{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0297482837528604,
  "eval_steps": 500,
  "global_step": 3000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4e-05,
      "loss": 2.6282,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 8e-05,
      "loss": 1.9839,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00012,
      "loss": 1.5426,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00016,
      "loss": 1.4084,
      "step": 40
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002,
      "loss": 1.387,
      "step": 50
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001993220338983051,
      "loss": 1.3427,
      "step": 60
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019864406779661017,
      "loss": 1.3838,
      "step": 70
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019796610169491526,
      "loss": 1.3672,
      "step": 80
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019728813559322035,
      "loss": 1.3454,
      "step": 90
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019661016949152545,
      "loss": 1.2267,
      "step": 100
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001959322033898305,
      "loss": 1.2541,
      "step": 110
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001952542372881356,
      "loss": 1.3141,
      "step": 120
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001945762711864407,
      "loss": 1.3119,
      "step": 130
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001938983050847458,
      "loss": 1.2873,
      "step": 140
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019322033898305085,
      "loss": 1.2628,
      "step": 150
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019254237288135595,
      "loss": 1.3217,
      "step": 160
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.000191864406779661,
      "loss": 1.302,
      "step": 170
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001911864406779661,
      "loss": 1.2469,
      "step": 180
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0001905084745762712,
      "loss": 1.2582,
      "step": 190
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0001898305084745763,
      "loss": 1.2481,
      "step": 200
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00018915254237288136,
      "loss": 1.198,
      "step": 210
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00018847457627118645,
      "loss": 1.2184,
      "step": 220
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00018779661016949151,
      "loss": 1.2307,
      "step": 230
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00018711864406779663,
      "loss": 1.2281,
      "step": 240
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001864406779661017,
      "loss": 1.2761,
      "step": 250
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001857627118644068,
      "loss": 1.2587,
      "step": 260
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00018508474576271186,
      "loss": 1.2824,
      "step": 270
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00018440677966101695,
      "loss": 1.2988,
      "step": 280
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00018372881355932204,
      "loss": 1.2746,
      "step": 290
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00018305084745762714,
      "loss": 1.2088,
      "step": 300
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0001823728813559322,
      "loss": 1.1606,
      "step": 310
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0001816949152542373,
      "loss": 1.2578,
      "step": 320
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00018101694915254239,
      "loss": 1.2815,
      "step": 330
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00018033898305084748,
      "loss": 1.2771,
      "step": 340
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00017966101694915257,
      "loss": 1.2231,
      "step": 350
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00017898305084745764,
      "loss": 1.2233,
      "step": 360
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00017830508474576273,
      "loss": 1.2291,
      "step": 370
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0001776271186440678,
      "loss": 1.2341,
      "step": 380
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0001769491525423729,
      "loss": 1.1777,
      "step": 390
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00017627118644067798,
      "loss": 1.2878,
      "step": 400
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00017559322033898307,
      "loss": 1.1953,
      "step": 410
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00017491525423728814,
      "loss": 1.1945,
      "step": 420
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00017423728813559323,
      "loss": 1.1931,
      "step": 430
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001735593220338983,
      "loss": 1.1969,
      "step": 440
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00017288135593220342,
      "loss": 1.3003,
      "step": 450
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00017220338983050848,
      "loss": 1.1747,
      "step": 460
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00017152542372881357,
      "loss": 1.2155,
      "step": 470
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00017084745762711864,
      "loss": 1.2154,
      "step": 480
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00017016949152542373,
      "loss": 1.2085,
      "step": 490
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00016949152542372882,
      "loss": 1.1937,
      "step": 500
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00016881355932203392,
      "loss": 1.2409,
      "step": 510
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00016813559322033898,
      "loss": 1.2248,
      "step": 520
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00016745762711864408,
      "loss": 1.2207,
      "step": 530
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00016677966101694914,
      "loss": 1.1788,
      "step": 540
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00016610169491525423,
      "loss": 1.169,
      "step": 550
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00016542372881355933,
      "loss": 1.1821,
      "step": 560
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00016474576271186442,
      "loss": 1.2172,
      "step": 570
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00016406779661016948,
      "loss": 1.1673,
      "step": 580
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00016338983050847458,
      "loss": 1.156,
      "step": 590
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00016271186440677967,
      "loss": 1.1926,
      "step": 600
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00016203389830508476,
      "loss": 1.2045,
      "step": 610
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00016135593220338985,
      "loss": 1.2378,
      "step": 620
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00016067796610169492,
      "loss": 1.1616,
      "step": 630
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00016,
      "loss": 1.2035,
      "step": 640
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00015932203389830508,
      "loss": 1.1623,
      "step": 650
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001586440677966102,
      "loss": 1.2262,
      "step": 660
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00015796610169491526,
      "loss": 1.1931,
      "step": 670
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00015728813559322036,
      "loss": 1.1441,
      "step": 680
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00015661016949152542,
      "loss": 1.16,
      "step": 690
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00015593220338983051,
      "loss": 1.1955,
      "step": 700
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001552542372881356,
      "loss": 1.1956,
      "step": 710
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0001545762711864407,
      "loss": 1.164,
      "step": 720
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00015389830508474577,
      "loss": 1.2317,
      "step": 730
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00015322033898305086,
      "loss": 1.1596,
      "step": 740
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00015254237288135592,
      "loss": 1.1851,
      "step": 750
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00015186440677966102,
      "loss": 1.167,
      "step": 760
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001511864406779661,
      "loss": 1.1623,
      "step": 770
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0001505084745762712,
      "loss": 1.1974,
      "step": 780
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00014983050847457627,
      "loss": 1.1948,
      "step": 790
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00014915254237288136,
      "loss": 1.1842,
      "step": 800
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00014847457627118645,
      "loss": 1.1257,
      "step": 810
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00014779661016949154,
      "loss": 1.1557,
      "step": 820
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0001471186440677966,
      "loss": 1.1412,
      "step": 830
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0001464406779661017,
      "loss": 1.2058,
      "step": 840
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00014576271186440677,
      "loss": 1.249,
      "step": 850
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00014508474576271186,
      "loss": 1.1608,
      "step": 860
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00014440677966101695,
      "loss": 1.1679,
      "step": 870
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00014372881355932205,
      "loss": 1.175,
      "step": 880
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00014305084745762714,
      "loss": 1.1369,
      "step": 890
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001423728813559322,
      "loss": 1.2629,
      "step": 900
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001416949152542373,
      "loss": 1.1771,
      "step": 910
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001410169491525424,
      "loss": 1.1791,
      "step": 920
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00014033898305084748,
      "loss": 1.1478,
      "step": 930
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00013966101694915255,
      "loss": 1.1767,
      "step": 940
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00013898305084745764,
      "loss": 1.1621,
      "step": 950
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001383050847457627,
      "loss": 1.1936,
      "step": 960
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001376271186440678,
      "loss": 1.1556,
      "step": 970
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0001369491525423729,
      "loss": 1.2168,
      "step": 980
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00013627118644067798,
      "loss": 1.1899,
      "step": 990
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00013559322033898305,
      "loss": 1.0627,
      "step": 1000
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00013491525423728814,
      "loss": 1.1734,
      "step": 1010
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0001342372881355932,
      "loss": 1.1362,
      "step": 1020
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00013355932203389833,
      "loss": 1.1481,
      "step": 1030
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001328813559322034,
      "loss": 1.1567,
      "step": 1040
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00013220338983050849,
      "loss": 1.1406,
      "step": 1050
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00013152542372881355,
      "loss": 1.1228,
      "step": 1060
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00013084745762711864,
      "loss": 1.1357,
      "step": 1070
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00013016949152542374,
      "loss": 1.1677,
      "step": 1080
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00012949152542372883,
      "loss": 1.1726,
      "step": 1090
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001288135593220339,
      "loss": 1.1771,
      "step": 1100
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.000128135593220339,
      "loss": 1.1527,
      "step": 1110
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00012745762711864405,
      "loss": 1.1748,
      "step": 1120
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00012677966101694917,
      "loss": 1.2424,
      "step": 1130
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00012610169491525426,
      "loss": 1.1711,
      "step": 1140
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00012542372881355933,
      "loss": 1.1538,
      "step": 1150
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00012474576271186442,
      "loss": 1.0941,
      "step": 1160
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0001240677966101695,
      "loss": 1.2163,
      "step": 1170
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012338983050847458,
      "loss": 1.1581,
      "step": 1180
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012271186440677967,
      "loss": 1.1345,
      "step": 1190
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00012203389830508477,
      "loss": 1.1448,
      "step": 1200
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012135593220338983,
      "loss": 1.1968,
      "step": 1210
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012067796610169492,
      "loss": 1.1306,
      "step": 1220
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00012,
      "loss": 1.1125,
      "step": 1230
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0001193220338983051,
      "loss": 1.1354,
      "step": 1240
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00011864406779661017,
      "loss": 1.1113,
      "step": 1250
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00011796610169491527,
      "loss": 1.1472,
      "step": 1260
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00011728813559322033,
      "loss": 1.1011,
      "step": 1270
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00011661016949152544,
      "loss": 1.1174,
      "step": 1280
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001159322033898305,
      "loss": 1.1574,
      "step": 1290
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001152542372881356,
      "loss": 1.1016,
      "step": 1300
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011457627118644068,
      "loss": 1.1353,
      "step": 1310
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00011389830508474577,
      "loss": 1.0817,
      "step": 1320
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011322033898305085,
      "loss": 1.1467,
      "step": 1330
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011254237288135594,
      "loss": 1.1301,
      "step": 1340
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00011186440677966102,
      "loss": 1.1716,
      "step": 1350
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00011118644067796611,
      "loss": 1.1534,
      "step": 1360
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00011050847457627118,
      "loss": 1.1629,
      "step": 1370
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00010983050847457627,
      "loss": 1.1446,
      "step": 1380
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010915254237288135,
      "loss": 1.1448,
      "step": 1390
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010847457627118644,
      "loss": 1.1453,
      "step": 1400
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00010779661016949153,
      "loss": 1.1202,
      "step": 1410
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010711864406779661,
      "loss": 1.1288,
      "step": 1420
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001064406779661017,
      "loss": 1.1126,
      "step": 1430
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00010576271186440679,
      "loss": 1.1432,
      "step": 1440
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010508474576271188,
      "loss": 1.133,
      "step": 1450
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010440677966101696,
      "loss": 1.1793,
      "step": 1460
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00010372881355932205,
      "loss": 1.1938,
      "step": 1470
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00010305084745762712,
      "loss": 1.1444,
      "step": 1480
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00010237288135593222,
      "loss": 1.1301,
      "step": 1490
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00010169491525423729,
      "loss": 1.1332,
      "step": 1500
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00010101694915254238,
      "loss": 1.1368,
      "step": 1510
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00010033898305084746,
      "loss": 1.0965,
      "step": 1520
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.966101694915255e-05,
      "loss": 1.168,
      "step": 1530
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.898305084745763e-05,
      "loss": 1.097,
      "step": 1540
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.830508474576272e-05,
      "loss": 1.1166,
      "step": 1550
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.76271186440678e-05,
      "loss": 1.1539,
      "step": 1560
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.69491525423729e-05,
      "loss": 1.1388,
      "step": 1570
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.627118644067797e-05,
      "loss": 1.1898,
      "step": 1580
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.559322033898305e-05,
      "loss": 1.155,
      "step": 1590
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.491525423728815e-05,
      "loss": 1.1212,
      "step": 1600
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.423728813559322e-05,
      "loss": 1.0912,
      "step": 1610
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.355932203389832e-05,
      "loss": 1.1017,
      "step": 1620
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.28813559322034e-05,
      "loss": 1.1294,
      "step": 1630
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.220338983050847e-05,
      "loss": 1.1105,
      "step": 1640
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.152542372881357e-05,
      "loss": 1.1159,
      "step": 1650
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.084745762711865e-05,
      "loss": 1.1195,
      "step": 1660
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.016949152542374e-05,
      "loss": 1.1514,
      "step": 1670
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.949152542372882e-05,
      "loss": 1.1156,
      "step": 1680
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.88135593220339e-05,
      "loss": 1.1175,
      "step": 1690
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.813559322033899e-05,
      "loss": 1.1928,
      "step": 1700
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.745762711864407e-05,
      "loss": 1.0526,
      "step": 1710
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.677966101694915e-05,
      "loss": 1.0861,
      "step": 1720
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.610169491525424e-05,
      "loss": 1.185,
      "step": 1730
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.542372881355932e-05,
      "loss": 1.132,
      "step": 1740
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.474576271186441e-05,
      "loss": 1.1164,
      "step": 1750
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.406779661016949e-05,
      "loss": 1.1232,
      "step": 1760
    },
    {
      "epoch": 0.61,
      "learning_rate": 8.338983050847457e-05,
      "loss": 1.0915,
      "step": 1770
    },
    {
      "epoch": 0.61,
      "learning_rate": 8.271186440677966e-05,
      "loss": 1.1382,
      "step": 1780
    },
    {
      "epoch": 0.61,
      "learning_rate": 8.203389830508474e-05,
      "loss": 1.0837,
      "step": 1790
    },
    {
      "epoch": 0.62,
      "learning_rate": 8.135593220338983e-05,
      "loss": 1.1287,
      "step": 1800
    },
    {
      "epoch": 0.62,
      "learning_rate": 8.067796610169493e-05,
      "loss": 1.1251,
      "step": 1810
    },
    {
      "epoch": 0.62,
      "learning_rate": 8e-05,
      "loss": 1.1527,
      "step": 1820
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.93220338983051e-05,
      "loss": 1.1309,
      "step": 1830
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.864406779661018e-05,
      "loss": 1.1185,
      "step": 1840
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.796610169491526e-05,
      "loss": 1.1131,
      "step": 1850
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.728813559322035e-05,
      "loss": 1.1497,
      "step": 1860
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.661016949152543e-05,
      "loss": 1.2128,
      "step": 1870
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.593220338983051e-05,
      "loss": 1.0872,
      "step": 1880
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.52542372881356e-05,
      "loss": 1.113,
      "step": 1890
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.457627118644068e-05,
      "loss": 1.0717,
      "step": 1900
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.389830508474577e-05,
      "loss": 1.0703,
      "step": 1910
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.322033898305085e-05,
      "loss": 1.1531,
      "step": 1920
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.254237288135593e-05,
      "loss": 1.1379,
      "step": 1930
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.186440677966102e-05,
      "loss": 1.1474,
      "step": 1940
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.11864406779661e-05,
      "loss": 1.0944,
      "step": 1950
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.05084745762712e-05,
      "loss": 1.0956,
      "step": 1960
    },
    {
      "epoch": 0.68,
      "learning_rate": 6.983050847457627e-05,
      "loss": 1.0674,
      "step": 1970
    },
    {
      "epoch": 0.68,
      "learning_rate": 6.915254237288135e-05,
      "loss": 1.0851,
      "step": 1980
    },
    {
      "epoch": 0.68,
      "learning_rate": 6.847457627118645e-05,
      "loss": 1.1439,
      "step": 1990
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.779661016949152e-05,
      "loss": 1.0869,
      "step": 2000
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.71186440677966e-05,
      "loss": 1.1284,
      "step": 2010
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.64406779661017e-05,
      "loss": 1.0684,
      "step": 2020
    },
    {
      "epoch": 0.7,
      "learning_rate": 6.576271186440678e-05,
      "loss": 1.1464,
      "step": 2030
    },
    {
      "epoch": 0.7,
      "learning_rate": 6.508474576271187e-05,
      "loss": 1.0331,
      "step": 2040
    },
    {
      "epoch": 0.7,
      "learning_rate": 6.440677966101695e-05,
      "loss": 1.0975,
      "step": 2050
    },
    {
      "epoch": 0.71,
      "learning_rate": 6.379661016949154e-05,
      "loss": 1.0438,
      "step": 2060
    },
    {
      "epoch": 0.71,
      "learning_rate": 6.311864406779661e-05,
      "loss": 1.146,
      "step": 2070
    },
    {
      "epoch": 0.71,
      "learning_rate": 6.244067796610171e-05,
      "loss": 1.135,
      "step": 2080
    },
    {
      "epoch": 0.72,
      "learning_rate": 6.176271186440679e-05,
      "loss": 1.0624,
      "step": 2090
    },
    {
      "epoch": 0.72,
      "learning_rate": 6.108474576271187e-05,
      "loss": 1.0677,
      "step": 2100
    },
    {
      "epoch": 0.72,
      "learning_rate": 6.040677966101695e-05,
      "loss": 1.0814,
      "step": 2110
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.972881355932204e-05,
      "loss": 1.1558,
      "step": 2120
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.905084745762712e-05,
      "loss": 1.0757,
      "step": 2130
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.837288135593221e-05,
      "loss": 1.0827,
      "step": 2140
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.7694915254237295e-05,
      "loss": 1.1251,
      "step": 2150
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.7016949152542374e-05,
      "loss": 1.0687,
      "step": 2160
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.633898305084746e-05,
      "loss": 1.0848,
      "step": 2170
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.5661016949152545e-05,
      "loss": 1.1592,
      "step": 2180
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.498305084745763e-05,
      "loss": 1.0944,
      "step": 2190
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.430508474576271e-05,
      "loss": 1.101,
      "step": 2200
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.3627118644067796e-05,
      "loss": 1.0942,
      "step": 2210
    },
    {
      "epoch": 0.76,
      "learning_rate": 5.294915254237288e-05,
      "loss": 1.0955,
      "step": 2220
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.227118644067797e-05,
      "loss": 1.0917,
      "step": 2230
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.1593220338983054e-05,
      "loss": 1.1238,
      "step": 2240
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.091525423728813e-05,
      "loss": 1.0737,
      "step": 2250
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.023728813559322e-05,
      "loss": 1.0966,
      "step": 2260
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.955932203389831e-05,
      "loss": 1.0809,
      "step": 2270
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.888135593220339e-05,
      "loss": 1.0894,
      "step": 2280
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.8203389830508476e-05,
      "loss": 1.1213,
      "step": 2290
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.752542372881356e-05,
      "loss": 1.0784,
      "step": 2300
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.684745762711865e-05,
      "loss": 1.123,
      "step": 2310
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.6169491525423734e-05,
      "loss": 1.1118,
      "step": 2320
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.549152542372881e-05,
      "loss": 1.1142,
      "step": 2330
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.48135593220339e-05,
      "loss": 1.1157,
      "step": 2340
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.4135593220338984e-05,
      "loss": 1.1249,
      "step": 2350
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.345762711864407e-05,
      "loss": 1.0925,
      "step": 2360
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.277966101694915e-05,
      "loss": 1.1079,
      "step": 2370
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.210169491525424e-05,
      "loss": 1.119,
      "step": 2380
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.142372881355933e-05,
      "loss": 1.0334,
      "step": 2390
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.0745762711864414e-05,
      "loss": 1.0725,
      "step": 2400
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.006779661016949e-05,
      "loss": 1.0998,
      "step": 2410
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.938983050847458e-05,
      "loss": 1.107,
      "step": 2420
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.8711864406779664e-05,
      "loss": 1.1407,
      "step": 2430
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.803389830508475e-05,
      "loss": 1.13,
      "step": 2440
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.735593220338983e-05,
      "loss": 1.1231,
      "step": 2450
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.6677966101694915e-05,
      "loss": 1.0928,
      "step": 2460
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.6e-05,
      "loss": 1.0743,
      "step": 2470
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.532203389830509e-05,
      "loss": 1.1144,
      "step": 2480
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.4644067796610166e-05,
      "loss": 1.0636,
      "step": 2490
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.396610169491525e-05,
      "loss": 1.1148,
      "step": 2500
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.3288135593220344e-05,
      "loss": 1.0876,
      "step": 2510
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.261016949152543e-05,
      "loss": 1.1073,
      "step": 2520
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.193220338983051e-05,
      "loss": 1.0801,
      "step": 2530
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.1254237288135595e-05,
      "loss": 1.0995,
      "step": 2540
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.057627118644068e-05,
      "loss": 1.1142,
      "step": 2550
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.9898305084745763e-05,
      "loss": 1.0569,
      "step": 2560
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.922033898305085e-05,
      "loss": 1.0937,
      "step": 2570
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.854237288135593e-05,
      "loss": 1.0893,
      "step": 2580
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.7864406779661017e-05,
      "loss": 1.0466,
      "step": 2590
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.7186440677966103e-05,
      "loss": 1.0828,
      "step": 2600
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.6508474576271186e-05,
      "loss": 1.0123,
      "step": 2610
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.583050847457627e-05,
      "loss": 1.0969,
      "step": 2620
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.5152542372881354e-05,
      "loss": 1.049,
      "step": 2630
    },
    {
      "epoch": 0.91,
      "learning_rate": 2.4474576271186443e-05,
      "loss": 1.0377,
      "step": 2640
    },
    {
      "epoch": 0.91,
      "learning_rate": 2.3796610169491526e-05,
      "loss": 1.103,
      "step": 2650
    },
    {
      "epoch": 0.91,
      "learning_rate": 2.311864406779661e-05,
      "loss": 1.0602,
      "step": 2660
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.2440677966101694e-05,
      "loss": 1.112,
      "step": 2670
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.1762711864406783e-05,
      "loss": 1.0804,
      "step": 2680
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.1084745762711866e-05,
      "loss": 1.1136,
      "step": 2690
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.040677966101695e-05,
      "loss": 1.064,
      "step": 2700
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.9728813559322034e-05,
      "loss": 1.1352,
      "step": 2710
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.905084745762712e-05,
      "loss": 1.0464,
      "step": 2720
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.8372881355932202e-05,
      "loss": 1.0936,
      "step": 2730
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.769491525423729e-05,
      "loss": 1.0586,
      "step": 2740
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.7016949152542374e-05,
      "loss": 1.076,
      "step": 2750
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.633898305084746e-05,
      "loss": 1.094,
      "step": 2760
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.5661016949152542e-05,
      "loss": 1.1068,
      "step": 2770
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.4983050847457628e-05,
      "loss": 1.0789,
      "step": 2780
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.4305084745762712e-05,
      "loss": 1.1326,
      "step": 2790
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.3627118644067796e-05,
      "loss": 1.0818,
      "step": 2800
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.2949152542372884e-05,
      "loss": 1.119,
      "step": 2810
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.2271186440677966e-05,
      "loss": 1.1013,
      "step": 2820
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.1593220338983052e-05,
      "loss": 1.0815,
      "step": 2830
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.0915254237288136e-05,
      "loss": 1.0326,
      "step": 2840
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.023728813559322e-05,
      "loss": 1.1054,
      "step": 2850
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.559322033898306e-06,
      "loss": 1.0615,
      "step": 2860
    },
    {
      "epoch": 0.99,
      "learning_rate": 8.88135593220339e-06,
      "loss": 1.1075,
      "step": 2870
    },
    {
      "epoch": 0.99,
      "learning_rate": 8.203389830508475e-06,
      "loss": 1.1043,
      "step": 2880
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.525423728813559e-06,
      "loss": 1.1056,
      "step": 2890
    },
    {
      "epoch": 1.0,
      "learning_rate": 6.8474576271186445e-06,
      "loss": 1.0234,
      "step": 2900
    },
    {
      "epoch": 1.0,
      "learning_rate": 6.169491525423729e-06,
      "loss": 1.0732,
      "step": 2910
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.491525423728814e-06,
      "loss": 1.0076,
      "step": 2920
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813559322033899e-06,
      "loss": 0.9715,
      "step": 2930
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.135593220338984e-06,
      "loss": 0.9092,
      "step": 2940
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.4576271186440682e-06,
      "loss": 0.8787,
      "step": 2950
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7796610169491524e-06,
      "loss": 0.841,
      "step": 2960
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.1016949152542374e-06,
      "loss": 0.9436,
      "step": 2970
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.423728813559322e-06,
      "loss": 0.9332,
      "step": 2980
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.457627118644068e-07,
      "loss": 0.8813,
      "step": 2990
    },
    {
      "epoch": 1.03,
      "learning_rate": 6.779661016949153e-08,
      "loss": 0.8643,
      "step": 3000
    },
    {
      "epoch": 1.03,
      "step": 3000,
      "total_flos": 3.2240003230162944e+16,
      "train_loss": 1.1516946287155152,
      "train_runtime": 29468.0473,
      "train_samples_per_second": 0.611,
      "train_steps_per_second": 0.102
    }
  ],
  "logging_steps": 10,
  "max_steps": 3000,
  "num_train_epochs": 2,
  "save_steps": 500,
  "total_flos": 3.2240003230162944e+16,
  "trial_name": null,
  "trial_params": null
}