| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 312870, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.992009460798415e-05, | |
| "loss": 0.0884, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9840189215968296e-05, | |
| "loss": 0.0335, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.976028382395244e-05, | |
| "loss": 0.028, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.968037843193659e-05, | |
| "loss": 0.0262, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.960047303992074e-05, | |
| "loss": 0.024, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.952056764790488e-05, | |
| "loss": 0.0214, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.944066225588903e-05, | |
| "loss": 0.0214, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.936075686387318e-05, | |
| "loss": 0.0216, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9280851471857325e-05, | |
| "loss": 0.0222, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.920094607984147e-05, | |
| "loss": 0.0205, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.912104068782561e-05, | |
| "loss": 0.02, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9041135295809766e-05, | |
| "loss": 0.0214, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.8961229903793907e-05, | |
| "loss": 0.0196, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.888132451177806e-05, | |
| "loss": 0.0191, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.880141911976221e-05, | |
| "loss": 0.0187, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.872151372774635e-05, | |
| "loss": 0.0194, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.86416083357305e-05, | |
| "loss": 0.0192, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.856170294371464e-05, | |
| "loss": 0.0168, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.8481797551698795e-05, | |
| "loss": 0.0186, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.8401892159682935e-05, | |
| "loss": 0.0191, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.832198676766708e-05, | |
| "loss": 0.0182, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.824208137565123e-05, | |
| "loss": 0.0152, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8162175983635376e-05, | |
| "loss": 0.0176, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.808227059161953e-05, | |
| "loss": 0.0176, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.800236519960367e-05, | |
| "loss": 0.0158, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.792245980758782e-05, | |
| "loss": 0.016, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7842554415571964e-05, | |
| "loss": 0.0174, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.776264902355611e-05, | |
| "loss": 0.0187, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.768274363154026e-05, | |
| "loss": 0.0172, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.7602838239524405e-05, | |
| "loss": 0.0188, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.752293284750855e-05, | |
| "loss": 0.0154, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.74430274554927e-05, | |
| "loss": 0.0175, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7363122063476846e-05, | |
| "loss": 0.0158, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.728321667146099e-05, | |
| "loss": 0.0151, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.720331127944514e-05, | |
| "loss": 0.017, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.712340588742928e-05, | |
| "loss": 0.0167, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.7043500495413434e-05, | |
| "loss": 0.0142, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.696359510339758e-05, | |
| "loss": 0.0164, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.688368971138173e-05, | |
| "loss": 0.0175, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6803784319365875e-05, | |
| "loss": 0.0152, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.6723878927350016e-05, | |
| "loss": 0.0175, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.664397353533417e-05, | |
| "loss": 0.0178, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.656406814331831e-05, | |
| "loss": 0.017, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.648416275130246e-05, | |
| "loss": 0.017, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.640425735928661e-05, | |
| "loss": 0.0151, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.632435196727075e-05, | |
| "loss": 0.0159, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6244446575254904e-05, | |
| "loss": 0.0163, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6164541183239044e-05, | |
| "loss": 0.0168, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.60846357912232e-05, | |
| "loss": 0.0163, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.600473039920734e-05, | |
| "loss": 0.0158, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.5924825007191485e-05, | |
| "loss": 0.016, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.584491961517563e-05, | |
| "loss": 0.0165, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.576501422315978e-05, | |
| "loss": 0.0136, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.568510883114393e-05, | |
| "loss": 0.0132, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5605203439128073e-05, | |
| "loss": 0.0154, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.552529804711222e-05, | |
| "loss": 0.0143, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.544539265509637e-05, | |
| "loss": 0.0155, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5365487263080514e-05, | |
| "loss": 0.0157, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.528558187106466e-05, | |
| "loss": 0.0155, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.520567647904881e-05, | |
| "loss": 0.0169, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5125771087032955e-05, | |
| "loss": 0.0144, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.50458656950171e-05, | |
| "loss": 0.0151, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.496596030300125e-05, | |
| "loss": 0.0152, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4886054910985396e-05, | |
| "loss": 0.0149, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.480614951896954e-05, | |
| "loss": 0.0172, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.472624412695369e-05, | |
| "loss": 0.015, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.464633873493784e-05, | |
| "loss": 0.0155, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.4566433342921984e-05, | |
| "loss": 0.0151, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.448652795090613e-05, | |
| "loss": 0.0155, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.440662255889028e-05, | |
| "loss": 0.0141, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.432671716687442e-05, | |
| "loss": 0.0156, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.424681177485857e-05, | |
| "loss": 0.0163, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.416690638284271e-05, | |
| "loss": 0.0148, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.4087000990826866e-05, | |
| "loss": 0.0151, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.400709559881101e-05, | |
| "loss": 0.0149, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.3927190206795153e-05, | |
| "loss": 0.0152, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.384728481477931e-05, | |
| "loss": 0.0152, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.376737942276345e-05, | |
| "loss": 0.0143, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.36874740307476e-05, | |
| "loss": 0.0155, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.360756863873174e-05, | |
| "loss": 0.0147, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.352766324671589e-05, | |
| "loss": 0.0139, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.3447757854700035e-05, | |
| "loss": 0.0147, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.336785246268418e-05, | |
| "loss": 0.0145, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.3287947070668336e-05, | |
| "loss": 0.015, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.3208041678652476e-05, | |
| "loss": 0.0139, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.312813628663662e-05, | |
| "loss": 0.016, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.304823089462077e-05, | |
| "loss": 0.0141, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.296832550260492e-05, | |
| "loss": 0.0155, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.2888420110589064e-05, | |
| "loss": 0.0167, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.280851471857321e-05, | |
| "loss": 0.0137, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.272860932655736e-05, | |
| "loss": 0.0136, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.2648703934541505e-05, | |
| "loss": 0.0144, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.256879854252565e-05, | |
| "loss": 0.0138, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.24888931505098e-05, | |
| "loss": 0.0129, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.2408987758493946e-05, | |
| "loss": 0.0137, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.232908236647809e-05, | |
| "loss": 0.0147, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.224917697446224e-05, | |
| "loss": 0.0121, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.216927158244639e-05, | |
| "loss": 0.015, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.2089366190430534e-05, | |
| "loss": 0.0146, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.200946079841468e-05, | |
| "loss": 0.0152, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.192955540639882e-05, | |
| "loss": 0.0145, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.1849650014382975e-05, | |
| "loss": 0.0151, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.1769744622367115e-05, | |
| "loss": 0.0132, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.168983923035127e-05, | |
| "loss": 0.0136, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.1609933838335416e-05, | |
| "loss": 0.0143, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.1530028446319556e-05, | |
| "loss": 0.015, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.145012305430371e-05, | |
| "loss": 0.014, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.137021766228785e-05, | |
| "loss": 0.0133, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.1290312270272004e-05, | |
| "loss": 0.0136, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.1210406878256144e-05, | |
| "loss": 0.0147, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.113050148624029e-05, | |
| "loss": 0.0144, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.105059609422444e-05, | |
| "loss": 0.0138, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.0970690702208585e-05, | |
| "loss": 0.0143, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.089078531019274e-05, | |
| "loss": 0.014, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.081087991817688e-05, | |
| "loss": 0.0123, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.0730974526161026e-05, | |
| "loss": 0.0128, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.065106913414517e-05, | |
| "loss": 0.0145, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.057116374212932e-05, | |
| "loss": 0.0143, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.049125835011347e-05, | |
| "loss": 0.0136, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.0411352958097614e-05, | |
| "loss": 0.014, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.033144756608176e-05, | |
| "loss": 0.0126, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.025154217406591e-05, | |
| "loss": 0.0138, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.0171636782050055e-05, | |
| "loss": 0.0144, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.00917313900342e-05, | |
| "loss": 0.0126, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.001182599801835e-05, | |
| "loss": 0.0144, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.9931920606002496e-05, | |
| "loss": 0.013, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.985201521398664e-05, | |
| "loss": 0.0116, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.977210982197079e-05, | |
| "loss": 0.0136, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.969220442995494e-05, | |
| "loss": 0.0139, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.9612299037939084e-05, | |
| "loss": 0.0122, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.9532393645923224e-05, | |
| "loss": 0.0125, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.945248825390738e-05, | |
| "loss": 0.0122, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.937258286189152e-05, | |
| "loss": 0.0135, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.929267746987567e-05, | |
| "loss": 0.0125, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.921277207785982e-05, | |
| "loss": 0.0135, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.913286668584396e-05, | |
| "loss": 0.0139, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.905296129382811e-05, | |
| "loss": 0.0143, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.8973055901812253e-05, | |
| "loss": 0.0128, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.889315050979641e-05, | |
| "loss": 0.0124, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.881324511778055e-05, | |
| "loss": 0.0135, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.8733339725764694e-05, | |
| "loss": 0.0136, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.865343433374884e-05, | |
| "loss": 0.0141, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.857352894173299e-05, | |
| "loss": 0.013, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.849362354971714e-05, | |
| "loss": 0.0127, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.841371815770128e-05, | |
| "loss": 0.0115, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.833381276568543e-05, | |
| "loss": 0.0121, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.8253907373669576e-05, | |
| "loss": 0.0133, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.817400198165372e-05, | |
| "loss": 0.013, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.809409658963787e-05, | |
| "loss": 0.013, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.801419119762202e-05, | |
| "loss": 0.0136, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.7934285805606164e-05, | |
| "loss": 0.012, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.785438041359031e-05, | |
| "loss": 0.0141, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.777447502157446e-05, | |
| "loss": 0.0131, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.7694569629558605e-05, | |
| "loss": 0.0125, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.761466423754275e-05, | |
| "loss": 0.013, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.75347588455269e-05, | |
| "loss": 0.0113, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.7454853453511046e-05, | |
| "loss": 0.0127, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.737494806149519e-05, | |
| "loss": 0.0127, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.729504266947934e-05, | |
| "loss": 0.0133, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.721513727746349e-05, | |
| "loss": 0.0131, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.713523188544763e-05, | |
| "loss": 0.0118, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.705532649343178e-05, | |
| "loss": 0.0133, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.697542110141592e-05, | |
| "loss": 0.0135, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.6895515709400075e-05, | |
| "loss": 0.0128, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.681561031738422e-05, | |
| "loss": 0.0133, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.673570492536836e-05, | |
| "loss": 0.0124, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.6655799533352516e-05, | |
| "loss": 0.0122, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.6575894141336656e-05, | |
| "loss": 0.0129, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.649598874932081e-05, | |
| "loss": 0.0129, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.641608335730495e-05, | |
| "loss": 0.0132, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.63361779652891e-05, | |
| "loss": 0.0124, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.625627257327325e-05, | |
| "loss": 0.0138, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.617636718125739e-05, | |
| "loss": 0.0134, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.6096461789241545e-05, | |
| "loss": 0.012, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.6016556397225685e-05, | |
| "loss": 0.012, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.593665100520983e-05, | |
| "loss": 0.0123, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.585674561319398e-05, | |
| "loss": 0.014, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.5776840221178126e-05, | |
| "loss": 0.0124, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.569693482916227e-05, | |
| "loss": 0.0131, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.561702943714642e-05, | |
| "loss": 0.0126, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.553712404513057e-05, | |
| "loss": 0.0129, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.5457218653114714e-05, | |
| "loss": 0.0129, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.537731326109886e-05, | |
| "loss": 0.0133, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.529740786908301e-05, | |
| "loss": 0.013, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.5217502477067155e-05, | |
| "loss": 0.0128, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.51375970850513e-05, | |
| "loss": 0.0124, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.505769169303545e-05, | |
| "loss": 0.0115, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.4977786301019596e-05, | |
| "loss": 0.0118, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.489788090900374e-05, | |
| "loss": 0.0118, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.481797551698789e-05, | |
| "loss": 0.0116, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.473807012497203e-05, | |
| "loss": 0.012, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.4658164732956184e-05, | |
| "loss": 0.0127, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.4578259340940324e-05, | |
| "loss": 0.0126, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.449835394892448e-05, | |
| "loss": 0.0124, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.4418448556908625e-05, | |
| "loss": 0.0115, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.4338543164892765e-05, | |
| "loss": 0.0131, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.425863777287692e-05, | |
| "loss": 0.0129, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.417873238086106e-05, | |
| "loss": 0.0121, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.409882698884521e-05, | |
| "loss": 0.0124, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.401892159682935e-05, | |
| "loss": 0.0136, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.39390162048135e-05, | |
| "loss": 0.0127, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.3859110812797654e-05, | |
| "loss": 0.0122, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.3779205420781794e-05, | |
| "loss": 0.0119, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.369930002876595e-05, | |
| "loss": 0.0128, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.361939463675009e-05, | |
| "loss": 0.0125, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.3539489244734235e-05, | |
| "loss": 0.0112, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.345958385271838e-05, | |
| "loss": 0.012, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.337967846070253e-05, | |
| "loss": 0.0125, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.3299773068686676e-05, | |
| "loss": 0.012, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.321986767667082e-05, | |
| "loss": 0.0115, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.313996228465497e-05, | |
| "loss": 0.0124, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.306005689263912e-05, | |
| "loss": 0.01, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.2980151500623264e-05, | |
| "loss": 0.0108, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.2900246108607404e-05, | |
| "loss": 0.0097, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.282034071659156e-05, | |
| "loss": 0.0114, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.2740435324575705e-05, | |
| "loss": 0.011, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.266052993255985e-05, | |
| "loss": 0.0105, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.2580624540544e-05, | |
| "loss": 0.0108, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.250071914852814e-05, | |
| "loss": 0.0116, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.242081375651229e-05, | |
| "loss": 0.0102, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.234090836449643e-05, | |
| "loss": 0.0113, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.226100297248059e-05, | |
| "loss": 0.0095, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.2181097580464734e-05, | |
| "loss": 0.0117, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.2101192188448874e-05, | |
| "loss": 0.0094, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.202128679643303e-05, | |
| "loss": 0.0113, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.194138140441717e-05, | |
| "loss": 0.0113, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.186147601240132e-05, | |
| "loss": 0.0115, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.178157062038546e-05, | |
| "loss": 0.0104, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.170166522836961e-05, | |
| "loss": 0.011, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.1621759836353756e-05, | |
| "loss": 0.0102, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.15418544443379e-05, | |
| "loss": 0.0107, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.146194905232206e-05, | |
| "loss": 0.0108, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.13820436603062e-05, | |
| "loss": 0.0128, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.1302138268290344e-05, | |
| "loss": 0.0114, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.122223287627449e-05, | |
| "loss": 0.0092, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.114232748425864e-05, | |
| "loss": 0.0113, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.1062422092242785e-05, | |
| "loss": 0.0127, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.098251670022693e-05, | |
| "loss": 0.0112, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.090261130821108e-05, | |
| "loss": 0.0099, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.0822705916195226e-05, | |
| "loss": 0.0086, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.074280052417937e-05, | |
| "loss": 0.0109, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.066289513216352e-05, | |
| "loss": 0.011, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.058298974014767e-05, | |
| "loss": 0.0108, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.050308434813181e-05, | |
| "loss": 0.0105, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.042317895611596e-05, | |
| "loss": 0.0118, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.0343273564100105e-05, | |
| "loss": 0.0098, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.0263368172084255e-05, | |
| "loss": 0.0107, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.01834627800684e-05, | |
| "loss": 0.0101, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.0103557388052546e-05, | |
| "loss": 0.0103, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.0023651996036696e-05, | |
| "loss": 0.0109, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 2.994374660402084e-05, | |
| "loss": 0.0109, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.986384121200499e-05, | |
| "loss": 0.0105, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.9783935819989134e-05, | |
| "loss": 0.0111, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9704030427973277e-05, | |
| "loss": 0.0123, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9624125035957428e-05, | |
| "loss": 0.0118, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.9544219643941575e-05, | |
| "loss": 0.0106, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.9464314251925722e-05, | |
| "loss": 0.0099, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.938440885990987e-05, | |
| "loss": 0.0122, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9304503467894012e-05, | |
| "loss": 0.0105, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.9224598075878163e-05, | |
| "loss": 0.0107, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.9144692683862306e-05, | |
| "loss": 0.011, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.9064787291846457e-05, | |
| "loss": 0.0112, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.89848818998306e-05, | |
| "loss": 0.0093, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.8904976507814747e-05, | |
| "loss": 0.011, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.8825071115798898e-05, | |
| "loss": 0.0099, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.874516572378304e-05, | |
| "loss": 0.0097, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.866526033176719e-05, | |
| "loss": 0.0094, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.8585354939751335e-05, | |
| "loss": 0.0109, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.850544954773548e-05, | |
| "loss": 0.0112, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.842554415571963e-05, | |
| "loss": 0.0104, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8345638763703776e-05, | |
| "loss": 0.0097, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8265733371687923e-05, | |
| "loss": 0.0103, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.818582797967207e-05, | |
| "loss": 0.0102, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.8105922587656214e-05, | |
| "loss": 0.0102, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.8026017195640364e-05, | |
| "loss": 0.0091, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.7946111803624508e-05, | |
| "loss": 0.0109, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7866206411608658e-05, | |
| "loss": 0.0109, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7786301019592802e-05, | |
| "loss": 0.01, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.770639562757695e-05, | |
| "loss": 0.0102, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.76264902355611e-05, | |
| "loss": 0.009, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.7546584843545243e-05, | |
| "loss": 0.0104, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.7466679451529393e-05, | |
| "loss": 0.0114, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.7386774059513537e-05, | |
| "loss": 0.0107, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.730686866749768e-05, | |
| "loss": 0.0115, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.722696327548183e-05, | |
| "loss": 0.0109, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7147057883465978e-05, | |
| "loss": 0.0108, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.7067152491450125e-05, | |
| "loss": 0.0112, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.698724709943427e-05, | |
| "loss": 0.0103, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.6907341707418415e-05, | |
| "loss": 0.0103, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.6827436315402566e-05, | |
| "loss": 0.0101, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.674753092338671e-05, | |
| "loss": 0.0101, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.666762553137086e-05, | |
| "loss": 0.0093, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.6587720139355003e-05, | |
| "loss": 0.0114, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.650781474733915e-05, | |
| "loss": 0.0108, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.64279093553233e-05, | |
| "loss": 0.0102, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.6348003963307444e-05, | |
| "loss": 0.0109, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.6268098571291595e-05, | |
| "loss": 0.0091, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.6188193179275738e-05, | |
| "loss": 0.011, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.6108287787259882e-05, | |
| "loss": 0.0107, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.6028382395244032e-05, | |
| "loss": 0.0104, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.594847700322818e-05, | |
| "loss": 0.0106, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.5868571611212326e-05, | |
| "loss": 0.0113, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.5788666219196473e-05, | |
| "loss": 0.0107, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.5708760827180617e-05, | |
| "loss": 0.0113, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.5628855435164767e-05, | |
| "loss": 0.0107, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.554895004314891e-05, | |
| "loss": 0.0094, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.546904465113306e-05, | |
| "loss": 0.0106, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.5389139259117205e-05, | |
| "loss": 0.009, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.5309233867101352e-05, | |
| "loss": 0.0101, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.5229328475085502e-05, | |
| "loss": 0.01, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.5149423083069646e-05, | |
| "loss": 0.0112, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.5069517691053796e-05, | |
| "loss": 0.0101, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.498961229903794e-05, | |
| "loss": 0.0102, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.4909706907022087e-05, | |
| "loss": 0.0094, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.4829801515006234e-05, | |
| "loss": 0.012, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.474989612299038e-05, | |
| "loss": 0.0096, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.4669990730974528e-05, | |
| "loss": 0.0116, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.4590085338958675e-05, | |
| "loss": 0.0114, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.451017994694282e-05, | |
| "loss": 0.0105, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.443027455492697e-05, | |
| "loss": 0.0113, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.4350369162911112e-05, | |
| "loss": 0.0111, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.427046377089526e-05, | |
| "loss": 0.0103, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.4190558378879406e-05, | |
| "loss": 0.0114, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.4110652986863557e-05, | |
| "loss": 0.0093, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.4030747594847704e-05, | |
| "loss": 0.0099, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.3950842202831847e-05, | |
| "loss": 0.0111, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.3870936810815994e-05, | |
| "loss": 0.0105, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.379103141880014e-05, | |
| "loss": 0.0106, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.3711126026784288e-05, | |
| "loss": 0.01, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.3631220634768435e-05, | |
| "loss": 0.0106, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.3551315242752582e-05, | |
| "loss": 0.0112, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.347140985073673e-05, | |
| "loss": 0.0096, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.3391504458720876e-05, | |
| "loss": 0.0101, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.3311599066705023e-05, | |
| "loss": 0.0097, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.323169367468917e-05, | |
| "loss": 0.0098, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.3151788282673314e-05, | |
| "loss": 0.0115, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.307188289065746e-05, | |
| "loss": 0.0098, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.2991977498641608e-05, | |
| "loss": 0.0101, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2912072106625758e-05, | |
| "loss": 0.0116, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2832166714609905e-05, | |
| "loss": 0.0104, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.275226132259405e-05, | |
| "loss": 0.0095, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.2672355930578196e-05, | |
| "loss": 0.0107, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.2592450538562343e-05, | |
| "loss": 0.0095, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.251254514654649e-05, | |
| "loss": 0.0113, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.2432639754530637e-05, | |
| "loss": 0.0108, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.2352734362514784e-05, | |
| "loss": 0.0108, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.227282897049893e-05, | |
| "loss": 0.0114, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.2192923578483078e-05, | |
| "loss": 0.0104, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.2113018186467225e-05, | |
| "loss": 0.0107, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.203311279445137e-05, | |
| "loss": 0.0102, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.1953207402435515e-05, | |
| "loss": 0.0091, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.1873302010419662e-05, | |
| "loss": 0.0106, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.179339661840381e-05, | |
| "loss": 0.0105, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.171349122638796e-05, | |
| "loss": 0.0097, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.1633585834372107e-05, | |
| "loss": 0.0109, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.155368044235625e-05, | |
| "loss": 0.0095, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1473775050340397e-05, | |
| "loss": 0.0098, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.1393869658324544e-05, | |
| "loss": 0.0111, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.131396426630869e-05, | |
| "loss": 0.0105, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.1234058874292838e-05, | |
| "loss": 0.0101, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.1154153482276985e-05, | |
| "loss": 0.0097, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.1074248090261132e-05, | |
| "loss": 0.0095, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.099434269824528e-05, | |
| "loss": 0.0105, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0914437306229426e-05, | |
| "loss": 0.0118, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0834531914213573e-05, | |
| "loss": 0.0102, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0754626522197717e-05, | |
| "loss": 0.0102, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.0674721130181864e-05, | |
| "loss": 0.0117, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.059481573816601e-05, | |
| "loss": 0.0098, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.051491034615016e-05, | |
| "loss": 0.0104, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.0435004954134308e-05, | |
| "loss": 0.0112, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.035509956211845e-05, | |
| "loss": 0.0093, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.02751941701026e-05, | |
| "loss": 0.0103, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0195288778086746e-05, | |
| "loss": 0.011, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0115383386070893e-05, | |
| "loss": 0.0106, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.003547799405504e-05, | |
| "loss": 0.0104, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.9955572602039187e-05, | |
| "loss": 0.0105, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9875667210023334e-05, | |
| "loss": 0.0097, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.979576181800748e-05, | |
| "loss": 0.01, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.9715856425991628e-05, | |
| "loss": 0.0084, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.9635951033975775e-05, | |
| "loss": 0.0103, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.9556045641959918e-05, | |
| "loss": 0.01, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.9476140249944065e-05, | |
| "loss": 0.0099, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.9396234857928212e-05, | |
| "loss": 0.0102, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.9316329465912363e-05, | |
| "loss": 0.0104, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.923642407389651e-05, | |
| "loss": 0.0097, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.9156518681880653e-05, | |
| "loss": 0.0119, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.90766132898648e-05, | |
| "loss": 0.0103, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.8996707897848947e-05, | |
| "loss": 0.0102, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.8916802505833094e-05, | |
| "loss": 0.0112, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.883689711381724e-05, | |
| "loss": 0.0096, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.8756991721801388e-05, | |
| "loss": 0.01, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8677086329785535e-05, | |
| "loss": 0.0105, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8597180937769682e-05, | |
| "loss": 0.0104, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.851727554575383e-05, | |
| "loss": 0.0118, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.8437370153737976e-05, | |
| "loss": 0.0105, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.835746476172212e-05, | |
| "loss": 0.0102, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.8277559369706267e-05, | |
| "loss": 0.0089, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.8197653977690414e-05, | |
| "loss": 0.0103, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.8117748585674564e-05, | |
| "loss": 0.0095, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.803784319365871e-05, | |
| "loss": 0.0102, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.7957937801642855e-05, | |
| "loss": 0.0109, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.7878032409627e-05, | |
| "loss": 0.0107, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.779812701761115e-05, | |
| "loss": 0.0106, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.7718221625595296e-05, | |
| "loss": 0.0098, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.7638316233579443e-05, | |
| "loss": 0.01, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.755841084156359e-05, | |
| "loss": 0.0096, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.7478505449547737e-05, | |
| "loss": 0.0099, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.7398600057531884e-05, | |
| "loss": 0.0101, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.731869466551603e-05, | |
| "loss": 0.009, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.7238789273500178e-05, | |
| "loss": 0.0108, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.715888388148432e-05, | |
| "loss": 0.0112, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.7078978489468468e-05, | |
| "loss": 0.0096, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.699907309745262e-05, | |
| "loss": 0.0098, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.6919167705436766e-05, | |
| "loss": 0.0113, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.6839262313420913e-05, | |
| "loss": 0.0095, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.6759356921405056e-05, | |
| "loss": 0.0099, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.6679451529389203e-05, | |
| "loss": 0.0099, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.659954613737335e-05, | |
| "loss": 0.0089, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.6519640745357497e-05, | |
| "loss": 0.009, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.6439735353341644e-05, | |
| "loss": 0.0081, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.635982996132579e-05, | |
| "loss": 0.008, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.6279924569309938e-05, | |
| "loss": 0.0081, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.6200019177294085e-05, | |
| "loss": 0.0089, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.6120113785278232e-05, | |
| "loss": 0.0086, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.604020839326238e-05, | |
| "loss": 0.0086, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.5960303001246523e-05, | |
| "loss": 0.0081, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.588039760923067e-05, | |
| "loss": 0.0076, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.580049221721482e-05, | |
| "loss": 0.0085, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.5720586825198967e-05, | |
| "loss": 0.0078, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.5640681433183114e-05, | |
| "loss": 0.0094, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.5560776041167258e-05, | |
| "loss": 0.0087, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.5480870649151405e-05, | |
| "loss": 0.009, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.540096525713555e-05, | |
| "loss": 0.0084, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.53210598651197e-05, | |
| "loss": 0.0077, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5241154473103847e-05, | |
| "loss": 0.009, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5161249081087991e-05, | |
| "loss": 0.0082, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.508134368907214e-05, | |
| "loss": 0.0087, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.5001438297056287e-05, | |
| "loss": 0.0086, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.4921532905040434e-05, | |
| "loss": 0.0072, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.484162751302458e-05, | |
| "loss": 0.008, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4761722121008726e-05, | |
| "loss": 0.0079, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4681816728992873e-05, | |
| "loss": 0.0088, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.460191133697702e-05, | |
| "loss": 0.0067, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.4522005944961167e-05, | |
| "loss": 0.0085, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.4442100552945314e-05, | |
| "loss": 0.0083, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.4362195160929459e-05, | |
| "loss": 0.009, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.4282289768913606e-05, | |
| "loss": 0.0086, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.4202384376897753e-05, | |
| "loss": 0.0085, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.4122478984881902e-05, | |
| "loss": 0.0081, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.4042573592866049e-05, | |
| "loss": 0.0094, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.3962668200850192e-05, | |
| "loss": 0.0088, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.3882762808834341e-05, | |
| "loss": 0.008, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.3802857416818488e-05, | |
| "loss": 0.0083, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.3722952024802635e-05, | |
| "loss": 0.007, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.3643046632786782e-05, | |
| "loss": 0.0084, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.3563141240770927e-05, | |
| "loss": 0.0077, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.3483235848755074e-05, | |
| "loss": 0.0088, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.3403330456739221e-05, | |
| "loss": 0.0089, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.3323425064723368e-05, | |
| "loss": 0.0086, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3243519672707515e-05, | |
| "loss": 0.008, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.316361428069166e-05, | |
| "loss": 0.0094, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3083708888675808e-05, | |
| "loss": 0.0081, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.3003803496659955e-05, | |
| "loss": 0.0084, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.2923898104644103e-05, | |
| "loss": 0.0081, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.284399271262825e-05, | |
| "loss": 0.0088, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.2764087320612394e-05, | |
| "loss": 0.0085, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.2684181928596543e-05, | |
| "loss": 0.0088, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.260427653658069e-05, | |
| "loss": 0.0087, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.2524371144564837e-05, | |
| "loss": 0.0083, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.2444465752548982e-05, | |
| "loss": 0.0085, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.236456036053313e-05, | |
| "loss": 0.0085, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.2284654968517276e-05, | |
| "loss": 0.0076, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.2204749576501423e-05, | |
| "loss": 0.0081, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.212484418448557e-05, | |
| "loss": 0.008, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.2044938792469717e-05, | |
| "loss": 0.0088, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.1965033400453864e-05, | |
| "loss": 0.0083, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.1885128008438009e-05, | |
| "loss": 0.0089, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.1805222616422156e-05, | |
| "loss": 0.0083, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.1725317224406303e-05, | |
| "loss": 0.0089, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.164541183239045e-05, | |
| "loss": 0.0078, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.1565506440374597e-05, | |
| "loss": 0.0087, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.1485601048358744e-05, | |
| "loss": 0.0081, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1405695656342891e-05, | |
| "loss": 0.0092, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1325790264327036e-05, | |
| "loss": 0.0088, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.1245884872311183e-05, | |
| "loss": 0.0079, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.1165979480295332e-05, | |
| "loss": 0.0085, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.1086074088279477e-05, | |
| "loss": 0.0089, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.1006168696263624e-05, | |
| "loss": 0.0092, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.0926263304247771e-05, | |
| "loss": 0.0089, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.0846357912231918e-05, | |
| "loss": 0.0085, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.0766452520216065e-05, | |
| "loss": 0.0074, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.068654712820021e-05, | |
| "loss": 0.0084, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0606641736184358e-05, | |
| "loss": 0.0084, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.0526736344168505e-05, | |
| "loss": 0.0076, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.0446830952152652e-05, | |
| "loss": 0.0088, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0366925560136799e-05, | |
| "loss": 0.0077, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0287020168120946e-05, | |
| "loss": 0.0082, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0207114776105093e-05, | |
| "loss": 0.0094, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0127209384089238e-05, | |
| "loss": 0.0083, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.0047303992073385e-05, | |
| "loss": 0.0084, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 9.967398600057534e-06, | |
| "loss": 0.0082, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.887493208041679e-06, | |
| "loss": 0.0075, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.807587816025826e-06, | |
| "loss": 0.0079, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.727682424009973e-06, | |
| "loss": 0.0084, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.64777703199412e-06, | |
| "loss": 0.009, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.567871639978267e-06, | |
| "loss": 0.0071, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.487966247962412e-06, | |
| "loss": 0.0086, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.40806085594656e-06, | |
| "loss": 0.0097, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.328155463930706e-06, | |
| "loss": 0.0088, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.248250071914853e-06, | |
| "loss": 0.0081, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.168344679899e-06, | |
| "loss": 0.0082, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.088439287883147e-06, | |
| "loss": 0.0086, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 9.008533895867294e-06, | |
| "loss": 0.0077, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 8.92862850385144e-06, | |
| "loss": 0.0083, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.848723111835586e-06, | |
| "loss": 0.0083, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.768817719819735e-06, | |
| "loss": 0.0084, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 8.68891232780388e-06, | |
| "loss": 0.009, | |
| "step": 258500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 8.609006935788027e-06, | |
| "loss": 0.0072, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.529101543772174e-06, | |
| "loss": 0.0081, | |
| "step": 259500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.449196151756321e-06, | |
| "loss": 0.007, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.369290759740468e-06, | |
| "loss": 0.0087, | |
| "step": 260500 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.289385367724614e-06, | |
| "loss": 0.0086, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.209479975708762e-06, | |
| "loss": 0.0085, | |
| "step": 261500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.129574583692908e-06, | |
| "loss": 0.0081, | |
| "step": 262000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 8.049669191677055e-06, | |
| "loss": 0.0096, | |
| "step": 262500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 7.969763799661202e-06, | |
| "loss": 0.0076, | |
| "step": 263000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.889858407645349e-06, | |
| "loss": 0.0074, | |
| "step": 263500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.809953015629496e-06, | |
| "loss": 0.0083, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.73004762361364e-06, | |
| "loss": 0.0067, | |
| "step": 264500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.650142231597788e-06, | |
| "loss": 0.0084, | |
| "step": 265000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.570236839581936e-06, | |
| "loss": 0.0078, | |
| "step": 265500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.490331447566082e-06, | |
| "loss": 0.0087, | |
| "step": 266000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.410426055550229e-06, | |
| "loss": 0.0095, | |
| "step": 266500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.330520663534375e-06, | |
| "loss": 0.0076, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.250615271518523e-06, | |
| "loss": 0.0095, | |
| "step": 267500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.17070987950267e-06, | |
| "loss": 0.008, | |
| "step": 268000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.090804487486816e-06, | |
| "loss": 0.0083, | |
| "step": 268500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.010899095470963e-06, | |
| "loss": 0.0076, | |
| "step": 269000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 6.930993703455109e-06, | |
| "loss": 0.0087, | |
| "step": 269500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 6.851088311439256e-06, | |
| "loss": 0.0087, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 6.771182919423404e-06, | |
| "loss": 0.0092, | |
| "step": 270500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.691277527407549e-06, | |
| "loss": 0.0078, | |
| "step": 271000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.611372135391697e-06, | |
| "loss": 0.0066, | |
| "step": 271500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.531466743375843e-06, | |
| "loss": 0.0091, | |
| "step": 272000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.45156135135999e-06, | |
| "loss": 0.0099, | |
| "step": 272500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 6.371655959344137e-06, | |
| "loss": 0.0071, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 6.291750567328283e-06, | |
| "loss": 0.0085, | |
| "step": 273500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.211845175312431e-06, | |
| "loss": 0.009, | |
| "step": 274000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.131939783296577e-06, | |
| "loss": 0.0085, | |
| "step": 274500 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 6.052034391280724e-06, | |
| "loss": 0.0085, | |
| "step": 275000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.97212899926487e-06, | |
| "loss": 0.0089, | |
| "step": 275500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.892223607249017e-06, | |
| "loss": 0.0081, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.812318215233164e-06, | |
| "loss": 0.009, | |
| "step": 276500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.732412823217311e-06, | |
| "loss": 0.0077, | |
| "step": 277000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.6525074312014575e-06, | |
| "loss": 0.0068, | |
| "step": 277500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.5726020391856045e-06, | |
| "loss": 0.0078, | |
| "step": 278000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.492696647169751e-06, | |
| "loss": 0.0078, | |
| "step": 278500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.412791255153898e-06, | |
| "loss": 0.008, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.332885863138045e-06, | |
| "loss": 0.0069, | |
| "step": 279500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.252980471122192e-06, | |
| "loss": 0.0083, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.173075079106339e-06, | |
| "loss": 0.008, | |
| "step": 280500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.093169687090485e-06, | |
| "loss": 0.0065, | |
| "step": 281000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.013264295074632e-06, | |
| "loss": 0.0081, | |
| "step": 281500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.933358903058779e-06, | |
| "loss": 0.0092, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.853453511042926e-06, | |
| "loss": 0.0078, | |
| "step": 282500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.773548119027072e-06, | |
| "loss": 0.0084, | |
| "step": 283000 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.693642727011219e-06, | |
| "loss": 0.0086, | |
| "step": 283500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.613737334995365e-06, | |
| "loss": 0.0087, | |
| "step": 284000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.533831942979513e-06, | |
| "loss": 0.0086, | |
| "step": 284500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.453926550963659e-06, | |
| "loss": 0.0088, | |
| "step": 285000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.374021158947806e-06, | |
| "loss": 0.0079, | |
| "step": 285500 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.294115766931953e-06, | |
| "loss": 0.0089, | |
| "step": 286000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.214210374916099e-06, | |
| "loss": 0.0096, | |
| "step": 286500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.134304982900247e-06, | |
| "loss": 0.0079, | |
| "step": 287000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.054399590884393e-06, | |
| "loss": 0.0084, | |
| "step": 287500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 3.97449419886854e-06, | |
| "loss": 0.008, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.894588806852686e-06, | |
| "loss": 0.0071, | |
| "step": 288500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.8146834148368332e-06, | |
| "loss": 0.0091, | |
| "step": 289000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.7347780228209802e-06, | |
| "loss": 0.0067, | |
| "step": 289500 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.6548726308051272e-06, | |
| "loss": 0.01, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.574967238789274e-06, | |
| "loss": 0.0075, | |
| "step": 290500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.4950618467734204e-06, | |
| "loss": 0.0085, | |
| "step": 291000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.415156454757567e-06, | |
| "loss": 0.0078, | |
| "step": 291500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.3352510627417144e-06, | |
| "loss": 0.0075, | |
| "step": 292000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.255345670725861e-06, | |
| "loss": 0.0088, | |
| "step": 292500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.1754402787100075e-06, | |
| "loss": 0.0084, | |
| "step": 293000 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.095534886694154e-06, | |
| "loss": 0.0083, | |
| "step": 293500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 3.015629494678301e-06, | |
| "loss": 0.008, | |
| "step": 294000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 2.9357241026624476e-06, | |
| "loss": 0.0065, | |
| "step": 294500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.8558187106465946e-06, | |
| "loss": 0.0073, | |
| "step": 295000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.775913318630741e-06, | |
| "loss": 0.0087, | |
| "step": 295500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.696007926614888e-06, | |
| "loss": 0.0084, | |
| "step": 296000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.616102534599035e-06, | |
| "loss": 0.0071, | |
| "step": 296500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.5361971425831817e-06, | |
| "loss": 0.0092, | |
| "step": 297000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.4562917505673283e-06, | |
| "loss": 0.008, | |
| "step": 297500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.3763863585514753e-06, | |
| "loss": 0.0085, | |
| "step": 298000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.296480966535622e-06, | |
| "loss": 0.0087, | |
| "step": 298500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.216575574519769e-06, | |
| "loss": 0.0088, | |
| "step": 299000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.1366701825039154e-06, | |
| "loss": 0.0077, | |
| "step": 299500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 2.056764790488062e-06, | |
| "loss": 0.008, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 1.976859398472209e-06, | |
| "loss": 0.0079, | |
| "step": 300500 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.8969540064563558e-06, | |
| "loss": 0.0082, | |
| "step": 301000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.8170486144405027e-06, | |
| "loss": 0.0076, | |
| "step": 301500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.7371432224246493e-06, | |
| "loss": 0.0076, | |
| "step": 302000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.6572378304087959e-06, | |
| "loss": 0.0084, | |
| "step": 302500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.5773324383929429e-06, | |
| "loss": 0.0086, | |
| "step": 303000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.4974270463770894e-06, | |
| "loss": 0.008, | |
| "step": 303500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.4175216543612362e-06, | |
| "loss": 0.0079, | |
| "step": 304000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.3376162623453832e-06, | |
| "loss": 0.0093, | |
| "step": 304500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.25771087032953e-06, | |
| "loss": 0.0077, | |
| "step": 305000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.1778054783136768e-06, | |
| "loss": 0.0091, | |
| "step": 305500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.0979000862978233e-06, | |
| "loss": 0.0077, | |
| "step": 306000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.0179946942819701e-06, | |
| "loss": 0.0081, | |
| "step": 306500 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 9.38089302266117e-07, | |
| "loss": 0.0073, | |
| "step": 307000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 8.581839102502638e-07, | |
| "loss": 0.008, | |
| "step": 307500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 7.782785182344105e-07, | |
| "loss": 0.0081, | |
| "step": 308000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 6.983731262185574e-07, | |
| "loss": 0.0086, | |
| "step": 308500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 6.18467734202704e-07, | |
| "loss": 0.0071, | |
| "step": 309000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 5.385623421868508e-07, | |
| "loss": 0.0087, | |
| "step": 309500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.5865695017099754e-07, | |
| "loss": 0.0078, | |
| "step": 310000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.7875155815514437e-07, | |
| "loss": 0.0073, | |
| "step": 310500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 2.988461661392911e-07, | |
| "loss": 0.0072, | |
| "step": 311000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 2.1894077412343788e-07, | |
| "loss": 0.008, | |
| "step": 311500 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 1.3903538210758463e-07, | |
| "loss": 0.0086, | |
| "step": 312000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 5.9129990091731396e-08, | |
| "loss": 0.0081, | |
| "step": 312500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 312870, | |
| "total_flos": 2.800067088663337e+17, | |
| "train_loss": 0.011290902120900085, | |
| "train_runtime": 83291.9658, | |
| "train_samples_per_second": 15.025, | |
| "train_steps_per_second": 3.756 | |
| } | |
| ], | |
| "max_steps": 312870, | |
| "num_train_epochs": 3, | |
| "total_flos": 2.800067088663337e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |