| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1068, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.8691588785046729e-07, | |
| "loss": 1.862, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.345794392523365e-07, | |
| "loss": 1.7887, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.869158878504673e-06, | |
| "loss": 1.5242, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.8037383177570094e-06, | |
| "loss": 1.2751, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.738317757009346e-06, | |
| "loss": 1.208, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.6728971962616825e-06, | |
| "loss": 1.1479, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.607476635514019e-06, | |
| "loss": 1.1126, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.542056074766355e-06, | |
| "loss": 1.0975, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.476635514018692e-06, | |
| "loss": 1.0781, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.411214953271028e-06, | |
| "loss": 1.0466, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.345794392523365e-06, | |
| "loss": 1.0434, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.02803738317757e-05, | |
| "loss": 1.057, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.1214953271028037e-05, | |
| "loss": 1.0466, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.2149532710280374e-05, | |
| "loss": 1.0452, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.308411214953271e-05, | |
| "loss": 1.0368, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.4018691588785047e-05, | |
| "loss": 1.0308, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.4953271028037384e-05, | |
| "loss": 1.0298, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.588785046728972e-05, | |
| "loss": 1.0334, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.6822429906542056e-05, | |
| "loss": 1.0297, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.7757009345794395e-05, | |
| "loss": 1.0464, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.869158878504673e-05, | |
| "loss": 1.0239, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9626168224299065e-05, | |
| "loss": 1.06, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9999519091998718e-05, | |
| "loss": 1.0427, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9996580377272997e-05, | |
| "loss": 1.0744, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9990970903110117e-05, | |
| "loss": 1.0433, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9982692168178177e-05, | |
| "loss": 1.0239, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.997174638428448e-05, | |
| "loss": 1.0644, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9958136475784596e-05, | |
| "loss": 1.0385, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9941866078801108e-05, | |
| "loss": 1.0404, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9922939540252124e-05, | |
| "loss": 1.0414, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9901361916689937e-05, | |
| "loss": 1.0151, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9877138972950084e-05, | |
| "loss": 1.0413, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.985027718061116e-05, | |
| "loss": 1.0299, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.982078371626582e-05, | |
| "loss": 1.0269, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9788666459603448e-05, | |
| "loss": 1.0227, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.975393399130495e-05, | |
| "loss": 1.0139, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9716595590750264e-05, | |
| "loss": 1.0407, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.967666123353924e-05, | |
| "loss": 1.0224, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9634141588826464e-05, | |
| "loss": 1.0385, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9589048016470816e-05, | |
| "loss": 1.0272, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.954139256400049e-05, | |
| "loss": 1.0278, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.9491187963394283e-05, | |
| "loss": 1.0096, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9438447627680032e-05, | |
| "loss": 1.0518, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9383185647351083e-05, | |
| "loss": 1.0361, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.9325416786601785e-05, | |
| "loss": 1.0256, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.9265156479382968e-05, | |
| "loss": 1.024, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9202420825278497e-05, | |
| "loss": 1.0309, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9137226585204006e-05, | |
| "loss": 1.0322, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9069591176928903e-05, | |
| "loss": 1.0342, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.899953267042293e-05, | |
| "loss": 1.041, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.8927069783028447e-05, | |
| "loss": 1.0405, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.885222187445979e-05, | |
| "loss": 1.0345, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.877500894163096e-05, | |
| "loss": 1.0386, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.8695451613313125e-05, | |
| "loss": 1.0244, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.8613571144623267e-05, | |
| "loss": 1.041, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.8529389411345534e-05, | |
| "loss": 1.0224, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.8442928904086717e-05, | |
| "loss": 1.0206, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.8354212722267516e-05, | |
| "loss": 1.0208, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.8263264567951132e-05, | |
| "loss": 1.0128, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.8170108739510825e-05, | |
| "loss": 1.0328, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.807477012513822e-05, | |
| "loss": 1.0304, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.7977274196193966e-05, | |
| "loss": 1.0057, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.787764700040263e-05, | |
| "loss": 1.0193, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.77759151548936e-05, | |
| "loss": 1.0387, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.7672105839089862e-05, | |
| "loss": 1.0062, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.7566246787446538e-05, | |
| "loss": 0.9987, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.7458366282041154e-05, | |
| "loss": 1.0177, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.7348493145017576e-05, | |
| "loss": 1.0183, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.7236656730885703e-05, | |
| "loss": 1.0134, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.7122886918678887e-05, | |
| "loss": 1.022, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.7007214103971224e-05, | |
| "loss": 1.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.688966919075687e-05, | |
| "loss": 1.0061, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.6770283583193478e-05, | |
| "loss": 1.0263, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.664908917721207e-05, | |
| "loss": 1.0131, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.6526118351995444e-05, | |
| "loss": 1.009, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.6401403961327558e-05, | |
| "loss": 1.0104, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.6274979324816043e-05, | |
| "loss": 1.0029, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.6146878218990302e-05, | |
| "loss": 1.0032, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.6017134868277522e-05, | |
| "loss": 0.9954, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.588578393585904e-05, | |
| "loss": 1.0294, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.5752860514409462e-05, | |
| "loss": 1.0063, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.5618400116721045e-05, | |
| "loss": 1.021, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.5482438666215848e-05, | |
| "loss": 1.0124, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.534501248734817e-05, | |
| "loss": 1.0082, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.5206158295899837e-05, | |
| "loss": 0.9861, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.5065913189170944e-05, | |
| "loss": 1.0259, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.4924314636068675e-05, | |
| "loss": 1.0183, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.4781400467096834e-05, | |
| "loss": 0.9975, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.4637208864248766e-05, | |
| "loss": 1.0088, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.4491778350806392e-05, | |
| "loss": 1.0142, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.434514778104803e-05, | |
| "loss": 1.0055, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.4197356329867813e-05, | |
| "loss": 0.9996, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.4048443482309429e-05, | |
| "loss": 1.0101, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.3898449023017024e-05, | |
| "loss": 1.0033, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.3747413025606027e-05, | |
| "loss": 0.9942, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.3595375841956804e-05, | |
| "loss": 1.0043, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.3442378091433947e-05, | |
| "loss": 0.9997, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.3288460650034111e-05, | |
| "loss": 0.9995, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.3133664639465287e-05, | |
| "loss": 0.9904, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.297803141616042e-05, | |
| "loss": 0.982, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.2821602560228326e-05, | |
| "loss": 1.0035, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.2664419864344843e-05, | |
| "loss": 1.0017, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.2506525322587207e-05, | |
| "loss": 0.9985, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.2347961119214584e-05, | |
| "loss": 0.9967, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.2188769617397839e-05, | |
| "loss": 0.9898, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.2028993347901467e-05, | |
| "loss": 0.9834, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.1868674997720767e-05, | |
| "loss": 0.9836, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.1707857398677277e-05, | |
| "loss": 0.9866, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.1546583515975511e-05, | |
| "loss": 0.9989, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.1384896436724047e-05, | |
| "loss": 0.9999, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.122283935842408e-05, | |
| "loss": 0.9808, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.1060455577428454e-05, | |
| "loss": 0.98, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.0897788477374289e-05, | |
| "loss": 0.9846, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.0734881517592297e-05, | |
| "loss": 0.9847, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.057177822149586e-05, | |
| "loss": 0.9809, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.0408522164953013e-05, | |
| "loss": 0.9947, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.0245156964644381e-05, | |
| "loss": 0.9867, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.0081726266410239e-05, | |
| "loss": 0.981, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.918273733589764e-06, | |
| "loss": 0.9738, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.754843035355624e-06, | |
| "loss": 0.9979, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.59147783504699e-06, | |
| "loss": 0.968, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.428221778504143e-06, | |
| "loss": 0.9837, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.265118482407708e-06, | |
| "loss": 0.9787, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.102211522625714e-06, | |
| "loss": 0.9646, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.939544422571547e-06, | |
| "loss": 0.9826, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.777160641575923e-06, | |
| "loss": 0.9773, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.615103563275958e-06, | |
| "loss": 0.968, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.453416484024492e-06, | |
| "loss": 0.9746, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.292142601322724e-06, | |
| "loss": 0.9642, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.131325002279235e-06, | |
| "loss": 0.9647, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.971006652098535e-06, | |
| "loss": 0.9784, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.811230382602163e-06, | |
| "loss": 0.9615, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.652038880785419e-06, | |
| "loss": 0.9641, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.493474677412795e-06, | |
| "loss": 0.9619, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.335580135655157e-06, | |
| "loss": 0.9677, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.178397439771676e-06, | |
| "loss": 0.9657, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.021968583839585e-06, | |
| "loss": 0.963, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 6.8663353605347185e-06, | |
| "loss": 0.9524, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 6.711539349965892e-06, | |
| "loss": 0.9721, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 6.557621908566058e-06, | |
| "loss": 0.963, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 6.4046241580432e-06, | |
| "loss": 0.9782, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 6.252586974393975e-06, | |
| "loss": 0.9578, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 6.101550976982979e-06, | |
| "loss": 0.9607, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.951556517690573e-06, | |
| "loss": 0.9819, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.802643670132193e-06, | |
| "loss": 0.9656, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.654852218951973e-06, | |
| "loss": 0.9672, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.508221649193609e-06, | |
| "loss": 0.9486, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5.362791135751237e-06, | |
| "loss": 0.9557, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5.21859953290317e-06, | |
| "loss": 0.9569, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 5.0756853639313245e-06, | |
| "loss": 0.9667, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.934086810829057e-06, | |
| "loss": 0.9662, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.793841704100162e-06, | |
| "loss": 0.9721, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.654987512651827e-06, | |
| "loss": 0.9644, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.517561333784154e-06, | |
| "loss": 0.9519, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.381599883278962e-06, | |
| "loss": 0.9523, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.247139485590543e-06, | |
| "loss": 0.9613, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.1142160641409626e-06, | |
| "loss": 0.9515, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.982865131722483e-06, | |
| "loss": 0.9638, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.853121781009703e-06, | |
| "loss": 0.9749, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.725020675183961e-06, | |
| "loss": 0.9509, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.5985960386724438e-06, | |
| "loss": 0.9403, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.4738816480045577e-06, | |
| "loss": 0.9602, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.350910822787937e-06, | |
| "loss": 0.9503, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.2297164168065253e-06, | |
| "loss": 0.9558, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.110330809243134e-06, | |
| "loss": 0.9427, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.9927858960287793e-06, | |
| "loss": 0.941, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.8771130813211147e-06, | |
| "loss": 0.9546, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.7633432691142957e-06, | |
| "loss": 0.9587, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.6515068549824274e-06, | |
| "loss": 0.9659, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.541633717958849e-06, | |
| "loss": 0.9557, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.4337532125534602e-06, | |
| "loss": 0.9352, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.327894160910139e-06, | |
| "loss": 0.9666, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.2240848451064047e-06, | |
| "loss": 0.9426, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.1223529995973747e-06, | |
| "loss": 0.9444, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.0227258038060372e-06, | |
| "loss": 0.9429, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.9252298748617827e-06, | |
| "loss": 0.9449, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.829891260489176e-06, | |
| "loss": 0.9336, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.7367354320488704e-06, | |
| "loss": 0.9292, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.6457872777324856e-06, | |
| "loss": 0.96, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.5570710959132861e-06, | |
| "loss": 0.9504, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.4706105886544674e-06, | |
| "loss": 0.9474, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.3864288553767335e-06, | |
| "loss": 0.9474, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.3045483866868757e-06, | |
| "loss": 0.9581, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.2249910583690394e-06, | |
| "loss": 0.9533, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.147778125540212e-06, | |
| "loss": 0.9414, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.0729302169715538e-06, | |
| "loss": 0.935, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.000467329577074e-06, | |
| "loss": 0.9579, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 9.304088230711006e-07, | |
| "loss": 0.9589, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.627734147959954e-07, | |
| "loss": 0.9434, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.975791747215023e-07, | |
| "loss": 0.9371, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.348435206170357e-07, | |
| "loss": 0.9266, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.745832133982188e-07, | |
| "loss": 0.9509, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 6.168143526489201e-07, | |
| "loss": 0.9363, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.615523723199723e-07, | |
| "loss": 0.9496, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5.088120366057192e-07, | |
| "loss": 0.9388, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.5860743599951186e-07, | |
| "loss": 0.9501, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.1095198352918643e-07, | |
| "loss": 0.9446, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.658584111735375e-07, | |
| "loss": 0.9444, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.2333876646076126e-07, | |
| "loss": 0.9354, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.834044092497379e-07, | |
| "loss": 0.9361, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.460660086950539e-07, | |
| "loss": 0.9314, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.113335403965522e-07, | |
| "loss": 0.9387, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.7921628373418264e-07, | |
| "loss": 0.9472, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.4972281938884337e-07, | |
| "loss": 0.9493, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.228610270499153e-07, | |
| "loss": 0.9297, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 9.863808331006197e-08, | |
| "loss": 0.9439, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 7.706045974787746e-08, | |
| "loss": 0.9522, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5.813392119889405e-08, | |
| "loss": 0.9491, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.1863524215406405e-08, | |
| "loss": 0.9329, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.825361571552554e-08, | |
| "loss": 0.9478, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.7307831821823385e-08, | |
| "loss": 0.9417, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.029096889884293e-09, | |
| "loss": 0.9452, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.4196227270055783e-09, | |
| "loss": 0.9355, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.80908001283531e-10, | |
| "loss": 0.9511, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.9552798271179199, | |
| "eval_runtime": 138.8736, | |
| "eval_samples_per_second": 108.905, | |
| "eval_steps_per_second": 1.707, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1068, | |
| "total_flos": 447234944532480.0, | |
| "train_loss": 0.99920238649354, | |
| "train_runtime": 5476.1677, | |
| "train_samples_per_second": 24.954, | |
| "train_steps_per_second": 0.195 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 1068, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 447234944532480.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |