.1089,
      "step": 3500
    },
    {
      "epoch": 0.12,
      "learning_rate": 8.966576513406124e-05,
      "loss": 3.0698,
      "step": 3600
    },
    {
      "epoch": 0.12,
      "learning_rate": 8.933186416908745e-05,
      "loss": 3.0202,
      "step": 3700
    },
    {
      "epoch": 0.12,
      "learning_rate": 8.899796320411367e-05,
      "loss": 3.0918,
      "step": 3800
    },
    {
      "epoch": 0.13,
      "learning_rate": 8.866406223913988e-05,
      "loss": 3.1013,
      "step": 3900
    },
    {
      "epoch": 0.13,
      "learning_rate": 8.833016127416609e-05,
      "loss": 3.3244,
      "step": 4000
    },
    {
      "epoch": 0.13,
      "learning_rate": 8.799626030919231e-05,
      "loss": 3.0985,
      "step": 4100
    },
    {
      "epoch": 0.14,
      "learning_rate": 8.766235934421851e-05,
      "loss": 3.0286,
      "step": 4200
    },
    {
      "epoch": 0.14,
      "learning_rate": 8.732845837924472e-05,
      "loss": 3.0809,
      "step": 4300
    },
    {
      "epoch": 0.14,
      "learning_rate": 8.699455741427092e-05,
      "loss": 3.0599,
      "step": 4400
    },
    {
      "epoch": 0.15,
      "learning_rate": 8.666065644929714e-05,
      "loss": 3.0703,
      "step": 4500
    },
    {
      "epoch": 0.15,
      "learning_rate": 8.632675548432335e-05,
      "loss": 3.0816,
      "step": 4600
    },
    {
      "epoch": 0.15,
      "learning_rate": 8.599285451934957e-05,
      "loss": 3.0659,
      "step": 4700
    },
    {
      "epoch": 0.16,
      "learning_rate": 8.565895355437577e-05,
      "loss": 3.0197,
      "step": 4800
    },
    {
      "epoch": 0.16,
      "learning_rate": 8.532505258940199e-05,
      "loss": 3.0335,
      "step": 4900
    },
    {
      "epoch": 0.16,
      "learning_rate": 8.49911516244282e-05,
      "loss": 3.1091,
      "step": 5000
    },
    {
      "epoch": 0.17,
      "learning_rate": 8.465725065945441e-05,
      "loss": 3.0402,
      "step": 5100
    },
    {
      "epoch": 0.17,
      "learning_rate": 8.432334969448062e-05,
      "loss": 3.0689,
      "step": 5200
    },
    {
      "epoch": 0.17,
      "learning_rate": 8.398944872950684e-05,
      "loss": 3.0209,
      "step": 5300
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.365554776453304e-05,
      "loss": 3.0484,
      "step": 5400
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.332164679955925e-05,
      "loss": 3.0419,
      "step": 5500
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.29910848442352e-05,
      "loss": 3.1417,
      "step": 5600
    },
    {
      "epoch": 0.19,
      "learning_rate": 8.265718387926142e-05,
      "loss": 3.03,
      "step": 5700
    },
    {
      "epoch": 0.19,
      "learning_rate": 8.232328291428762e-05,
      "loss": 3.0755,
      "step": 5800
    },
    {
      "epoch": 0.19,
      "learning_rate": 8.198938194931384e-05,
      "loss": 3.0759,
      "step": 5900
    },
    {
      "epoch": 0.2,
      "learning_rate": 8.165548098434005e-05,
      "loss": 3.0822,
      "step": 6000
    },
    {
      "epoch": 0.2,
      "learning_rate": 8.132158001936627e-05,
      "loss": 3.3225,
      "step": 6100
    },
    {
      "epoch": 0.2,
      "learning_rate": 8.098767905439247e-05,
      "loss": 4.2645,
      "step": 6200
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.065377808941869e-05,
      "loss": 3.3215,
      "step": 6300
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.03198771244449e-05,
      "loss": 3.3495,
      "step": 6400
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.998597615947111e-05,
      "loss": 3.9503,
      "step": 6500
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.96520751944973e-05,
      "loss": 3.2799,
      "step": 6600
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.931817422952352e-05,
      "loss": 3.3251,
      "step": 6700
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.898427326454973e-05,
      "loss": 3.3824,
      "step": 6800
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.865037229957595e-05,
      "loss": 3.2103,
      "step": 6900
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.831647133460215e-05,
      "loss": 3.4477,
      "step": 7000
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.798257036962837e-05,
      "loss": 3.3625,
      "step": 7100
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.764866940465458e-05,
      "loss": 3.4537,
      "step": 7200
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.73147684396808e-05,
      "loss": 3.5719,
      "step": 7300
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.698086747470701e-05,
      "loss": 3.1677,
