| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 26325, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.962631077216398e-06, | |
| "loss": 0.1958, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.886367969494757e-06, | |
| "loss": 0.1228, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.810104861773118e-06, | |
| "loss": 0.1051, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.733841754051478e-06, | |
| "loss": 0.0977, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.657578646329838e-06, | |
| "loss": 0.0891, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.581315538608199e-06, | |
| "loss": 0.0827, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.50505243088656e-06, | |
| "loss": 0.0822, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.42878932316492e-06, | |
| "loss": 0.0796, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.352526215443281e-06, | |
| "loss": 0.0777, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.27626310772164e-06, | |
| "loss": 0.0769, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.200000000000002e-06, | |
| "loss": 0.0736, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.123736892278361e-06, | |
| "loss": 0.0721, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.047473784556723e-06, | |
| "loss": 0.0748, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.971210676835082e-06, | |
| "loss": 0.0736, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.894947569113442e-06, | |
| "loss": 0.0714, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.818684461391803e-06, | |
| "loss": 0.0718, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.742421353670163e-06, | |
| "loss": 0.0729, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.666158245948522e-06, | |
| "loss": 0.0711, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.589895138226883e-06, | |
| "loss": 0.0663, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.513632030505243e-06, | |
| "loss": 0.0675, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.437368922783604e-06, | |
| "loss": 0.0689, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.361105815061964e-06, | |
| "loss": 0.0685, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.284842707340325e-06, | |
| "loss": 0.0663, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.208579599618686e-06, | |
| "loss": 0.0651, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.132316491897046e-06, | |
| "loss": 0.0652, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.056053384175407e-06, | |
| "loss": 0.0653, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.979790276453767e-06, | |
| "loss": 0.0656, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.903527168732126e-06, | |
| "loss": 0.0635, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.827264061010486e-06, | |
| "loss": 0.061, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.751000953288847e-06, | |
| "loss": 0.0637, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.674737845567207e-06, | |
| "loss": 0.0621, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.598856053384176e-06, | |
| "loss": 0.0657, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.522592945662537e-06, | |
| "loss": 0.0618, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.446329837940897e-06, | |
| "loss": 0.0617, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.370066730219257e-06, | |
| "loss": 0.0623, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.293803622497617e-06, | |
| "loss": 0.0581, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.217540514775978e-06, | |
| "loss": 0.0612, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.141277407054338e-06, | |
| "loss": 0.0584, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.065014299332699e-06, | |
| "loss": 0.0595, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.9887511916110584e-06, | |
| "loss": 0.0605, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.912488083889419e-06, | |
| "loss": 0.0584, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.836606291706388e-06, | |
| "loss": 0.06, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.760343183984748e-06, | |
| "loss": 0.0589, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.12254729866981506, | |
| "eval_runtime": 5547.0518, | |
| "eval_samples_per_second": 6.063, | |
| "eval_steps_per_second": 0.189, | |
| "eval_wer": 0.06041116269525739, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.684080076263109e-06, | |
| "loss": 0.0547, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.6078169685414686e-06, | |
| "loss": 0.0363, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.531553860819829e-06, | |
| "loss": 0.0357, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.4552907530981886e-06, | |
| "loss": 0.0375, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.37902764537655e-06, | |
| "loss": 0.0347, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.302764537654909e-06, | |
| "loss": 0.0347, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.226501429933271e-06, | |
| "loss": 0.0342, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.15023832221163e-06, | |
| "loss": 0.0358, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.073975214489991e-06, | |
| "loss": 0.0344, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.997712106768351e-06, | |
| "loss": 0.0333, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.9218303145853195e-06, | |
| "loss": 0.0357, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.845567206863681e-06, | |
| "loss": 0.0355, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.76930409914204e-06, | |
| "loss": 0.0348, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.693040991420401e-06, | |
| "loss": 0.0339, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.61677788369876e-06, | |
| "loss": 0.0344, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.540514775977122e-06, | |
| "loss": 0.0345, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 5.464251668255481e-06, | |
| "loss": 0.0329, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.3879885605338424e-06, | |
| "loss": 0.0334, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.311725452812202e-06, | |
| "loss": 0.0355, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.235462345090562e-06, | |
| "loss": 0.0338, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.159199237368924e-06, | |
| "loss": 0.0339, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.082936129647283e-06, | |
| "loss": 0.0357, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 5.0066730219256445e-06, | |
| "loss": 0.0323, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.930409914204004e-06, | |
| "loss": 0.034, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.8541468064823645e-06, | |
