| { | |
| "best_metric": 70.54026450257521, | |
| "best_model_checkpoint": "/home/jcanete/ft-data/all_results/tar/albeto_base_4/epochs_3_bs_16_lr_5e-5/checkpoint-9000", | |
| "epoch": 3.0, | |
| "global_step": 16455, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "eval_exact_match": 34.645222327341536, | |
| "eval_f1": 52.13460975937906, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.848678213309025e-05, | |
| "loss": 2.8212, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "eval_exact_match": 40.38789025543993, | |
| "eval_f1": 58.80108695949609, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "eval_exact_match": 44.1438032166509, | |
| "eval_f1": 61.68021475413032, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.6967487085992104e-05, | |
| "loss": 2.147, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "eval_exact_match": 45.08041627246925, | |
| "eval_f1": 63.22958368722072, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.544819203889396e-05, | |
| "loss": 2.0333, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "eval_exact_match": 46.09271523178808, | |
| "eval_f1": 63.34953738572746, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "eval_exact_match": 47.13339640491959, | |
| "eval_f1": 64.76366505969077, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.392889699179581e-05, | |
| "loss": 1.9667, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "eval_exact_match": 47.4077578051088, | |
| "eval_f1": 65.80706119842145, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "eval_exact_match": 48.55250709555345, | |
| "eval_f1": 66.49751567340297, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.2409601944697666e-05, | |
| "loss": 1.8743, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "eval_exact_match": 48.84578997161778, | |
| "eval_f1": 66.76044504218349, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.089030689759952e-05, | |
| "loss": 1.8655, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "eval_exact_match": 49.55534531693472, | |
| "eval_f1": 67.55609852983001, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_exact_match": 49.45127719962157, | |
| "eval_f1": 67.28955811469469, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.9371011850501374e-05, | |
| "loss": 1.8223, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "eval_exact_match": 50.08514664143803, | |
| "eval_f1": 68.11294509259712, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "eval_exact_match": 50.38789025543993, | |
| "eval_f1": 68.15488930053279, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.785171680340323e-05, | |
| "loss": 1.7684, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "eval_exact_match": 51.05960264900662, | |
| "eval_f1": 69.07576620387981, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.6335460346399276e-05, | |
| "loss": 1.7677, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "eval_exact_match": 51.561021759697255, | |
| "eval_f1": 69.20806146876814, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "eval_exact_match": 51.61778618732261, | |
| "eval_f1": 69.20269381359182, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.481616529930112e-05, | |
| "loss": 1.7566, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "eval_exact_match": 51.68401135288553, | |
| "eval_f1": 69.22415299954444, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "eval_exact_match": 51.82592242194891, | |
| "eval_f1": 69.68048492944583, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.329990884229718e-05, | |
| "loss": 1.6976, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "eval_exact_match": 51.84484389782403, | |
| "eval_f1": 69.61844469538968, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.1780613795199024e-05, | |
| "loss": 1.3996, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "eval_exact_match": 51.79754020813623, | |
| "eval_f1": 69.51904893791095, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "eval_exact_match": 51.72185430463576, | |
| "eval_f1": 69.18971361078592, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.0261318748100882e-05, | |
| "loss": 1.3947, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "eval_exact_match": 51.82592242194891, | |
| "eval_f1": 69.9010405722466, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "eval_exact_match": 51.74077578051088, | |
| "eval_f1": 69.21425877541581, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.8742023701002736e-05, | |
| "loss": 1.3898, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "eval_exact_match": 52.64900662251656, | |
| "eval_f1": 70.03047551437781, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.722272865390459e-05, | |
| "loss": 1.3861, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "eval_exact_match": 52.232734153263955, | |
| "eval_f1": 69.84348950849247, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "eval_exact_match": 52.05298013245033, | |
| "eval_f1": 69.85616305536696, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.5703433606806444e-05, | |
| "loss": 1.4272, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "eval_exact_match": 52.42194891201514, | |
| "eval_f1": 70.12191052992236, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "eval_exact_match": 52.327341532639544, | |
| "eval_f1": 70.1945849880174, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.4184138559708297e-05, | |
| "loss": 1.387, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "eval_exact_match": 52.73415326395459, | |
| "eval_f1": 70.53063216515172, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.266484351261015e-05, | |
| "loss": 1.4181, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "eval_exact_match": 52.980132450331126, | |
| "eval_f1": 70.54026450257521, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "eval_exact_match": 52.327341532639544, | |
| "eval_f1": 70.03067034409061, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.1145548465512002e-05, | |
| "loss": 1.3941, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "eval_exact_match": 52.75307473982971, | |
| "eval_f1": 70.53593438103856, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "eval_exact_match": 52.37464522232734, | |
| "eval_f1": 69.9459164585827, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.9629292008508053e-05, | |
| "loss": 1.4019, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "eval_exact_match": 52.544938505203405, | |
| "eval_f1": 70.19906201457854, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.8109996961409907e-05, | |
| "loss": 1.4031, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "eval_exact_match": 52.61116367076632, | |
| "eval_f1": 70.3522720546097, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "eval_exact_match": 53.02743614001892, | |
| "eval_f1": 70.4807280516286, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.659070191431176e-05, | |
| "loss": 1.3625, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "eval_exact_match": 51.88268684957427, | |
| "eval_f1": 69.74458138686404, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "eval_exact_match": 51.66508987701041, | |
| "eval_f1": 69.55251413256356, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.5071406867213613e-05, | |
| "loss": 1.0261, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "eval_exact_match": 51.12582781456954, | |
| "eval_f1": 69.45293644822506, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.3555150410209664e-05, | |
| "loss": 0.9969, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "eval_exact_match": 51.84484389782403, | |
| "eval_f1": 69.82132514934786, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "eval_exact_match": 51.106906338694415, | |
| "eval_f1": 69.26705027187695, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.2035855363111518e-05, | |
| "loss": 1.0186, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "eval_exact_match": 51.59886471144749, | |
| "eval_f1": 69.76823383346489, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "eval_exact_match": 51.45695364238411, | |
| "eval_f1": 69.76810827781901, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.051656031601337e-05, | |
| "loss": 1.0272, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "eval_exact_match": 51.35288552507095, | |
| "eval_f1": 69.64753731877455, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 8.997265268915224e-06, | |
| "loss": 1.0469, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "eval_exact_match": 51.324503311258276, | |
| "eval_f1": 69.46890832651873, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "eval_exact_match": 51.343424787133394, | |
| "eval_f1": 69.28540200400431, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.4810088119112735e-06, | |
| "loss": 1.0324, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "eval_exact_match": 51.27719962157048, | |
| "eval_f1": 69.21410385147476, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "eval_exact_match": 51.400189214758754, | |
| "eval_f1": 69.48987451431931, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.964752354907324e-06, | |
| "loss": 1.026, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "eval_exact_match": 51.14474929044466, | |
| "eval_f1": 69.48096843416343, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.445457307809177e-06, | |
| "loss": 1.0191, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "eval_exact_match": 51.333964049195835, | |
| "eval_f1": 69.47772191155812, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "eval_exact_match": 51.68401135288553, | |
| "eval_f1": 69.54969415984502, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.9261622607110303e-06, | |
| "loss": 0.9864, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "eval_exact_match": 51.37180700094607, | |
| "eval_f1": 69.56621150352814, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "eval_exact_match": 51.561021759697255, | |
| "eval_f1": 69.58651422829355, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.4068672136128838e-06, | |
| "loss": 1.0319, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "eval_exact_match": 51.627246925260174, | |
| "eval_f1": 69.53955054930253, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 16455, | |
| "total_flos": 3819871314614016.0, | |
| "train_loss": 1.4580001696595593, | |
| "train_runtime": 2422.3692, | |
| "train_samples_per_second": 108.671, | |
| "train_steps_per_second": 6.793 | |
| } | |
| ], | |
| "max_steps": 16455, | |
| "num_train_epochs": 3, | |
| "total_flos": 3819871314614016.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |