DISC-FinLLM / trainer_state.json
Go4miii's picture
Upload 17 files
b68d805
raw history blame
No virus
24.3 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"global_step": 1974,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.9996834033646177e-05,
"loss": 0.9175,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 4.998733693645213e-05,
"loss": 0.7595,
"step": 20
},
{
"epoch": 0.03,
"learning_rate": 4.997151111381707e-05,
"loss": 0.7375,
"step": 30
},
{
"epoch": 0.04,
"learning_rate": 4.9949360574062774e-05,
"loss": 0.7227,
"step": 40
},
{
"epoch": 0.05,
"learning_rate": 4.9920890927418316e-05,
"loss": 0.7147,
"step": 50
},
{
"epoch": 0.06,
"learning_rate": 4.988610938459917e-05,
"loss": 0.7102,
"step": 60
},
{
"epoch": 0.07,
"learning_rate": 4.9845024754980876e-05,
"loss": 0.7056,
"step": 70
},
{
"epoch": 0.08,
"learning_rate": 4.979764744436784e-05,
"loss": 0.7128,
"step": 80
},
{
"epoch": 0.09,
"learning_rate": 4.9743989452357756e-05,
"loss": 0.6982,
"step": 90
},
{
"epoch": 0.1,
"learning_rate": 4.968406436930243e-05,
"loss": 0.7258,
"step": 100
},
{
"epoch": 0.11,
"learning_rate": 4.961788737286559e-05,
"loss": 0.7095,
"step": 110
},
{
"epoch": 0.12,
"learning_rate": 4.954547522417877e-05,
"loss": 0.7048,
"step": 120
},
{
"epoch": 0.13,
"learning_rate": 4.946684626359607e-05,
"loss": 0.6805,
"step": 130
},
{
"epoch": 0.14,
"learning_rate": 4.938202040604898e-05,
"loss": 0.6798,
"step": 140
},
{
"epoch": 0.15,
"learning_rate": 4.929101913600238e-05,
"loss": 0.7134,
"step": 150
},
{
"epoch": 0.16,
"learning_rate": 4.919386550201299e-05,
"loss": 0.6895,
"step": 160
},
{
"epoch": 0.17,
"learning_rate": 4.909058411089174e-05,
"loss": 0.705,
"step": 170
},
{
"epoch": 0.18,
"learning_rate": 4.8981201121471356e-05,
"loss": 0.6712,
"step": 180
},
{
"epoch": 0.19,
"learning_rate": 4.886574423798097e-05,
"loss": 0.6744,
"step": 190
},
{
"epoch": 0.2,
"learning_rate": 4.874424270302927e-05,
"loss": 0.6675,
"step": 200
},
{
"epoch": 0.21,
"learning_rate": 4.861672729019797e-05,
"loss": 0.6726,
"step": 210
},
{
"epoch": 0.22,
"learning_rate": 4.848323029624761e-05,
"loss": 0.6821,
"step": 220
},
{
"epoch": 0.23,
"learning_rate": 4.834378553293748e-05,
"loss": 0.7133,
"step": 230
},
{
"epoch": 0.24,
"learning_rate": 4.81984283184619e-05,
"loss": 0.6681,
"step": 240
},
{
"epoch": 0.25,
"learning_rate": 4.804719546850487e-05,
"loss": 0.682,
"step": 250
},
{
"epoch": 0.26,
"learning_rate": 4.789012528691558e-05,
"loss": 0.6755,
"step": 260
},
{
"epoch": 0.27,
"learning_rate": 4.772725755600682e-05,
"loss": 0.68,
"step": 270
},
{
"epoch": 0.28,
"learning_rate": 4.755863352647909e-05,
"loss": 0.6663,
"step": 280
},
{
"epoch": 0.29,
"learning_rate": 4.738429590697271e-05,
"loss": 0.6672,
"step": 290
},
{
"epoch": 0.3,
"learning_rate": 4.720428885325069e-05,
"loss": 0.6466,
"step": 300
},
{
"epoch": 0.31,
"learning_rate": 4.701865795701505e-05,
"loss": 0.6668,
"step": 310
},
{
"epoch": 0.32,
"learning_rate": 4.684682059461469e-05,
"loss": 0.6591,
"step": 320
},
{
"epoch": 0.33,
"learning_rate": 4.665063509461097e-05,
"loss": 0.661,
"step": 330
},
{
"epoch": 0.34,
"learning_rate": 4.644896598002736e-05,
"loss": 0.6749,
"step": 340
},
{
"epoch": 0.35,
"learning_rate": 4.624186432907437e-05,
"loss": 0.6654,
"step": 350
},
{
"epoch": 0.36,
"learning_rate": 4.602938259590072e-05,
"loss": 0.6716,
"step": 360
},
{
"epoch": 0.37,
"learning_rate": 4.581157459730783e-05,
"loss": 0.6796,
"step": 370
},
{
"epoch": 0.39,
"learning_rate": 4.558849549911931e-05,
"loss": 0.6794,
"step": 380
},
{
"epoch": 0.4,
"learning_rate": 4.536020180220871e-05,
"loss": 0.6651,
"step": 390
},
{
"epoch": 0.41,
"learning_rate": 4.515032675559024e-05,
"loss": 0.6823,
"step": 400
},
{
"epoch": 0.42,
"learning_rate": 4.4912285699446786e-05,
"loss": 0.6677,
"step": 410
},
{
"epoch": 0.43,
"learning_rate": 4.4669201313179155e-05,
"loss": 0.6704,
"step": 420
},
{
"epoch": 0.44,
"learning_rate": 4.442113516454638e-05,
"loss": 0.6481,
"step": 430
},
{
"epoch": 0.45,
"learning_rate": 4.416815008307488e-05,
"loss": 0.665,
"step": 440
},
{
"epoch": 0.46,
"learning_rate": 4.391031014414514e-05,
"loss": 0.6658,
"step": 450
},
{
"epoch": 0.47,
"learning_rate": 4.364768065276284e-05,
"loss": 0.6699,
"step": 460
},
{
"epoch": 0.48,
"learning_rate": 4.338032812701867e-05,
"loss": 0.6664,
"step": 470
},
{
"epoch": 0.49,
"learning_rate": 4.310832028124069e-05,
"loss": 0.6817,
"step": 480
},
{
"epoch": 0.5,
"learning_rate": 4.283172600884393e-05,
"loss": 0.6791,
"step": 490
},
{
"epoch": 0.51,
"learning_rate": 4.2550615364881194e-05,
"loss": 0.6679,
"step": 500
},
{
"epoch": 0.52,
"learning_rate": 4.226505954829973e-05,
"loss": 0.6476,
"step": 510
},
{
"epoch": 0.53,
"learning_rate": 4.197513088390813e-05,
"loss": 0.6721,
"step": 520
},
{
"epoch": 0.54,
"learning_rate": 4.1680902804058095e-05,
"loss": 0.6563,
"step": 530
},
{
"epoch": 0.55,
"learning_rate": 4.138244983004574e-05,
"loss": 0.6391,
"step": 540
},
{
"epoch": 0.56,
"learning_rate": 4.107984755323697e-05,
"loss": 0.6672,
"step": 550
},
{
"epoch": 0.57,
"learning_rate": 4.077317261592194e-05,
"loss": 0.6497,
"step": 560
},
{
"epoch": 0.58,
"learning_rate": 4.04625026919033e-05,
"loss": 0.668,
"step": 570
},
{
"epoch": 0.59,
"learning_rate": 4.0147916466823174e-05,
"loss": 0.6682,
"step": 580
},
{
"epoch": 0.6,
"learning_rate": 3.982949361823388e-05,
"loss": 0.6352,
"step": 590
},
{
"epoch": 0.61,
"learning_rate": 3.950731479541743e-05,
"loss": 0.6698,
"step": 600
},
{
"epoch": 0.62,
"learning_rate": 3.918146159895882e-05,
"loss": 0.6549,
"step": 610
},
{
"epoch": 0.63,
"learning_rate": 3.8852016560078605e-05,
"loss": 0.6516,
"step": 620
},
{
"epoch": 0.64,
"learning_rate": 3.851906311972943e-05,
"loss": 0.6629,
"step": 630
},
{
"epoch": 0.65,
"learning_rate": 3.821647502051616e-05,
"loss": 0.6764,
"step": 640
},
{
"epoch": 0.66,
"learning_rate": 3.787708866250794e-05,
"loss": 0.6415,
"step": 650
},
{
"epoch": 0.67,
"learning_rate": 3.7534440830144466e-05,
"loss": 0.6463,
"step": 660
},
{
"epoch": 0.68,
"learning_rate": 3.71886183083464e-05,
"loss": 0.6508,
"step": 670
},
{
"epoch": 0.69,
"learning_rate": 3.683970868611123e-05,
"loss": 0.6411,
"step": 680
},
{
"epoch": 0.7,
"learning_rate": 3.648780033432891e-05,
"loss": 0.6266,
"step": 690
},
{
"epoch": 0.71,
"learning_rate": 3.613298238339955e-05,
"loss": 0.6409,
"step": 700
},
{
"epoch": 0.72,
"learning_rate": 3.5775344700658705e-05,
"loss": 0.6594,
"step": 710
},
{
"epoch": 0.73,
"learning_rate": 3.5414977867616006e-05,
"loss": 0.6427,
"step": 720
},
{
"epoch": 0.74,
"learning_rate": 3.505197315701292e-05,
"loss": 0.6462,
"step": 730
},
{
"epoch": 0.75,
"learning_rate": 3.468642250970547e-05,
"loss": 0.6277,
"step": 740
},
{
"epoch": 0.76,
"learning_rate": 3.431841851137764e-05,
"loss": 0.6551,
"step": 750
},
{
"epoch": 0.77,
"learning_rate": 3.394805436909157e-05,
"loss": 0.6402,
"step": 760
},
{
"epoch": 0.78,
"learning_rate": 3.357542388768033e-05,
"loss": 0.6515,
"step": 770
},
{
"epoch": 0.79,
"learning_rate": 3.3200621445989226e-05,
"loss": 0.6489,
"step": 780
},
{
"epoch": 0.8,
"learning_rate": 3.282374197297185e-05,
"loss": 0.6568,
"step": 790
},
{
"epoch": 0.81,
"learning_rate": 3.2444880923646674e-05,
"loss": 0.615,
"step": 800
},
{
"epoch": 0.82,
"learning_rate": 3.20641342549205e-05,
"loss": 0.6447,
"step": 810
},
{
"epoch": 0.83,
"learning_rate": 3.168159840128472e-05,
"loss": 0.6159,
"step": 820
},
{
"epoch": 0.84,
"learning_rate": 3.129737025039068e-05,
"loss": 0.6347,
"step": 830
},
{
"epoch": 0.85,
"learning_rate": 3.091154711851022e-05,
"loss": 0.6361,
"step": 840
},
{
"epoch": 0.86,
"learning_rate": 3.052422672588765e-05,
"loss": 0.6504,
"step": 850
},
{
"epoch": 0.87,
"learning_rate": 3.013550717198948e-05,
"loss": 0.6467,
"step": 860
},
{
"epoch": 0.88,
"learning_rate": 2.9745486910657993e-05,
"loss": 0.6364,
"step": 870
},
{
"epoch": 0.89,
"learning_rate": 2.9354264725175185e-05,
"loss": 0.6361,
"step": 880
},
{
"epoch": 0.9,
"learning_rate": 2.8961939703243122e-05,
"loss": 0.6441,
"step": 890
},
{
"epoch": 0.91,
"learning_rate": 2.856861121188735e-05,
"loss": 0.6404,
"step": 900
},
{
"epoch": 0.92,
"learning_rate": 2.8174378872289446e-05,
"loss": 0.6307,
"step": 910
},
{
"epoch": 0.93,
"learning_rate": 2.777934253455522e-05,
"loss": 0.6484,
"step": 920
},
{
"epoch": 0.94,
"learning_rate": 2.7383602252424985e-05,
"loss": 0.6237,
"step": 930
},
{
"epoch": 0.95,
"learning_rate": 2.6987258257932175e-05,
"loss": 0.6161,
"step": 940
},
{
"epoch": 0.96,
"learning_rate": 2.6590410936016895e-05,
"loss": 0.6381,
"step": 950
},
{
"epoch": 0.97,
"learning_rate": 2.619316079910063e-05,
"loss": 0.6366,
"step": 960
},
{
"epoch": 0.98,
"learning_rate": 2.5795608461628802e-05,
"loss": 0.6202,
"step": 970
},
{
"epoch": 0.99,
"learning_rate": 2.5397854614587334e-05,
"loss": 0.6334,
"step": 980
},
{
"epoch": 1.0,
"learning_rate": 2.5e-05,
"loss": 0.5954,
"step": 990
},
{
"epoch": 1.01,
"learning_rate": 2.460214538541267e-05,
"loss": 0.4963,
"step": 1000
},
{
"epoch": 1.02,
"learning_rate": 2.4204391538371207e-05,
"loss": 0.487,
"step": 1010
},
{
"epoch": 1.03,
"learning_rate": 2.3806839200899377e-05,
"loss": 0.489,
"step": 1020
},
{
"epoch": 1.04,
"learning_rate": 2.3409589063983117e-05,
"loss": 0.4805,
"step": 1030
},
{
"epoch": 1.05,
"learning_rate": 2.3012741742067838e-05,
"loss": 0.4907,
"step": 1040
},
{
"epoch": 1.06,
"learning_rate": 2.261639774757503e-05,
"loss": 0.4719,
"step": 1050
},
{
"epoch": 1.07,
"learning_rate": 2.2220657465444782e-05,
"loss": 0.4914,
"step": 1060
},
{
"epoch": 1.08,
"learning_rate": 2.182562112771056e-05,
"loss": 0.4775,
"step": 1070
},
{
"epoch": 1.09,
"learning_rate": 2.143138878811265e-05,
"loss": 0.4935,
"step": 1080
},
{
"epoch": 1.1,
"learning_rate": 2.1038060296756883e-05,
"loss": 0.5082,
"step": 1090
},
{
"epoch": 1.11,
"learning_rate": 2.064573527482482e-05,
"loss": 0.4868,
"step": 1100
},
{
"epoch": 1.12,
"learning_rate": 2.025451308934201e-05,
"loss": 0.4988,
"step": 1110
},
{
"epoch": 1.13,
"learning_rate": 1.9864492828010526e-05,
"loss": 0.4653,
"step": 1120
},
{
"epoch": 1.14,
"learning_rate": 1.9475773274112354e-05,
"loss": 0.4915,
"step": 1130
},
{
"epoch": 1.16,
"learning_rate": 1.9088452881489787e-05,
"loss": 0.4763,
"step": 1140
},
{
"epoch": 1.17,
"learning_rate": 1.8702629749609324e-05,
"loss": 0.4807,
"step": 1150
},
{
"epoch": 1.18,
"learning_rate": 1.8318401598715284e-05,
"loss": 0.4653,
"step": 1160
},
{
"epoch": 1.19,
"learning_rate": 1.793586574507951e-05,
"loss": 0.4778,
"step": 1170
},
{
"epoch": 1.2,
"learning_rate": 1.7555119076353338e-05,
"loss": 0.4839,
"step": 1180
},
{
"epoch": 1.21,
"learning_rate": 1.7176258027028152e-05,
"loss": 0.4718,
"step": 1190
},
{
"epoch": 1.22,
"learning_rate": 1.6799378554010773e-05,
"loss": 0.4793,
"step": 1200
},
{
"epoch": 1.23,
"learning_rate": 1.6424576112319672e-05,
"loss": 0.4825,
"step": 1210
},
{
"epoch": 1.24,
"learning_rate": 1.6051945630908426e-05,
"loss": 0.4857,
"step": 1220
},
{
"epoch": 1.25,
"learning_rate": 1.5681581488622367e-05,
"loss": 0.4802,
"step": 1230
},
{
"epoch": 1.26,
"learning_rate": 1.5313577490294538e-05,
"loss": 0.4812,
"step": 1240
},
{
"epoch": 1.27,
"learning_rate": 1.4948026842987084e-05,
"loss": 0.4682,
"step": 1250
},
{
"epoch": 1.28,
"learning_rate": 1.4585022132384008e-05,
"loss": 0.4974,
"step": 1260
},
{
"epoch": 1.29,
"learning_rate": 1.4224655299341304e-05,
"loss": 0.4737,
"step": 1270
},
{
"epoch": 1.3,
"learning_rate": 1.3867017616600456e-05,
"loss": 0.4877,
"step": 1280
},
{
"epoch": 1.31,
"learning_rate": 1.3512199665671094e-05,
"loss": 0.4753,
"step": 1290
},
{
"epoch": 1.32,
"learning_rate": 1.316029131388878e-05,
"loss": 0.4638,
"step": 1300
},
{
"epoch": 1.33,
"learning_rate": 1.2811381691653607e-05,
"loss": 0.4626,
"step": 1310
},
{
"epoch": 1.34,
"learning_rate": 1.2465559169855535e-05,
"loss": 0.4786,
"step": 1320
},
{
"epoch": 1.35,
"learning_rate": 1.212291133749206e-05,
"loss": 0.4717,
"step": 1330
},
{
"epoch": 1.36,
"learning_rate": 1.178352497948384e-05,
"loss": 0.4803,
"step": 1340
},
{
"epoch": 1.37,
"learning_rate": 1.1447486054694112e-05,
"loss": 0.4803,
"step": 1350
},
{
"epoch": 1.38,
"learning_rate": 1.1114879674157233e-05,
"loss": 0.4739,
"step": 1360
},
{
"epoch": 1.39,
"learning_rate": 1.0785790079522001e-05,
"loss": 0.471,
"step": 1370
},
{
"epoch": 1.4,
"learning_rate": 1.046030062171512e-05,
"loss": 0.4799,
"step": 1380
},
{
"epoch": 1.41,
"learning_rate": 1.0138493739830352e-05,
"loss": 0.4689,
"step": 1390
},
{
"epoch": 1.42,
"learning_rate": 9.820450940248544e-06,
"loss": 0.4599,
"step": 1400
},
{
"epoch": 1.43,
"learning_rate": 9.506252775993882e-06,
"loss": 0.5019,
"step": 1410
},
{
"epoch": 1.44,
"learning_rate": 9.195978826331697e-06,
"loss": 0.4764,
"step": 1420
},
{
"epoch": 1.45,
"learning_rate": 8.889707676612791e-06,
"loss": 0.4579,
"step": 1430
},
{
"epoch": 1.46,
"learning_rate": 8.587516898369589e-06,
"loss": 0.4592,
"step": 1440
},
{
"epoch": 1.47,
"learning_rate": 8.289483029668972e-06,
"loss": 0.4861,
"step": 1450
},
{
"epoch": 1.48,
"learning_rate": 7.99568155572701e-06,
"loss": 0.4753,
"step": 1460
},
{
"epoch": 1.49,
"learning_rate": 7.706186889790209e-06,
"loss": 0.4929,
"step": 1470
},
{
"epoch": 1.5,
"learning_rate": 7.421072354288302e-06,
"loss": 0.4594,
"step": 1480
},
{
"epoch": 1.51,
"learning_rate": 7.140410162263414e-06,
"loss": 0.4912,
"step": 1490
},
{
"epoch": 1.52,
"learning_rate": 6.86427139908008e-06,
"loss": 0.4681,
"step": 1500
},
{
"epoch": 1.53,
"learning_rate": 6.5927260044209655e-06,
"loss": 0.4816,
"step": 1510
},
{
"epoch": 1.54,
"learning_rate": 6.3258427545727e-06,
"loss": 0.4723,
"step": 1520
},
{
"epoch": 1.55,
"learning_rate": 6.063689245006443e-06,
"loss": 0.4856,
"step": 1530
},
{
"epoch": 1.56,
"learning_rate": 5.806331873257462e-06,
"loss": 0.4829,
"step": 1540
},
{
"epoch": 1.57,
"learning_rate": 5.553835822108152e-06,
"loss": 0.4741,
"step": 1550
},
{
"epoch": 1.58,
"learning_rate": 5.306265043078693e-06,
"loss": 0.4654,
"step": 1560
},
{
"epoch": 1.59,
"learning_rate": 5.0636822402296165e-06,
"loss": 0.4668,
"step": 1570
},
{
"epoch": 1.6,
"learning_rate": 4.826148854280277e-06,
"loss": 0.4723,
"step": 1580
},
{
"epoch": 1.61,
"learning_rate": 4.593725047047293e-06,
"loss": 0.4639,
"step": 1590
},
{
"epoch": 1.62,
"learning_rate": 4.3664696862069505e-06,
"loss": 0.4777,
"step": 1600
},
{
"epoch": 1.63,
"learning_rate": 4.144440330385347e-06,
"loss": 0.4546,
"step": 1610
},
{
"epoch": 1.64,
"learning_rate": 3.927693214580075e-06,
"loss": 0.4543,
"step": 1620
},
{
"epoch": 1.65,
"learning_rate": 3.71628323591722e-06,
"loss": 0.4543,
"step": 1630
},
{
"epoch": 1.66,
"learning_rate": 3.5102639397471214e-06,
"loss": 0.4659,
"step": 1640
},
{
"epoch": 1.67,
"learning_rate": 3.3096875060825845e-06,
"loss": 0.485,
"step": 1650
},
{
"epoch": 1.68,
"learning_rate": 3.11460473638282e-06,
"loss": 0.4768,
"step": 1660
},
{
"epoch": 1.69,
"learning_rate": 2.925065040686642e-06,
"loss": 0.4635,
"step": 1670
},
{
"epoch": 1.7,
"learning_rate": 2.741116425097995e-06,
"loss": 0.4681,
"step": 1680
},
{
"epoch": 1.71,
"learning_rate": 2.5628054796271063e-06,
"loss": 0.4492,
"step": 1690
},
{
"epoch": 1.72,
"learning_rate": 2.390177366390273e-06,
"loss": 0.4664,
"step": 1700
},
{
"epoch": 1.73,
"learning_rate": 2.22327580817136e-06,
"loss": 0.48,
"step": 1710
},
{
"epoch": 1.74,
"learning_rate": 2.0621430773477947e-06,
"loss": 0.4616,
"step": 1720
},
{
"epoch": 1.75,
"learning_rate": 1.906819985183908e-06,
"loss": 0.4854,
"step": 1730
},
{
"epoch": 1.76,
"learning_rate": 1.7573458714944063e-06,
"loss": 0.4846,
"step": 1740
},
{
"epoch": 1.77,
"learning_rate": 1.6137585946804674e-06,
"loss": 0.4552,
"step": 1750
},
{
"epoch": 1.78,
"learning_rate": 1.4760945221410638e-06,
"loss": 0.4615,
"step": 1760
},
{
"epoch": 1.79,
"learning_rate": 1.3443885210619428e-06,
"loss": 0.4735,
"step": 1770
},
{
"epoch": 1.8,
"learning_rate": 1.2186739495845477e-06,
"loss": 0.4705,
"step": 1780
},
{
"epoch": 1.81,
"learning_rate": 1.0989826483571552e-06,
"loss": 0.4653,
"step": 1790
},
{
"epoch": 1.82,
"learning_rate": 9.85344932470364e-07,
"loss": 0.4632,
"step": 1800
},
{
"epoch": 1.83,
"learning_rate": 8.77789583778979e-07,
"loss": 0.481,
"step": 1810
},
{
"epoch": 1.84,
"learning_rate": 7.763438436122122e-07,
"loss": 0.4773,
"step": 1820
},
{
"epoch": 1.85,
"learning_rate": 6.810334058740736e-07,
"loss": 0.4791,
"step": 1830
},
{
"epoch": 1.86,
"learning_rate": 5.918824105356797e-07,
"loss": 0.4793,
"step": 1840
},
{
"epoch": 1.87,
"learning_rate": 5.08913437521169e-07,
"loss": 0.4647,
"step": 1850
},
{
"epoch": 1.88,
"learning_rate": 4.3214750098869995e-07,
"loss": 0.4765,
"step": 1860
},
{
"epoch": 1.89,
"learning_rate": 3.616040440080432e-07,
"loss": 0.4787,
"step": 1870
},
{
"epoch": 1.9,
"learning_rate": 2.973009336361021e-07,
"loss": 0.4723,
"step": 1880
},
{
"epoch": 1.91,
"learning_rate": 2.392544563915883e-07,
"loss": 0.453,
"step": 1890
},
{
"epoch": 1.93,
"learning_rate": 1.8747931413001795e-07,
"loss": 0.4561,
"step": 1900
},
{
"epoch": 1.94,
"learning_rate": 1.4198862032005488e-07,
"loss": 0.4612,
"step": 1910
},
{
"epoch": 1.95,
"learning_rate": 1.0279389672218365e-07,
"loss": 0.4774,
"step": 1920
},
{
"epoch": 1.96,
"learning_rate": 6.990507047049676e-08,
"loss": 0.4635,
"step": 1930
},
{
"epoch": 1.97,
"learning_rate": 4.3330471558378213e-08,
"loss": 0.4761,
"step": 1940
},
{
"epoch": 1.98,
"learning_rate": 2.3076830728713252e-08,
"loss": 0.4593,
"step": 1950
},
{
"epoch": 1.99,
"learning_rate": 9.149277769132658e-09,
"loss": 0.4736,
"step": 1960
},
{
"epoch": 2.0,
"learning_rate": 1.551340212760377e-09,
"loss": 0.4486,
"step": 1970
},
{
"epoch": 2.0,
"step": 1974,
"total_flos": 3.437359013145084e+19,
"train_loss": 0.5708327819994277,
"train_runtime": 105327.89,
"train_samples_per_second": 4.797,
"train_steps_per_second": 0.019
}
],
"max_steps": 1974,
"num_train_epochs": 2,
"total_flos": 3.437359013145084e+19,
"trial_name": null,
"trial_params": null
}