| "loss": 0.0333, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.777883698760725e-06, | |
| "loss": 0.034, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.701620591039085e-06, | |
| "loss": 0.033, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.625357483317446e-06, | |
| "loss": 0.0359, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.549475691134414e-06, | |
| "loss": 0.0339, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.473212583412775e-06, | |
| "loss": 0.0343, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.396949475691134e-06, | |
| "loss": 0.0335, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.320686367969495e-06, | |
| "loss": 0.0327, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.244423260247855e-06, | |
| "loss": 0.033, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.1681601525262155e-06, | |
| "loss": 0.0339, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.091897044804577e-06, | |
| "loss": 0.0334, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.015633937082936e-06, | |
| "loss": 0.0306, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 3.939370829361297e-06, | |
| "loss": 0.0315, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.863107721639657e-06, | |
| "loss": 0.0326, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 3.7868446139180175e-06, | |
| "loss": 0.0309, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 3.710581506196378e-06, | |
| "loss": 0.0321, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.6343183984747384e-06, | |
| "loss": 0.033, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 3.5580552907530984e-06, | |
| "loss": 0.0338, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 3.4817921830314588e-06, | |
| "loss": 0.0319, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 3.405529075309819e-06, | |
| "loss": 0.031, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.11860840022563934, | |
| "eval_runtime": 5539.7182, | |
| "eval_samples_per_second": 6.071, | |
| "eval_steps_per_second": 0.19, | |
| "eval_wer": 0.056729379878763, | |
| "step": 17550 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.3292659675881796e-06, | |
| "loss": 0.0288, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 3.2530028598665396e-06, | |
| "loss": 0.0175, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 3.1767397521449e-06, | |
| "loss": 0.0172, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 3.1004766444232604e-06, | |
| "loss": 0.0168, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 3.024213536701621e-06, | |
| "loss": 0.0158, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.9483317445185893e-06, | |
| "loss": 0.0169, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2.8720686367969497e-06, | |
| "loss": 0.0174, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 2.79580552907531e-06, | |
| "loss": 0.0171, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 2.71954242135367e-06, | |
| "loss": 0.0169, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 2.643660629170639e-06, | |
| "loss": 0.0181, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 2.5673975214489994e-06, | |
| "loss": 0.0166, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 2.4911344137273594e-06, | |
| "loss": 0.0159, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 2.41487130600572e-06, | |
| "loss": 0.0174, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.3386081982840803e-06, | |
| "loss": 0.0168, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 2.2623450905624407e-06, | |
| "loss": 0.0167, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 2.1860819828408007e-06, | |
| "loss": 0.0166, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.109818875119161e-06, | |
| "loss": 0.0166, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 2.033555767397522e-06, | |
| "loss": 0.016, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1.957292659675882e-06, | |
| "loss": 0.0162, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 1.8810295519542423e-06, | |
| "loss": 0.0159, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 1.8047664442326028e-06, | |
| "loss": 0.0156, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 1.728503336510963e-06, | |
| "loss": 0.0168, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 1.6522402287893232e-06, | |
| "loss": 0.0167, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 1.5759771210676836e-06, | |
| "loss": 0.0163, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 1.4997140133460438e-06, | |
| "loss": 0.0161, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 1.4234509056244042e-06, | |
| "loss": 0.0162, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 1.3471877979027648e-06, | |
| "loss": 0.015, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 1.270924690181125e-06, | |
| "loss": 0.0157, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 1.1946615824594855e-06, | |
| "loss": 0.0157, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 1.1183984747378457e-06, | |
| "loss": 0.0153, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 1.042135367016206e-06, | |
| "loss": 0.0149, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 9.662535748331745e-07, | |
| "loss": 0.0157, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 8.899904671115349e-07, | |
| "loss": 0.0144, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 8.137273593898952e-07, | |
| "loss": 0.0164, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 7.374642516682555e-07, | |
| "loss": 0.0156, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 6.612011439466159e-07, | |
| "loss": 0.0158, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 5.849380362249762e-07, | |
| "loss": 0.0152, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 5.086749285033365e-07, | |
| "loss": 0.0149, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.324118207816969e-07, | |
| "loss": 0.0155, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 3.5614871306005724e-07, | |
| "loss": 0.0164, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 2.7988560533841755e-07, | |
| "loss": 0.0147, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 2.0362249761677792e-07, | |
| "loss": 0.0143, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 1.2774070543374643e-07, | |
| "loss": 0.0145, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 5.147759771210677e-08, | |
| "loss": 0.0148, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.12545564770698547, | |
| "eval_runtime": 5578.7222, | |
| "eval_samples_per_second": 6.029, | |
| "eval_steps_per_second": 0.188, | |
| "eval_wer": 0.05513910619779243, | |
| "step": 26325 | |
| } | |
| ], | |
| "max_steps": 26325, | |
| "num_train_epochs": 3, | |
| "total_flos": 6.447477588325171e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |