| { | |
| "best_metric": 0.23276936632271397, | |
| "best_model_checkpoint": "esm2_t6_8M_finetune_2023-10-08_00-58-24/checkpoint-42015", | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 42015, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0005701249285950224, | |
| "loss": 0.65, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0005700293047710233, | |
| "loss": 0.5381, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0005698699554923001, | |
| "loss": 0.5093, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0005696469163951857, | |
| "loss": 0.4747, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0005693602373593884, | |
| "loss": 0.4731, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0005690099824968371, | |
| "loss": 0.454, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000568596230137343, | |
| "loss": 0.4384, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0005681190728110825, | |
| "loss": 0.4432, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0005675786172279044, | |
| "loss": 0.4288, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0005669749842534648, | |
| "loss": 0.4278, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000566308308882198, | |
| "loss": 0.4289, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0005655787402071262, | |
| "loss": 0.4171, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0005647864413865174, | |
| "loss": 0.4262, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0005639315896073965, | |
| "loss": 0.4155, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0005630143760459205, | |
| "loss": 0.4042, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000562035005824624, | |
| "loss": 0.4102, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0005609936979665467, | |
| "loss": 0.403, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0005598906853462516, | |
| "loss": 0.3957, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0005587262146377457, | |
| "loss": 0.3939, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005575005462593147, | |
| "loss": 0.4017, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005562139543152841, | |
| "loss": 0.3957, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005548736130124954, | |
| "loss": 0.3983, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005534663515857835, | |
| "loss": 0.3824, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005519990687871066, | |
| "loss": 0.3809, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0005504720927546211, | |
| "loss": 0.3862, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0005488857649760757, | |
| "loss": 0.3723, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0005472404402124425, | |
| "loss": 0.3708, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0005455364864185798, | |
| "loss": 0.372, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0005437742846609435, | |
| "loss": 0.3652, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000541954229032367, | |
| "loss": 0.3534, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0005400767265639273, | |
| "loss": 0.3557, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0005381421971339184, | |
| "loss": 0.3563, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0005361510733739505, | |
| "loss": 0.3641, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0005341141758547804, | |
| "loss": 0.365, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0005320114891528482, | |
| "loss": 0.3507, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.000529853579171728, | |
| "loss": 0.347, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0005276409284991003, | |
| "loss": 0.349, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0005253740319646683, | |
| "loss": 0.3518, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0005230533965294961, | |
| "loss": 0.3329, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0005206795411726335, | |
| "loss": 0.3401, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0005182529967750532, | |
| "loss": 0.3342, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.000515774306000927, | |
| "loss": 0.3409, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0005132440231762652, | |
| "loss": 0.331, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00051066271416495, | |
| "loss": 0.3356, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0005080309562421869, | |
| "loss": 0.3235, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0005053493379654053, | |
| "loss": 0.3292, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0005026184590426352, | |
| "loss": 0.3198, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0004998389301983908, | |
| "loss": 0.3262, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0004970113730370897, | |
| "loss": 0.3239, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0004941364199040399, | |
| "loss": 0.3177, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0004912147137440239, | |
| "loss": 0.3103, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.000488246907957513, | |
| "loss": 0.3169, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00048524884437437616, | |
| "loss": 0.3071, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0004822064618785885, | |
| "loss": 0.3091, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00047910481733056954, | |
| "loss": 0.305, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.000475959781372279, | |
| "loss": 0.3086, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0004727720573489078, | |
| "loss": 0.3034, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00046954235815226125, | |
| "loss": 0.2921, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00046627140606133027, | |
| "loss": 0.3043, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00046295993258076314, | |
| "loss": 0.2905, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00045960867827727475, | |
| "loss": 0.2969, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00045621839261402846, | |
| "loss": 0.2938, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00045278983378302856, | |
| "loss": 0.2934, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0004493237685355611, | |
| "loss": 0.2844, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00044582097201072027, | |
| "loss": 0.2882, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00044228222756205834, | |
| "loss": 0.2936, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00043870832658239985, | |
| "loss": 0.2776, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00043510006832685687, | |
| "loss": 0.2811, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0004314582597340859, | |
| "loss": 0.2821, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00042778371524582784, | |
| "loss": 0.287, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00042407725662476785, | |
| "loss": 0.2789, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00042033971277076005, | |
| "loss": 0.2763, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0004165719195354547, | |
| "loss": 0.2695, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00041277471953537104, | |
