aws-reinvent-2023-cmp302 / trainer_state.json
Ubuntu
just the model from aws reinvent 2023 CMP302
0d6d299
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 2368,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.9915540540540542e-05,
"loss": 4.6179,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 1.9831081081081082e-05,
"loss": 4.6077,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 1.9746621621621623e-05,
"loss": 4.5968,
"step": 30
},
{
"epoch": 0.02,
"learning_rate": 1.9662162162162163e-05,
"loss": 4.5821,
"step": 40
},
{
"epoch": 0.02,
"learning_rate": 1.9577702702702703e-05,
"loss": 4.5755,
"step": 50
},
{
"epoch": 0.03,
"learning_rate": 1.9493243243243244e-05,
"loss": 4.5376,
"step": 60
},
{
"epoch": 0.03,
"learning_rate": 1.9408783783783784e-05,
"loss": 4.5339,
"step": 70
},
{
"epoch": 0.03,
"learning_rate": 1.9324324324324328e-05,
"loss": 4.5202,
"step": 80
},
{
"epoch": 0.04,
"learning_rate": 1.9239864864864865e-05,
"loss": 4.5162,
"step": 90
},
{
"epoch": 0.04,
"learning_rate": 1.9155405405405408e-05,
"loss": 4.4989,
"step": 100
},
{
"epoch": 0.05,
"learning_rate": 1.9070945945945945e-05,
"loss": 4.4788,
"step": 110
},
{
"epoch": 0.05,
"learning_rate": 1.898648648648649e-05,
"loss": 4.4514,
"step": 120
},
{
"epoch": 0.05,
"learning_rate": 1.890202702702703e-05,
"loss": 4.4425,
"step": 130
},
{
"epoch": 0.06,
"learning_rate": 1.881756756756757e-05,
"loss": 4.4391,
"step": 140
},
{
"epoch": 0.06,
"learning_rate": 1.873310810810811e-05,
"loss": 4.411,
"step": 150
},
{
"epoch": 0.07,
"learning_rate": 1.864864864864865e-05,
"loss": 4.3889,
"step": 160
},
{
"epoch": 0.07,
"learning_rate": 1.856418918918919e-05,
"loss": 4.3677,
"step": 170
},
{
"epoch": 0.08,
"learning_rate": 1.847972972972973e-05,
"loss": 4.3688,
"step": 180
},
{
"epoch": 0.08,
"learning_rate": 1.839527027027027e-05,
"loss": 4.3612,
"step": 190
},
{
"epoch": 0.08,
"learning_rate": 1.831081081081081e-05,
"loss": 4.3355,
"step": 200
},
{
"epoch": 0.09,
"learning_rate": 1.8226351351351352e-05,
"loss": 4.3058,
"step": 210
},
{
"epoch": 0.09,
"learning_rate": 1.8141891891891895e-05,
"loss": 4.2977,
"step": 220
},
{
"epoch": 0.1,
"learning_rate": 1.8057432432432432e-05,
"loss": 4.3002,
"step": 230
},
{
"epoch": 0.1,
"learning_rate": 1.7972972972972976e-05,
"loss": 4.2797,
"step": 240
},
{
"epoch": 0.11,
"learning_rate": 1.7888513513513513e-05,
"loss": 4.2535,
"step": 250
},
{
"epoch": 0.11,
"learning_rate": 1.7804054054054057e-05,
"loss": 4.2415,
"step": 260
},
{
"epoch": 0.11,
"learning_rate": 1.7719594594594597e-05,
"loss": 4.2308,
"step": 270
},
{
"epoch": 0.12,
"learning_rate": 1.7635135135135137e-05,
"loss": 4.215,
"step": 280
},
{
"epoch": 0.12,
"learning_rate": 1.7550675675675678e-05,
"loss": 4.1792,
"step": 290
},
{
"epoch": 0.13,
"learning_rate": 1.7466216216216218e-05,
"loss": 4.2011,
"step": 300
},
{
"epoch": 0.13,
"learning_rate": 1.7381756756756758e-05,
"loss": 4.1563,
"step": 310
},
{
"epoch": 0.14,
"learning_rate": 1.72972972972973e-05,
"loss": 4.1189,
"step": 320
},
{
"epoch": 0.14,
"learning_rate": 1.721283783783784e-05,
"loss": 4.1216,
"step": 330
},
{
"epoch": 0.14,
"learning_rate": 1.712837837837838e-05,
"loss": 4.1172,
"step": 340
},
{
"epoch": 0.15,
"learning_rate": 1.704391891891892e-05,
"loss": 4.0805,
"step": 350
},
{
"epoch": 0.15,
"learning_rate": 1.6959459459459463e-05,
"loss": 4.0812,
"step": 360
},
{
"epoch": 0.16,
"learning_rate": 1.6875e-05,
"loss": 4.0771,
"step": 370
},
{
"epoch": 0.16,
"learning_rate": 1.679054054054054e-05,
"loss": 4.0753,
"step": 380
},
{
"epoch": 0.16,
"learning_rate": 1.670608108108108e-05,
"loss": 4.0547,
"step": 390
},
{
"epoch": 0.17,
"learning_rate": 1.662162162162162e-05,
"loss": 4.0507,
"step": 400
},
{
"epoch": 0.17,
"learning_rate": 1.6537162162162165e-05,
"loss": 4.0397,
"step": 410
},
{
"epoch": 0.18,
"learning_rate": 1.6452702702702702e-05,
"loss": 4.0156,
"step": 420
},
{
"epoch": 0.18,
"learning_rate": 1.6368243243243246e-05,
"loss": 3.9697,
"step": 430
},
{
"epoch": 0.19,
"learning_rate": 1.6283783783783786e-05,
"loss": 3.971,
"step": 440
},
{
"epoch": 0.19,
"learning_rate": 1.6199324324324326e-05,
"loss": 3.9323,
"step": 450
},
{
"epoch": 0.19,
"learning_rate": 1.6114864864864866e-05,
"loss": 3.9576,
"step": 460
},
{
"epoch": 0.2,
"learning_rate": 1.6030405405405407e-05,
"loss": 3.905,
"step": 470
},
{
"epoch": 0.2,
"learning_rate": 1.5945945945945947e-05,
"loss": 3.931,
"step": 480
},
{
"epoch": 0.21,
"learning_rate": 1.5861486486486487e-05,
"loss": 3.8975,
"step": 490
},
{
"epoch": 0.21,
"learning_rate": 1.5777027027027028e-05,
"loss": 3.9065,
"step": 500
},
{
"epoch": 0.22,
"learning_rate": 1.5692567567567568e-05,
"loss": 3.8508,
"step": 510
},
{
"epoch": 0.22,
"learning_rate": 1.560810810810811e-05,
"loss": 3.8269,
"step": 520
},
{
"epoch": 0.22,
"learning_rate": 1.552364864864865e-05,
"loss": 3.8256,
"step": 530
},
{
"epoch": 0.23,
"learning_rate": 1.543918918918919e-05,
"loss": 3.8214,
"step": 540
},
{
"epoch": 0.23,
"learning_rate": 1.5354729729729733e-05,
"loss": 3.7999,
"step": 550
},
{
"epoch": 0.24,
"learning_rate": 1.527027027027027e-05,
"loss": 3.8517,
"step": 560
},
{
"epoch": 0.24,
"learning_rate": 1.5185810810810813e-05,
"loss": 3.7893,
"step": 570
},
{
"epoch": 0.24,
"learning_rate": 1.5101351351351352e-05,
"loss": 3.8025,
"step": 580
},
{
"epoch": 0.25,
"learning_rate": 1.5016891891891894e-05,
"loss": 3.7113,
"step": 590
},
{
"epoch": 0.25,
"learning_rate": 1.4932432432432433e-05,
"loss": 3.7626,
"step": 600
},
{
"epoch": 0.26,
"learning_rate": 1.4847972972972975e-05,
"loss": 3.7344,
"step": 610
},
{
"epoch": 0.26,
"learning_rate": 1.4763513513513515e-05,
"loss": 3.7154,
"step": 620
},
{
"epoch": 0.27,
"learning_rate": 1.4679054054054055e-05,
"loss": 3.7228,
"step": 630
},
{
"epoch": 0.27,
"learning_rate": 1.4594594594594596e-05,
"loss": 3.7254,
"step": 640
},
{
"epoch": 0.27,
"learning_rate": 1.4510135135135138e-05,
"loss": 3.6979,
"step": 650
},
{
"epoch": 0.28,
"learning_rate": 1.4425675675675676e-05,
"loss": 3.714,
"step": 660
},
{
"epoch": 0.28,
"learning_rate": 1.4341216216216218e-05,
"loss": 3.6422,
"step": 670
},
{
"epoch": 0.29,
"learning_rate": 1.4256756756756759e-05,
"loss": 3.647,
"step": 680
},
{
"epoch": 0.29,
"learning_rate": 1.4172297297297299e-05,
"loss": 3.6642,
"step": 690
},
{
"epoch": 0.3,
"learning_rate": 1.408783783783784e-05,
"loss": 3.6719,
"step": 700
},
{
"epoch": 0.3,
"learning_rate": 1.4003378378378381e-05,
"loss": 3.6281,
"step": 710
},
{
"epoch": 0.3,
"learning_rate": 1.391891891891892e-05,
"loss": 3.6067,
"step": 720
},
{
"epoch": 0.31,
"learning_rate": 1.3834459459459462e-05,
"loss": 3.647,
"step": 730
},
{
"epoch": 0.31,
"learning_rate": 1.375e-05,
"loss": 3.5887,
"step": 740
},
{
"epoch": 0.32,
"learning_rate": 1.366554054054054e-05,
"loss": 3.5994,
"step": 750
},
{
"epoch": 0.32,
"learning_rate": 1.3581081081081083e-05,
"loss": 3.6033,
"step": 760
},
{
"epoch": 0.33,
"learning_rate": 1.3496621621621621e-05,
"loss": 3.5927,
"step": 770
},
{
"epoch": 0.33,
"learning_rate": 1.3412162162162163e-05,
"loss": 3.5493,
"step": 780
},
{
"epoch": 0.33,
"learning_rate": 1.3327702702702702e-05,
"loss": 3.5578,
"step": 790
},
{
"epoch": 0.34,
"learning_rate": 1.3243243243243244e-05,
"loss": 3.5862,
"step": 800
},
{
"epoch": 0.34,
"learning_rate": 1.3158783783783784e-05,
"loss": 3.5373,
"step": 810
},
{
"epoch": 0.35,
"learning_rate": 1.3074324324324326e-05,
"loss": 3.5463,
"step": 820
},
{
"epoch": 0.35,
"learning_rate": 1.2989864864864865e-05,
"loss": 3.5247,
"step": 830
},
{
"epoch": 0.35,
"learning_rate": 1.2905405405405407e-05,
"loss": 3.533,
"step": 840
},
{
"epoch": 0.36,
"learning_rate": 1.2820945945945946e-05,
"loss": 3.5241,
"step": 850
},
{
"epoch": 0.36,
"learning_rate": 1.2736486486486488e-05,
"loss": 3.4578,
"step": 860
},
{
"epoch": 0.37,
"learning_rate": 1.2652027027027028e-05,
"loss": 3.4349,
"step": 870
},
{
"epoch": 0.37,
"learning_rate": 1.2567567567567568e-05,
"loss": 3.505,
"step": 880
},
{
"epoch": 0.38,
"learning_rate": 1.2483108108108109e-05,
"loss": 3.5384,
"step": 890
},
{
"epoch": 0.38,
"learning_rate": 1.239864864864865e-05,
"loss": 3.5353,
"step": 900
},
{
"epoch": 0.38,
"learning_rate": 1.231418918918919e-05,
"loss": 3.4609,
"step": 910
},
{
"epoch": 0.39,
"learning_rate": 1.2229729729729731e-05,
"loss": 3.4578,
"step": 920
},
{
"epoch": 0.39,
"learning_rate": 1.214527027027027e-05,
"loss": 3.4581,
"step": 930
},
{
"epoch": 0.4,
"learning_rate": 1.2060810810810812e-05,
"loss": 3.4776,
"step": 940
},
{
"epoch": 0.4,
"learning_rate": 1.1976351351351352e-05,
"loss": 3.3736,
"step": 950
},
{
"epoch": 0.41,
"learning_rate": 1.1891891891891894e-05,
"loss": 3.4251,
"step": 960
},
{
"epoch": 0.41,
"learning_rate": 1.1807432432432433e-05,
"loss": 3.412,
"step": 970
},
{
"epoch": 0.41,
"learning_rate": 1.1722972972972975e-05,
"loss": 3.4269,
"step": 980
},
{
"epoch": 0.42,
"learning_rate": 1.1638513513513514e-05,
"loss": 3.3543,
"step": 990
},
{
"epoch": 0.42,
"learning_rate": 1.1554054054054056e-05,
"loss": 3.3914,
"step": 1000
},
{
"epoch": 0.43,
"learning_rate": 1.1469594594594596e-05,
"loss": 3.3941,
"step": 1010
},
{
"epoch": 0.43,
"learning_rate": 1.1385135135135136e-05,
"loss": 3.3541,
"step": 1020
},
{
"epoch": 0.43,
"learning_rate": 1.1300675675675677e-05,
"loss": 3.3634,
"step": 1030
},
{
"epoch": 0.44,
"learning_rate": 1.1216216216216219e-05,
"loss": 3.3366,
"step": 1040
},
{
"epoch": 0.44,
"learning_rate": 1.1131756756756757e-05,
"loss": 3.3162,
"step": 1050
},
{
"epoch": 0.45,
"learning_rate": 1.10472972972973e-05,
"loss": 3.3058,
"step": 1060
},
{
"epoch": 0.45,
"learning_rate": 1.0962837837837838e-05,
"loss": 3.3119,
"step": 1070
},
{
"epoch": 0.46,
"learning_rate": 1.087837837837838e-05,
"loss": 3.2882,
"step": 1080
},
{
"epoch": 0.46,
"learning_rate": 1.079391891891892e-05,
"loss": 3.3195,
"step": 1090
},
{
"epoch": 0.46,
"learning_rate": 1.0709459459459462e-05,
"loss": 3.3216,
"step": 1100
},
{
"epoch": 0.47,
"learning_rate": 1.0625e-05,
"loss": 3.288,
"step": 1110
},
{
"epoch": 0.47,
"learning_rate": 1.0540540540540541e-05,
"loss": 3.338,
"step": 1120
},
{
"epoch": 0.48,
"learning_rate": 1.0456081081081081e-05,
"loss": 3.3246,
"step": 1130
},
{
"epoch": 0.48,
"learning_rate": 1.0371621621621622e-05,
"loss": 3.2378,
"step": 1140
},
{
"epoch": 0.49,
"learning_rate": 1.0287162162162164e-05,
"loss": 3.2111,
"step": 1150
},
{
"epoch": 0.49,
"learning_rate": 1.0202702702702702e-05,
"loss": 3.2133,
"step": 1160
},
{
"epoch": 0.49,
"learning_rate": 1.0118243243243244e-05,
"loss": 3.3285,
"step": 1170
},
{
"epoch": 0.5,
"learning_rate": 1.0033783783783783e-05,
"loss": 3.2534,
"step": 1180
},
{
"epoch": 0.5,
"learning_rate": 9.949324324324325e-06,
"loss": 3.2644,
"step": 1190
},
{
"epoch": 0.51,
"learning_rate": 9.864864864864865e-06,
"loss": 3.1807,
"step": 1200
},
{
"epoch": 0.51,
"learning_rate": 9.780405405405407e-06,
"loss": 3.2043,
"step": 1210
},
{
"epoch": 0.52,
"learning_rate": 9.695945945945948e-06,
"loss": 3.2501,
"step": 1220
},
{
"epoch": 0.52,
"learning_rate": 9.611486486486488e-06,
"loss": 3.1766,
"step": 1230
},
{
"epoch": 0.52,
"learning_rate": 9.527027027027028e-06,
"loss": 3.2242,
"step": 1240
},
{
"epoch": 0.53,
"learning_rate": 9.442567567567569e-06,
"loss": 3.2155,
"step": 1250
},
{
"epoch": 0.53,
"learning_rate": 9.358108108108109e-06,
"loss": 3.2137,
"step": 1260
},
{
"epoch": 0.54,
"learning_rate": 9.27364864864865e-06,
"loss": 3.231,
"step": 1270
},
{
"epoch": 0.54,
"learning_rate": 9.189189189189191e-06,
"loss": 3.1844,
"step": 1280
},
{
"epoch": 0.54,
"learning_rate": 9.104729729729732e-06,
"loss": 3.1806,
"step": 1290
},
{
"epoch": 0.55,
"learning_rate": 9.02027027027027e-06,
"loss": 3.1665,
"step": 1300
},
{
"epoch": 0.55,
"learning_rate": 8.93581081081081e-06,
"loss": 3.195,
"step": 1310
},
{
"epoch": 0.56,
"learning_rate": 8.851351351351351e-06,
"loss": 3.1486,
"step": 1320
},
{
"epoch": 0.56,
"learning_rate": 8.766891891891893e-06,
"loss": 3.1789,
"step": 1330
},
{
"epoch": 0.57,
"learning_rate": 8.682432432432433e-06,
"loss": 3.203,
"step": 1340
},
{
"epoch": 0.57,
"learning_rate": 8.597972972972974e-06,
"loss": 3.2001,
"step": 1350
},
{
"epoch": 0.57,
"learning_rate": 8.513513513513514e-06,
"loss": 3.1426,
"step": 1360
},
{
"epoch": 0.58,
"learning_rate": 8.429054054054054e-06,
"loss": 3.1418,
"step": 1370
},
{
"epoch": 0.58,
"learning_rate": 8.344594594594594e-06,
"loss": 3.1561,
"step": 1380
},
{
"epoch": 0.59,
"learning_rate": 8.260135135135135e-06,
"loss": 3.12,
"step": 1390
},
{
"epoch": 0.59,
"learning_rate": 8.175675675675677e-06,
"loss": 3.1581,
"step": 1400
},
{
"epoch": 0.6,
"learning_rate": 8.091216216216217e-06,
"loss": 3.1576,
"step": 1410
},
{
"epoch": 0.6,
"learning_rate": 8.006756756756757e-06,
"loss": 3.1007,
"step": 1420
},
{
"epoch": 0.6,
"learning_rate": 7.922297297297298e-06,
"loss": 3.0515,
"step": 1430
},
{
"epoch": 0.61,
"learning_rate": 7.837837837837838e-06,
"loss": 3.1258,
"step": 1440
},
{
"epoch": 0.61,
"learning_rate": 7.753378378378378e-06,
"loss": 3.1647,
"step": 1450
},
{
"epoch": 0.62,
"learning_rate": 7.668918918918919e-06,
"loss": 3.0716,
"step": 1460
},
{
"epoch": 0.62,
"learning_rate": 7.58445945945946e-06,
"loss": 3.1271,
"step": 1470
},
{
"epoch": 0.62,
"learning_rate": 7.500000000000001e-06,
"loss": 3.0764,
"step": 1480
},
{
"epoch": 0.63,
"learning_rate": 7.415540540540541e-06,
"loss": 3.0852,
"step": 1490
},
{
"epoch": 0.63,
"learning_rate": 7.331081081081082e-06,
"loss": 3.0582,
"step": 1500
},
{
"epoch": 0.64,
"learning_rate": 7.246621621621622e-06,
"loss": 3.0305,
"step": 1510
},
{
"epoch": 0.64,
"learning_rate": 7.162162162162163e-06,
"loss": 2.9988,
"step": 1520
},
{
"epoch": 0.65,
"learning_rate": 7.0777027027027035e-06,
"loss": 3.0951,
"step": 1530
},
{
"epoch": 0.65,
"learning_rate": 6.993243243243244e-06,
"loss": 3.0171,
"step": 1540
},
{
"epoch": 0.65,
"learning_rate": 6.908783783783785e-06,
"loss": 3.0459,
"step": 1550
},
{
"epoch": 0.66,
"learning_rate": 6.824324324324325e-06,
"loss": 3.1043,
"step": 1560
},
{
"epoch": 0.66,
"learning_rate": 6.739864864864866e-06,
"loss": 3.0592,
"step": 1570
},
{
"epoch": 0.67,
"learning_rate": 6.655405405405406e-06,
"loss": 3.0215,
"step": 1580
},
{
"epoch": 0.67,
"learning_rate": 6.570945945945947e-06,
"loss": 3.0157,
"step": 1590
},
{
"epoch": 0.68,
"learning_rate": 6.486486486486487e-06,
"loss": 2.9574,
"step": 1600
},
{
"epoch": 0.68,
"learning_rate": 6.402027027027028e-06,
"loss": 3.0353,
"step": 1610
},
{
"epoch": 0.68,
"learning_rate": 6.317567567567569e-06,
"loss": 3.0541,
"step": 1620
},
{
"epoch": 0.69,
"learning_rate": 6.233108108108109e-06,
"loss": 3.1111,
"step": 1630
},
{
"epoch": 0.69,
"learning_rate": 6.1486486486486495e-06,
"loss": 3.101,
"step": 1640
},
{
"epoch": 0.7,
"learning_rate": 6.06418918918919e-06,
"loss": 2.9928,
"step": 1650
},
{
"epoch": 0.7,
"learning_rate": 5.979729729729731e-06,
"loss": 2.9631,
"step": 1660
},
{
"epoch": 0.71,
"learning_rate": 5.8952702702702705e-06,
"loss": 3.0677,
"step": 1670
},
{
"epoch": 0.71,
"learning_rate": 5.810810810810811e-06,
"loss": 2.9865,
"step": 1680
},
{
"epoch": 0.71,
"learning_rate": 5.726351351351351e-06,
"loss": 3.0248,
"step": 1690
},
{
"epoch": 0.72,
"learning_rate": 5.641891891891892e-06,
"loss": 2.9862,
"step": 1700
},
{
"epoch": 0.72,
"learning_rate": 5.557432432432433e-06,
"loss": 3.0454,
"step": 1710
},
{
"epoch": 0.73,
"learning_rate": 5.472972972972973e-06,
"loss": 3.0235,
"step": 1720
},
{
"epoch": 0.73,
"learning_rate": 5.388513513513513e-06,
"loss": 2.9474,
"step": 1730
},
{
"epoch": 0.73,
"learning_rate": 5.304054054054054e-06,
"loss": 2.9162,
"step": 1740
},
{
"epoch": 0.74,
"learning_rate": 5.219594594594595e-06,
"loss": 2.8715,
"step": 1750
},
{
"epoch": 0.74,
"learning_rate": 5.135135135135135e-06,
"loss": 3.0131,
"step": 1760
},
{
"epoch": 0.75,
"learning_rate": 5.050675675675676e-06,
"loss": 3.059,
"step": 1770
},
{
"epoch": 0.75,
"learning_rate": 4.9662162162162165e-06,
"loss": 3.0185,
"step": 1780
},
{
"epoch": 0.76,
"learning_rate": 4.881756756756757e-06,
"loss": 2.9794,
"step": 1790
},
{
"epoch": 0.76,
"learning_rate": 4.797297297297297e-06,
"loss": 2.9555,
"step": 1800
},
{
"epoch": 0.76,
"learning_rate": 4.712837837837838e-06,
"loss": 2.9719,
"step": 1810
},
{
"epoch": 0.77,
"learning_rate": 4.628378378378379e-06,
"loss": 2.9083,
"step": 1820
},
{
"epoch": 0.77,
"learning_rate": 4.543918918918919e-06,
"loss": 2.9747,
"step": 1830
},
{
"epoch": 0.78,
"learning_rate": 4.45945945945946e-06,
"loss": 2.8957,
"step": 1840
},
{
"epoch": 0.78,
"learning_rate": 4.3750000000000005e-06,
"loss": 3.0,
"step": 1850
},
{
"epoch": 0.79,
"learning_rate": 4.290540540540541e-06,
"loss": 2.9275,
"step": 1860
},
{
"epoch": 0.79,
"learning_rate": 4.206081081081081e-06,
"loss": 2.956,
"step": 1870
},
{
"epoch": 0.79,
"learning_rate": 4.121621621621622e-06,
"loss": 2.9962,
"step": 1880
},
{
"epoch": 0.8,
"learning_rate": 4.037162162162163e-06,
"loss": 2.9561,
"step": 1890
},
{
"epoch": 0.8,
"learning_rate": 3.952702702702703e-06,
"loss": 3.0017,
"step": 1900
},
{
"epoch": 0.81,
"learning_rate": 3.868243243243244e-06,
"loss": 2.9761,
"step": 1910
},
{
"epoch": 0.81,
"learning_rate": 3.7837837837837844e-06,
"loss": 2.8731,
"step": 1920
},
{
"epoch": 0.82,
"learning_rate": 3.6993243243243247e-06,
"loss": 2.9409,
"step": 1930
},
{
"epoch": 0.82,
"learning_rate": 3.6148648648648655e-06,
"loss": 2.9489,
"step": 1940
},
{
"epoch": 0.82,
"learning_rate": 3.5304054054054053e-06,
"loss": 2.883,
"step": 1950
},
{
"epoch": 0.83,
"learning_rate": 3.445945945945946e-06,
"loss": 2.9262,
"step": 1960
},
{
"epoch": 0.83,
"learning_rate": 3.3614864864864864e-06,
"loss": 2.8513,
"step": 1970
},
{
"epoch": 0.84,
"learning_rate": 3.277027027027027e-06,
"loss": 2.9792,
"step": 1980
},
{
"epoch": 0.84,
"learning_rate": 3.192567567567568e-06,
"loss": 2.8856,
"step": 1990
},
{
"epoch": 0.84,
"learning_rate": 3.1081081081081082e-06,
"loss": 2.9072,
"step": 2000
},
{
"epoch": 0.85,
"learning_rate": 3.023648648648649e-06,
"loss": 2.8919,
"step": 2010
},
{
"epoch": 0.85,
"learning_rate": 2.9391891891891893e-06,
"loss": 2.9052,
"step": 2020
},
{
"epoch": 0.86,
"learning_rate": 2.85472972972973e-06,
"loss": 2.9392,
"step": 2030
},
{
"epoch": 0.86,
"learning_rate": 2.7702702702702703e-06,
"loss": 2.91,
"step": 2040
},
{
"epoch": 0.87,
"learning_rate": 2.685810810810811e-06,
"loss": 2.9045,
"step": 2050
},
{
"epoch": 0.87,
"learning_rate": 2.601351351351352e-06,
"loss": 2.8834,
"step": 2060
},
{
"epoch": 0.87,
"learning_rate": 2.516891891891892e-06,
"loss": 2.9285,
"step": 2070
},
{
"epoch": 0.88,
"learning_rate": 2.432432432432433e-06,
"loss": 2.8568,
"step": 2080
},
{
"epoch": 0.88,
"learning_rate": 2.347972972972973e-06,
"loss": 2.9196,
"step": 2090
},
{
"epoch": 0.89,
"learning_rate": 2.2635135135135135e-06,
"loss": 2.911,
"step": 2100
},
{
"epoch": 0.89,
"learning_rate": 2.1790540540540543e-06,
"loss": 2.8964,
"step": 2110
},
{
"epoch": 0.9,
"learning_rate": 2.0945945945945946e-06,
"loss": 2.9037,
"step": 2120
},
{
"epoch": 0.9,
"learning_rate": 2.0101351351351353e-06,
"loss": 2.8877,
"step": 2130
},
{
"epoch": 0.9,
"learning_rate": 1.925675675675676e-06,
"loss": 2.8896,
"step": 2140
},
{
"epoch": 0.91,
"learning_rate": 1.8412162162162164e-06,
"loss": 2.8499,
"step": 2150
},
{
"epoch": 0.91,
"learning_rate": 1.756756756756757e-06,
"loss": 2.9109,
"step": 2160
},
{
"epoch": 0.92,
"learning_rate": 1.6722972972972977e-06,
"loss": 2.8985,
"step": 2170
},
{
"epoch": 0.92,
"learning_rate": 1.5878378378378378e-06,
"loss": 2.8156,
"step": 2180
},
{
"epoch": 0.92,
"learning_rate": 1.5033783783783785e-06,
"loss": 2.849,
"step": 2190
},
{
"epoch": 0.93,
"learning_rate": 1.418918918918919e-06,
"loss": 2.9026,
"step": 2200
},
{
"epoch": 0.93,
"learning_rate": 1.3344594594594596e-06,
"loss": 2.9037,
"step": 2210
},
{
"epoch": 0.94,
"learning_rate": 1.25e-06,
"loss": 2.9295,
"step": 2220
},
{
"epoch": 0.94,
"learning_rate": 1.1655405405405406e-06,
"loss": 2.7775,
"step": 2230
},
{
"epoch": 0.95,
"learning_rate": 1.0810810810810812e-06,
"loss": 3.0173,
"step": 2240
},
{
"epoch": 0.95,
"learning_rate": 9.966216216216217e-07,
"loss": 2.8453,
"step": 2250
},
{
"epoch": 0.95,
"learning_rate": 9.121621621621622e-07,
"loss": 2.9383,
"step": 2260
},
{
"epoch": 0.96,
"learning_rate": 8.277027027027028e-07,
"loss": 2.8994,
"step": 2270
},
{
"epoch": 0.96,
"learning_rate": 7.432432432432434e-07,
"loss": 2.8407,
"step": 2280
},
{
"epoch": 0.97,
"learning_rate": 6.587837837837838e-07,
"loss": 2.9052,
"step": 2290
},
{
"epoch": 0.97,
"learning_rate": 5.743243243243245e-07,
"loss": 2.9265,
"step": 2300
},
{
"epoch": 0.98,
"learning_rate": 4.898648648648649e-07,
"loss": 2.9159,
"step": 2310
},
{
"epoch": 0.98,
"learning_rate": 4.0540540540540546e-07,
"loss": 2.855,
"step": 2320
},
{
"epoch": 0.98,
"learning_rate": 3.2094594594594594e-07,
"loss": 2.9238,
"step": 2330
},
{
"epoch": 0.99,
"learning_rate": 2.3648648648648652e-07,
"loss": 2.8947,
"step": 2340
},
{
"epoch": 0.99,
"learning_rate": 1.5202702702702706e-07,
"loss": 2.9042,
"step": 2350
},
{
"epoch": 1.0,
"learning_rate": 6.756756756756757e-08,
"loss": 2.9491,
"step": 2360
},
{
"epoch": 1.0,
"step": 2368,
"total_flos": 2.9376133569216e+18,
"train_loss": 3.4182597672617114,
"train_runtime": 551.9339,
"train_samples_per_second": 68.646,
"train_steps_per_second": 4.29
}
],
"logging_steps": 10,
"max_steps": 2368,
"num_train_epochs": 1,
"save_steps": 500,
"total_flos": 2.9376133569216e+18,
"trial_name": null,
"trial_params": null
}