{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9997040544539805,
  "global_step": 1689,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 5.000000000000001e-07,
      "loss": 2.7237,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 5e-06,
      "loss": 2.7145,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 1e-05,
      "loss": 2.7579,
      "step": 20
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.5e-05,
      "loss": 2.747,
      "step": 30
    },
    {
      "epoch": 0.02,
      "learning_rate": 2e-05,
      "loss": 2.6339,
      "step": 40
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.5e-05,
      "loss": 2.6645,
      "step": 50
    },
    {
      "epoch": 0.04,
      "learning_rate": 3e-05,
      "loss": 2.6576,
      "step": 60
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.5e-05,
      "loss": 2.6678,
      "step": 70
    },
    {
      "epoch": 0.05,
      "learning_rate": 4e-05,
      "loss": 2.5723,
      "step": 80
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.5e-05,
      "loss": 2.514,
      "step": 90
    },
    {
      "epoch": 0.06,
      "learning_rate": 5e-05,
      "loss": 2.575,
      "step": 100
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.968533668974198e-05,
      "loss": 2.5101,
      "step": 110
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.937067337948396e-05,
      "loss": 2.5532,
      "step": 120
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9056010069225936e-05,
      "loss": 2.5329,
      "step": 130
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.874134675896791e-05,
      "loss": 2.4838,
      "step": 140
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.842668344870988e-05,
      "loss": 2.5586,
      "step": 150
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8112020138451856e-05,
      "loss": 2.5078,
      "step": 160
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.7797356828193834e-05,
      "loss": 2.5224,
      "step": 170
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.748269351793581e-05,
      "loss": 2.5317,
      "step": 180
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.716803020767779e-05,
      "loss": 2.5802,
      "step": 190
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.685336689741977e-05,
      "loss": 2.4744,
      "step": 200
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.653870358716174e-05,
      "loss": 2.549,
      "step": 210
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.622404027690371e-05,
      "loss": 2.5513,
      "step": 220
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.590937696664569e-05,
      "loss": 2.4999,
      "step": 230
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.5594713656387665e-05,
      "loss": 2.5966,
      "step": 240
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.528005034612964e-05,
      "loss": 2.5307,
      "step": 250
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.496538703587162e-05,
      "loss": 2.5179,
      "step": 260
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.46507237256136e-05,
      "loss": 2.5107,
      "step": 270
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.433606041535558e-05,
      "loss": 2.473,
      "step": 280
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.402139710509755e-05,
      "loss": 2.5107,
      "step": 290
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.370673379483952e-05,
      "loss": 2.5232,
      "step": 300
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.33920704845815e-05,
      "loss": 2.5239,
      "step": 310
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.3077407174323475e-05,
      "loss": 2.559,
      "step": 320
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.276274386406545e-05,
      "loss": 2.5414,
      "step": 330
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.244808055380743e-05,
      "loss": 2.5183,
      "step": 340
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.213341724354941e-05,
      "loss": 2.5468,
      "step": 350
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.181875393329138e-05,
      "loss": 2.5972,
      "step": 360
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.150409062303336e-05,
      "loss": 2.5869,
      "step": 370
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.118942731277533e-05,
      "loss": 2.5745,
      "step": 380
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.0874764002517306e-05,
      "loss": 2.6045,
      "step": 390
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.0560100692259284e-05,
      "loss": 2.5491,
      "step": 400
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.024543738200126e-05,
      "loss": 2.5674,
      "step": 410
    },
    {
      "epoch": 0.25,
      "learning_rate": 3.993077407174324e-05,
      "loss": 2.5299,
      "step": 420
    },
    {
      "epoch": 0.25,
      "learning_rate": 3.961611076148522e-05,
      "loss": 2.6367,
      "step": 430
    },
    {
      "epoch": 0.26,
      "learning_rate": 3.930144745122719e-05,
      "loss": 2.5684,
      "step": 440
    },
    {
      "epoch": 0.27,
      "learning_rate": 3.898678414096916e-05,
      "loss": 2.5414,
      "step": 450
    },
    {
      "epoch": 0.27,
      "learning_rate": 3.867212083071114e-05,
      "loss": 2.5335,
      "step": 460
    },
    {
      "epoch": 0.28,
      "learning_rate": 3.8357457520453115e-05,
      "loss": 2.5569,
      "step": 470
    },
    {
      "epoch": 0.28,
      "learning_rate": 3.804279421019509e-05,
      "loss": 2.602,
      "step": 480
    },
    {
      "epoch": 0.29,
      "learning_rate": 3.772813089993707e-05,
      "loss": 2.5796,
      "step": 490
    },
    {
      "epoch": 0.3,
      "learning_rate": 3.741346758967905e-05,
      "loss": 2.6123,
      "step": 500
    },
    {
      "epoch": 0.3,
      "learning_rate": 3.7098804279421027e-05,
      "loss": 2.5303,
      "step": 510
    },
    {
      "epoch": 0.31,
      "learning_rate": 3.6784140969163e-05,
      "loss": 2.6367,
      "step": 520
    },
    {
      "epoch": 0.31,
      "learning_rate": 3.646947765890497e-05,
      "loss": 2.5969,
      "step": 530
    },
    {
      "epoch": 0.32,
      "learning_rate": 3.615481434864695e-05,
      "loss": 2.6128,
      "step": 540
    },
    {
      "epoch": 0.33,
      "learning_rate": 3.5840151038388924e-05,
      "loss": 2.6029,
      "step": 550
    },
    {
      "epoch": 0.33,
      "learning_rate": 3.55254877281309e-05,
      "loss": 2.5175,
      "step": 560
    },
    {
      "epoch": 0.34,
      "learning_rate": 3.521082441787288e-05,
      "loss": 2.5812,
      "step": 570
    },
    {
      "epoch": 0.34,
      "learning_rate": 3.489616110761486e-05,
      "loss": 2.6197,
      "step": 580
    },
    {
      "epoch": 0.35,
      "learning_rate": 3.458149779735683e-05,
      "loss": 2.6282,
      "step": 590
    },
    {
      "epoch": 0.36,
      "learning_rate": 3.42668344870988e-05,
      "loss": 2.6018,
      "step": 600
    },
    {
      "epoch": 0.36,
      "learning_rate": 3.395217117684078e-05,
      "loss": 2.5908,
      "step": 610
    },
    {
      "epoch": 0.37,
      "learning_rate": 3.3637507866582756e-05,
      "loss": 2.6068,
      "step": 620
    },
    {
      "epoch": 0.37,
      "learning_rate": 3.3322844556324734e-05,
      "loss": 2.5934,
      "step": 630
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.300818124606671e-05,
      "loss": 2.6117,
      "step": 640
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.269351793580869e-05,
      "loss": 2.5347,
      "step": 650
    },
    {
      "epoch": 0.39,
      "learning_rate": 3.237885462555067e-05,
      "loss": 2.6547,
      "step": 660
    },
    {
      "epoch": 0.4,
      "learning_rate": 3.206419131529264e-05,
      "loss": 2.6311,
      "step": 670
    },
    {
      "epoch": 0.4,
      "learning_rate": 3.174952800503461e-05,
      "loss": 2.6168,
      "step": 680
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.143486469477659e-05,
      "loss": 2.5944,
      "step": 690
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.1120201384518565e-05,
      "loss": 2.5865,
      "step": 700
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.080553807426054e-05,
      "loss": 2.6284,
      "step": 710
    },
    {
      "epoch": 0.43,
      "learning_rate": 3.049087476400252e-05,
      "loss": 2.5945,
      "step": 720
    },
    {
      "epoch": 0.43,
      "learning_rate": 3.0176211453744495e-05,
      "loss": 2.5627,
      "step": 730
    },
    {
      "epoch": 0.44,
      "learning_rate": 2.9861548143486473e-05,
      "loss": 2.5967,
      "step": 740
    },
    {
      "epoch": 0.44,
      "learning_rate": 2.9546884833228444e-05,
      "loss": 2.6523,
      "step": 750
    },
    {
      "epoch": 0.45,
      "learning_rate": 2.9232221522970422e-05,
      "loss": 2.6584,
      "step": 760
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.8917558212712397e-05,
      "loss": 2.6243,
      "step": 770
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.8602894902454374e-05,
      "loss": 2.5734,
      "step": 780
    },
    {
      "epoch": 0.47,
      "learning_rate": 2.8288231592196352e-05,
      "loss": 2.5459,
      "step": 790
    },
    {
      "epoch": 0.47,
      "learning_rate": 2.7973568281938327e-05,
      "loss": 2.5985,
      "step": 800
    },
    {
      "epoch": 0.48,
      "learning_rate": 2.7658904971680305e-05,
      "loss": 2.6552,
      "step": 810
    },
    {
      "epoch": 0.49,
      "learning_rate": 2.7344241661422282e-05,
      "loss": 2.5976,
      "step": 820
    },
    {
      "epoch": 0.49,
      "learning_rate": 2.7029578351164254e-05,
      "loss": 2.5838,
      "step": 830
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.671491504090623e-05,
      "loss": 2.6057,
      "step": 840
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.6400251730648206e-05,
      "loss": 2.6784,
      "step": 850
    },
    {
      "epoch": 0.51,
      "learning_rate": 2.6085588420390184e-05,
      "loss": 2.6413,
      "step": 860
    },
    {
      "epoch": 0.51,
      "learning_rate": 2.577092511013216e-05,
      "loss": 2.5734,
      "step": 870
    },
    {
      "epoch": 0.52,
      "learning_rate": 2.5456261799874136e-05,
      "loss": 2.6205,
      "step": 880
    },
    {
      "epoch": 0.53,
      "learning_rate": 2.5141598489616114e-05,
      "loss": 2.6198,
      "step": 890
    },
    {
      "epoch": 0.53,
      "learning_rate": 2.482693517935809e-05,
      "loss": 2.5844,
      "step": 900
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.4512271869100066e-05,
      "loss": 2.6097,
      "step": 910
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.419760855884204e-05,
      "loss": 2.6332,
      "step": 920
    },
    {
      "epoch": 0.55,
      "learning_rate": 2.3882945248584015e-05,
      "loss": 2.6285,
      "step": 930
    },
    {
      "epoch": 0.56,
      "learning_rate": 2.3568281938325993e-05,
      "loss": 2.668,
      "step": 940
    },
    {
      "epoch": 0.56,
      "learning_rate": 2.3253618628067967e-05,
      "loss": 2.6187,
      "step": 950
    },
    {
      "epoch": 0.57,
      "learning_rate": 2.2938955317809942e-05,
      "loss": 2.6292,
      "step": 960
    },
    {
      "epoch": 0.57,
      "learning_rate": 2.262429200755192e-05,
      "loss": 2.6736,
      "step": 970
    },
    {
      "epoch": 0.58,
      "learning_rate": 2.2309628697293898e-05,
      "loss": 2.6426,
      "step": 980
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.1994965387035872e-05,
      "loss": 2.6719,
      "step": 990
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.1680302076777847e-05,
      "loss": 2.6793,
      "step": 1000
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.1365638766519824e-05,
      "loss": 2.7038,
      "step": 1010
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.1050975456261802e-05,
      "loss": 2.6515,
      "step": 1020
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.0736312146003777e-05,
      "loss": 2.6198,
      "step": 1030
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.042164883574575e-05,
      "loss": 2.6505,
      "step": 1040
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.010698552548773e-05,
      "loss": 2.6638,
      "step": 1050
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.9792322215229707e-05,
      "loss": 2.6602,
      "step": 1060
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.947765890497168e-05,
      "loss": 2.599,
      "step": 1070
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.9162995594713656e-05,
      "loss": 2.6421,
      "step": 1080
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8848332284455634e-05,
      "loss": 2.6311,
      "step": 1090
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.853366897419761e-05,
      "loss": 2.617,
      "step": 1100
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8219005663939586e-05,
      "loss": 2.7233,
      "step": 1110
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.790434235368156e-05,
      "loss": 2.6468,
      "step": 1120
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.7589679043423538e-05,
      "loss": 2.6206,
      "step": 1130
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.7275015733165513e-05,
      "loss": 2.6542,
      "step": 1140
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.696035242290749e-05,
      "loss": 2.6512,
      "step": 1150
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.6645689112649465e-05,
      "loss": 2.6035,
      "step": 1160
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.6331025802391443e-05,
      "loss": 2.6696,
      "step": 1170
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.6016362492133417e-05,
      "loss": 2.6677,
      "step": 1180
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.5701699181875392e-05,
      "loss": 2.6828,
      "step": 1190
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.538703587161737e-05,
      "loss": 2.6724,
      "step": 1200
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.5072372561359346e-05,
      "loss": 2.6375,
      "step": 1210
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.4757709251101324e-05,
      "loss": 2.6768,
      "step": 1220
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4443045940843297e-05,
      "loss": 2.6233,
      "step": 1230
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4128382630585274e-05,
      "loss": 2.6259,
      "step": 1240
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.381371932032725e-05,
      "loss": 2.6801,
      "step": 1250
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.3499056010069228e-05,
      "loss": 2.7294,
      "step": 1260
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.3184392699811201e-05,
      "loss": 2.7005,
      "step": 1270
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.2869729389553179e-05,
      "loss": 2.7266,
      "step": 1280
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.2555066079295155e-05,
      "loss": 2.6862,
      "step": 1290
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.2240402769037131e-05,
      "loss": 2.6623,
      "step": 1300
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.1925739458779107e-05,
      "loss": 2.731,
      "step": 1310
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.1611076148521084e-05,
      "loss": 2.6724,
      "step": 1320
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.129641283826306e-05,
      "loss": 2.64,
      "step": 1330
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.0981749528005034e-05,
      "loss": 2.7042,
      "step": 1340
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.0667086217747012e-05,
      "loss": 2.6966,
      "step": 1350
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.0352422907488987e-05,
      "loss": 2.7149,
      "step": 1360
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.0037759597230964e-05,
      "loss": 2.7139,
      "step": 1370
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.723096286972939e-06,
      "loss": 2.6948,
      "step": 1380
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.408432976714915e-06,
      "loss": 2.7423,
      "step": 1390
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.093769666456891e-06,
      "loss": 2.6981,
      "step": 1400
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.779106356198867e-06,
      "loss": 2.7368,
      "step": 1410
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.464443045940844e-06,
      "loss": 2.6836,
      "step": 1420
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.14977973568282e-06,
      "loss": 2.7165,
      "step": 1430
    },
    {
      "epoch": 0.85,
      "learning_rate": 7.835116425424796e-06,
      "loss": 2.7344,
      "step": 1440
    },
    {
      "epoch": 0.86,
      "learning_rate": 7.520453115166771e-06,
      "loss": 2.7199,
      "step": 1450
    },
    {
      "epoch": 0.86,
      "learning_rate": 7.205789804908748e-06,
      "loss": 2.6674,
      "step": 1460
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.8911264946507235e-06,
      "loss": 2.755,
      "step": 1470
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.5764631843927005e-06,
      "loss": 2.6625,
      "step": 1480
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.261799874134676e-06,
      "loss": 2.712,
      "step": 1490
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.947136563876652e-06,
      "loss": 2.7506,
      "step": 1500
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.632473253618628e-06,
      "loss": 2.7283,
      "step": 1510
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.317809943360604e-06,
      "loss": 2.7416,
      "step": 1520
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.0031466331025804e-06,
      "loss": 2.7262,
      "step": 1530
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.688483322844557e-06,
      "loss": 2.7172,
      "step": 1540
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.373820012586533e-06,
      "loss": 2.7281,
      "step": 1550
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.059156702328509e-06,
      "loss": 2.7196,
      "step": 1560
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.7444933920704846e-06,
      "loss": 2.7212,
      "step": 1570
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.429830081812461e-06,
      "loss": 2.6828,
      "step": 1580
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.115166771554437e-06,
      "loss": 2.6544,
      "step": 1590
    },
    {
      "epoch": 0.95,
      "learning_rate": 2.800503461296413e-06,
      "loss": 2.7682,
      "step": 1600
    },
    {
      "epoch": 0.95,
      "learning_rate": 2.4858401510383893e-06,
      "loss": 2.7133,
      "step": 1610
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.1711768407803654e-06,
      "loss": 2.7196,
      "step": 1620
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.8565135305223414e-06,
      "loss": 2.7418,
      "step": 1630
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.5418502202643173e-06,
      "loss": 2.7931,
      "step": 1640
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.2271869100062933e-06,
      "loss": 2.7782,
      "step": 1650
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.125235997482694e-07,
      "loss": 2.7426,
      "step": 1660
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.978602894902455e-07,
      "loss": 2.7689,
      "step": 1670
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.8319697923222154e-07,
      "loss": 2.7109,
      "step": 1680
    },
    {
      "epoch": 1.0,
      "step": 1689,
      "total_flos": 5.644869713893786e+16,
      "train_loss": 2.6318450465295498,
      "train_runtime": 20073.5525,
      "train_samples_per_second": 0.337,
      "train_steps_per_second": 0.084
    }
  ],
  "max_steps": 1689,
  "num_train_epochs": 1,
  "total_flos": 5.644869713893786e+16,
  "trial_name": null,
  "trial_params": null
}