      "step": 7400
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.664696650973322e-05,
      "loss": 3.3993,
      "step": 7500
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.631306554475944e-05,
      "loss": 3.1136,
      "step": 7600
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.597916457978564e-05,
      "loss": 3.1078,
      "step": 7700
    },
    {
      "epoch": 0.26,
      "learning_rate": 7.564526361481185e-05,
      "loss": 3.0449,
      "step": 7800
    },
    {
      "epoch": 0.26,
      "learning_rate": 7.53147016594878e-05,
      "loss": 3.1574,
      "step": 7900
    },
    {
      "epoch": 0.26,
      "learning_rate": 7.4980800694514e-05,
      "loss": 3.2764,
      "step": 8000
    },
    {
      "epoch": 0.27,
      "learning_rate": 7.464689972954022e-05,
      "loss": 3.0553,
      "step": 8100
    },
    {
      "epoch": 0.27,
      "learning_rate": 7.431299876456643e-05,
      "loss": 3.1408,
      "step": 8200
    },
    {
      "epoch": 0.27,
      "learning_rate": 7.397909779959265e-05,
      "loss": 3.1765,
      "step": 8300
    },
    {
      "epoch": 0.28,
      "learning_rate": 7.364519683461885e-05,
      "loss": 3.0974,
      "step": 8400
    },
    {
      "epoch": 0.28,
      "learning_rate": 7.331129586964507e-05,
      "loss": 3.0998,
      "step": 8500
    },
    {
      "epoch": 0.28,
      "learning_rate": 7.297739490467128e-05,
      "loss": 3.1488,
      "step": 8600
    },
    {
      "epoch": 0.29,
      "learning_rate": 7.26434939396975e-05,
      "loss": 3.1722,
      "step": 8700
    },
    {
      "epoch": 0.29,
      "learning_rate": 7.23095929747237e-05,
      "loss": 3.1605,
      "step": 8800
    },
    {
      "epoch": 0.29,
      "learning_rate": 7.19756920097499e-05,
      "loss": 3.0762,
      "step": 8900
    },
    {
      "epoch": 0.3,
      "learning_rate": 7.164179104477612e-05,
      "loss": 3.1019,
      "step": 9000
    },
    {
      "epoch": 0.3,
      "learning_rate": 7.130789007980233e-05,
      "loss": 3.0633,
      "step": 9100
    },
    {
      "epoch": 0.3,
      "learning_rate": 7.097398911482855e-05,
      "loss": 3.1291,
      "step": 9200
    },
    {
      "epoch": 0.31,
      "learning_rate": 7.064008814985475e-05,
      "loss": 3.1701,
      "step": 9300
    },
    {
      "epoch": 0.31,
      "learning_rate": 7.030618718488097e-05,
      "loss": 3.1169,
      "step": 9400
    },
    {
      "epoch": 0.31,
      "learning_rate": 6.997228621990718e-05,
      "loss": 3.3379,
      "step": 9500
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.96383852549334e-05,
      "loss": 3.0393,
      "step": 9600
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.93044842899596e-05,
      "loss": 3.1069,
      "step": 9700
    },
    {
      "epoch": 0.32,
      "learning_rate": 6.897058332498582e-05,
      "loss": 3.1976,
      "step": 9800
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.863668236001203e-05,
      "loss": 3.0431,
      "step": 9900
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.830278139503824e-05,
      "loss": 3.063,
      "step": 10000
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.796888043006444e-05,
      "loss": 3.0741,
      "step": 10100
    },
    {
      "epoch": 0.33,
      "learning_rate": 6.763497946509065e-05,
      "loss": 3.1159,
      "step": 10200
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.730107850011686e-05,
      "loss": 3.1405,
      "step": 10300
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.697051654479281e-05,
      "loss": 3.45,
      "step": 10400
    },
    {
      "epoch": 0.34,
      "learning_rate": 6.663661557981903e-05,
      "loss": 3.0671,
      "step": 10500
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.630271461484523e-05,
      "loss": 3.1104,
      "step": 10600
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.596881364987145e-05,
      "loss": 3.262,
      "step": 10700
    },
    {
      "epoch": 0.35,
      "learning_rate": 6.563491268489766e-05,
      "loss": 3.3126,
      "step": 10800
    },
    {
      "epoch": 0.36,
      "learning_rate": 6.530101171992388e-05,
      "loss": 3.136,
      "step": 10900
    },
    {
      "epoch": 0.36,
      "learning_rate": 6.49671107549501e-05,
      "loss": 3.0665,
      "step": 11000
    },
    {
      "epoch": 0.36,
      "learning_rate": 6.46332097899763e-05,
      "loss": 3.1046,
      "step": 11100
    },
    {
      "epoch": 0.37,
      "learning_rate": 6.42993088250025e-05,
      "loss": 3.072,
      "step": 11200
    },
    {
      "epoch": 0.37,
      "learning_rate": 6.396540786002871e-05,
      "loss": 3.1027,
      "step": 11300
    },
    {
      "epoch": 0.37,
      "learning_rate": 6.363150689505493e-05,
      "loss": 3.1536,
      "step": 11400
    },
    {
      "epoch": 0.38,
      "learning_rate": 6.329760593008114e-05,
      "loss": 3.0932,
      "step": 11500
    },
    {
      "epoch": 0.38,
      "learning_rate": 6.296370496510735e-05,
      "loss": 3.0798,
      "step": 11600
    },
    {
      "epoch": 0.38,
      "learning_rate": 6.262980400013356e-05,
      "loss": 3.1229,
      "step": 11700
    },
    {
      "epoch": 0.39,
      "learning_rate": 6.229590303515978e-05,
      "loss": 3.1848,
      "step": 11800
    },
    {
      "epoch": 0.39,
      "learning_rate": 6.196200207018598e-05,
      "loss": 3.2213,
      "step": 11900
    },
    {
      "epoch": 0.39,
      "learning_rate": 6.16281011052122e-05,
      "loss": 3.3552,
      "step": 12000
    },
    {
      "epoch": 0.4,
      "learning_rate": 6.129420014023841e-05,
      "loss": 3.2678,
      "step": 12100
    },
    {
      "epoch": 0.4,
      "learning_rate": 6.0960299175264626e-05,
      "loss": 3.2389,
      "step": 12200
    },
    {
      "epoch": 0.4,
      "learning_rate": 6.062639821029084e-05,
      "loss": 3.3173,
      "step": 12300
    },
    {
      "epoch": 0.41,
      "learning_rate": 6.0292497245317036e-05,
      "loss": 3.0254,
      "step": 12400
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.995859628034325e-05,
      "loss": 3.0548,
      "step": 12500
    },
    {
      "epoch": 0.41,
      "learning_rate": 5.962469531536946e-05,
      "loss": 3.0759,
      "step": 12600
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.929079435039567e-05,
      "loss": 3.0964,
      "step": 12700
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.8956893385421884e-05,
      "loss": 3.2731,
      "step": 12800
    },
    {
      "epoch": 0.42,
      "learning_rate": 5.8622992420448096e-05,
      "loss": 3.0974,
      "step": 12900
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.828909145547431e-05,
      "loss": 3.2184,
      "step": 13000
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.795519049050052e-05,
      "loss": 3.1414,
      "step": 13100
    },
    {
      "epoch": 0.43,
      "learning_rate": 5.762128952552673e-05,
      "loss": 3.3148,
      "step": 13200
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.7287388560552944e-05,
      "loss": 3.2656,
      "step": 13300
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.6953487595579156e-05,
      "loss": 3.0808,
      "step": 13400
    },
    {
      "epoch": 0.44,
      "learning_rate": 5.661958663060537e-05,
      "loss": 3.1319,
      "step": 13500
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.628568566563157e-05,
      "loss": 3.0652,
      "step": 13600
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.5951784700657785e-05,
      "loss": 3.2544,
      "step": 13700
    },
    {
      "epoch": 0.45,
      "learning_rate": 5.5617883735684e-05,
      "loss": 3.1063,
      "step": 13800
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.528398277071021e-05,
      "loss": 3.146,
      "step": 13900
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.495008180573642e-05,
      "loss": 3.0885,
      "step": 14000
    },
    {
      "epoch": 0.46,
      "learning_rate": 5.461618084076263e-05,
      "loss": 3.0842,
      "step": 14100
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.4282279875788845e-05,
      "loss": 3.2433,
      "step": 14200
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.3948378910815057e-05,
      "loss": 3.2303,
      "step": 14300
    },
    {
      "epoch": 0.47,
      "learning_rate": 5.361447794584127e-05,
      "loss": 3.0701,
      "step": 14400
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.328391599051722e-05,
      "loss": 3.303,
      "step": 14500
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.295001502554343e-05,
      "loss": 3.1973,
      "step": 14600
    },
    {
      "epoch": 0.48,
      "learning_rate": 5.261611406056963e-05,
      "loss": 3.1066,
      "step": 14700
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.228221309559584e-05,
      "loss": 3.2033,
      "step": 14800
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.195165114027179e-05,
      "loss": 3.2673,
      "step": 14900
    },
    {
      "epoch": 0.49,
      "learning_rate": 5.161775017529801e-05,
      "loss": 3.0243,
      "step": 15000
    }
  ],
  "logging_steps": 100,
  "max_steps": 30449,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 1.105248502226903e+18,
  "trial_name": null,
  "trial_params": null
}