| "loss": 0.2635, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0004089489619634578, | |
| "loss": 0.2636, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0004050955023991823, | |
| "loss": 0.2632, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0004012152026171918, | |
| "loss": 0.2542, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.000397328524919439, | |
| "loss": 0.2705, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0003933972771535795, | |
| "loss": 0.2513, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0003894418053214825, | |
| "loss": 0.245, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00038546299401147295, | |
| "loss": 0.2578, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0003814617330314382, | |
| "loss": 0.2577, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0003774389172098332, | |
| "loss": 0.2491, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0003733954461955649, | |
| "loss": 0.2563, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00036933222425679775, | |
| "loss": 0.2552, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00036525016007872573, | |
| "loss": 0.236, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00036115016656035776, | |
| "loss": 0.2482, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00035703316061035936, | |
| "loss": 0.2412, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0003529000629419989, | |
| "loss": 0.2402, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00034877257538830126, | |
| "loss": 0.2383, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00034461013949659567, | |
| "loss": 0.2303, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00034043439012889894, | |
| "loss": 0.2368, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00033624626113565846, | |
| "loss": 0.227, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0003320466891358589, | |
| "loss": 0.2321, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00032783661330756047, | |
| "loss": 0.2307, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00032361697517786374, | |
| "loss": 0.2209, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0003193887184123502, | |
| "loss": 0.2283, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00031515278860404374, | |
| "loss": 0.2231, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00031091013306194153, | |
| "loss": 0.213, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0003066617005991611, | |
| "loss": 0.2227, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0003024084413207512, | |
| "loss": 0.2106, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0002981726001518289, | |
| "loss": 0.2156, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00029391255391089834, | |
| "loss": 0.2185, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0002896505320303282, | |
| "loss": 0.2133, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00028538748765425824, | |
| "loss": 0.2169, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0002811243741554959, | |
| "loss": 0.2082, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00027686214492230696, | |
| "loss": 0.2052, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00027262304895711635, | |
| "loss": 0.2166, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00026836543109541225, | |
| "loss": 0.211, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00026411155086609875, | |
| "loss": 0.2014, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0002598623595925447, | |
| "loss": 0.2029, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0002556188075494962, | |
| "loss": 0.1988, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00025138184375056156, | |
| "loss": 0.2003, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0002471524157359758, | |
| "loss": 0.197, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0002429314693606973, | |
| "loss": 0.1958, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00023871994858287913, | |
| "loss": 0.1848, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00023451879525276522, | |
| "loss": 0.1925, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0002303289489020594, | |
| "loss": 0.1923, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00022615134653381023, | |
| "loss": 0.191, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00022200771021072758, | |
| "loss": 0.1906, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00021785732279544682, | |
| "loss": 0.1905, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0002137219684748498, | |
| "loss": 0.1884, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00020960257206557216, | |
| "loss": 0.1844, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0002055000548154766, | |
| "loss": 0.1865, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00020141533419762732, | |
| "loss": 0.1729, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00019734932370511055, | |
| "loss": 0.1717, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00019330293264674297, | |
| "loss": 0.1797, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0001892770659437178, | |
| "loss": 0.1756, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00018527262392723065, | |
| "loss": 0.184, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00018129050213713306, | |
| "loss": 0.1806, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00017733159112165648, | |
| "loss": 0.175, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00017339677623825358, | |
| "loss": 0.1746, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00016948693745559923, | |
| "loss": 0.1667, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00016560294915679837, | |
| "loss": 0.1737, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00016174567994384052, | |
| "loss": 0.1624, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.00015793507085198864, | |
| "loss": 0.1667, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.00015413367721145552, | |
| "loss": 0.1679, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00015036156760590242, | |
| "loss": 0.1626, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00014661958561714702, | |
| "loss": 0.1614, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00014290856808936924, | |
| "loss": 0.1603, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0001392293449419613, | |
| "loss": 0.1611, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0001355827389839281, | |
| "loss": 0.156, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0001319695657298769, | |
| "loss": 0.1618, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.000128390633217638, | |
| "loss": 0.1653, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00012484674182755833, | |
| "loss": 0.1573, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00012133868410350671, | |
| "loss": 0.1629, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00011786724457563275, | |
| "loss": 0.1531, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00011443319958491592, | |
| "loss": 0.1491, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00011103731710954869, | |
| "loss": 0.1443, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00010768035659318755, | |
| "loss": 0.1454, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00010436306877511392, | |
| "loss": 0.1524, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00010108619552234125, | |
| "loss": 0.1476, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.78665447379777e-05, | |
| "loss": 0.1519, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.467247876055098e-05, | |
| "loss": 0.1421, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.152099451915894e-05, | |
| "loss": 0.1445, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.841279680106426e-05, | |
| "loss": 0.1378, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.534858071307888e-05, | |
| "loss": 0.1484, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.232903152611165e-05, | |
| "loss": 0.1434, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.935482452191686e-05, | |
| "loss": 0.1342, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.642662484207666e-05, | |
| "loss": 0.139, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.354508733925122e-05, | |
| "loss": 0.1411, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.071085643072893e-05, | |
| "loss": 0.1319, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.792456595431225e-05, | |
| "loss": 0.1287, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.518683902656733e-05, | |
| "loss": 0.1273, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.249828790347295e-05, | |
| "loss": 0.129, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 5.9859513843497324e-05, | |
| "loss": 0.1326, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 5.727110697313506e-05, | |
| "loss": 0.1302, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.47336461549329e-05, | |
| "loss": 0.1487, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.224769885803498e-05, | |
| "loss": 0.1282, | |
| "step": 33800 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.981382103127605e-05, | |
| "loss": 0.1278, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.7432556978851023e-05, | |
| "loss": 0.13, | |
| "step": 34200 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.51044392385881e-05, | |
| "loss": 0.1281, | |
| "step": 34400 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.28412263737391e-05, | |
| "loss": 0.1285, | |
| "step": 34600 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.063164624396037e-05, | |
| "loss": 0.1288, | |
| "step": 34800 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.846549409375182e-05, | |
| "loss": 0.1275, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.635449361928196e-05, | |
| "loss": 0.1249, | |
| "step": 35200 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.42991169175479e-05, | |
| "loss": 0.1258, | |
| "step": 35400 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.229982364603461e-05, | |
| "loss": 0.1285, | |
| "step": 35600 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.0357060919918606e-05, | |
| "loss": 0.1303, | |
| "step": 35800 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.8471263212076872e-05, | |
| "loss": 0.1306, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.6642852255922315e-05, | |
| "loss": 0.1253, | |
| "step": 36200 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.487223695108933e-05, | |
| "loss": 0.1175, | |
| "step": 36400 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3159813271988392e-05, | |
| "loss": 0.1273, | |
| "step": 36600 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.151408710314706e-05, | |
| "loss": 0.1316, | |
| "step": 36800 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.9918886837489768e-05, | |
| "loss": 0.121, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.838298594800929e-05, | |
| "loss": 0.1307, | |
| "step": 37200 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.69067279183791e-05, | |
| "loss": 0.1169, | |
| "step": 37400 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.549044289394571e-05, | |
| "loss": 0.1202, | |
| "step": 37600 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.4134447607895461e-05, | |
| "loss": 0.1139, | |
| "step": 37800 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.2839045310422171e-05, | |
| "loss": 0.1289, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.1604525700908673e-05, | |
| "loss": 0.1333, | |
| "step": 38200 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.0431164863139752e-05, | |
| "loss": 0.1165, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.319225203559418e-06, | |
| "loss": 0.121, | |
| "step": 38600 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.268955392587713e-06, | |
| "loss": 0.1265, | |
| "step": 38800 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.2805903090082e-06, | |
| "loss": 0.114, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.3543509874411355e-06, | |
| "loss": 0.1191, | |
| "step": 39200 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.4904445689116835e-06, | |
| "loss": 0.1085, | |
| "step": 39400 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.689064254525586e-06, | |
| "loss": 0.1171, | |
| "step": 39600 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.9503892622624656e-06, | |
| "loss": 0.1127, | |
| "step": 39800 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.2745847868960813e-06, | |
| "loss": 0.1206, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.6618019630506124e-06, | |
| "loss": 0.1167, | |
| "step": 40200 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.1121778314016173e-06, | |
| "loss": 0.116, | |
| "step": 40400 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.6258353080285549e-06, | |
| "loss": 0.1155, | |
| "step": 40600 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.202883156926388e-06, | |
| "loss": 0.1112, | |
| "step": 40800 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.450552439147354e-07, | |
| "loss": 0.124, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.488353993419233e-07, | |
| "loss": 0.1186, | |
| "step": 41200 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.1624678365224445e-07, | |
| "loss": 0.1128, | |
| "step": 41400 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.473414121760922e-07, | |
| "loss": 0.1269, | |
| "step": 41600 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.215705833867654e-08, | |
| "loss": 0.1186, | |
| "step": 41800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 7.172452126860464e-10, | |
| "loss": 0.1156, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9232681404353718, | |
| "eval_auc": 0.7327222270892191, | |
| "eval_f1": 0.23276936632271397, | |
| "eval_loss": 1.4070135354995728, | |
| "eval_mcc": 0.2532717443913893, | |
| "eval_precision": 0.14886006058928025, | |
| "eval_recall": 0.5334820519181167, | |
| "eval_runtime": 1911.9976, | |
| "eval_samples_per_second": 66.763, | |
| "eval_steps_per_second": 5.564, | |
| "step": 42015 | |
| } | |
| ], | |
| "logging_steps": 200, | |
| "max_steps": 42015, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "total_flos": 2.0189159728820676e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |