| { | |
| "best_metric": 51.829229828981525, | |
| "best_model_checkpoint": "/data/jcanete/all_results/mlqa/albeto_tiny/epochs_4_bs_16_lr_5e-5/checkpoint-9600", | |
| "epoch": 4.0, | |
| "global_step": 20516, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "eval_exact_match": 16.2, | |
| "eval_f1": 34.63862054971996, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.879118736595828e-05, | |
| "loss": 3.3701, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "eval_exact_match": 20.0, | |
| "eval_f1": 40.53628289728667, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "eval_exact_match": 23.0, | |
| "eval_f1": 43.6672661416669, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.757262624293235e-05, | |
| "loss": 2.7044, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "eval_exact_match": 22.4, | |
| "eval_f1": 43.72196312764883, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.635406511990641e-05, | |
| "loss": 2.4915, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "eval_exact_match": 23.2, | |
| "eval_f1": 45.18874378332089, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "eval_exact_match": 26.4, | |
| "eval_f1": 45.62609962984108, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.5135503996880485e-05, | |
| "loss": 2.4256, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "eval_exact_match": 27.0, | |
| "eval_f1": 45.0759507011648, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "eval_exact_match": 25.8, | |
| "eval_f1": 46.74317829494617, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.391694287385456e-05, | |
| "loss": 2.3753, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "eval_exact_match": 25.4, | |
| "eval_f1": 46.09934216831509, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.2698381750828625e-05, | |
| "loss": 2.3624, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "eval_exact_match": 27.6, | |
| "eval_f1": 47.56532306554079, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "eval_exact_match": 28.8, | |
| "eval_f1": 48.8118537425649, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.147982062780269e-05, | |
| "loss": 2.3088, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_exact_match": 26.6, | |
| "eval_f1": 48.11031277383674, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "eval_exact_match": 27.0, | |
| "eval_f1": 48.57193828888043, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.0261259504776764e-05, | |
| "loss": 2.28, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "eval_exact_match": 29.0, | |
| "eval_f1": 49.95989384569003, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.904513550399688e-05, | |
| "loss": 2.2532, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "eval_exact_match": 28.2, | |
| "eval_f1": 47.749518460354565, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "eval_exact_match": 28.0, | |
| "eval_f1": 47.95533226405339, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.7826574380970954e-05, | |
| "loss": 2.2263, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_exact_match": 30.4, | |
| "eval_f1": 50.35031495839047, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "eval_exact_match": 29.6, | |
| "eval_f1": 48.53740624490754, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.660801325794502e-05, | |
| "loss": 1.9813, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "eval_exact_match": 29.2, | |
| "eval_f1": 48.8968379190519, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.5389452134919087e-05, | |
| "loss": 1.9535, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "eval_exact_match": 29.0, | |
| "eval_f1": 49.5654691041312, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "eval_exact_match": 28.8, | |
| "eval_f1": 48.345747271344585, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.417332813413921e-05, | |
| "loss": 1.9314, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "eval_exact_match": 29.4, | |
| "eval_f1": 48.14012643359697, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "eval_exact_match": 29.2, | |
| "eval_f1": 50.52215104527601, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 3.295476701111328e-05, | |
| "loss": 1.911, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "eval_exact_match": 28.4, | |
| "eval_f1": 49.46281966862897, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.173620588808735e-05, | |
| "loss": 1.9372, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "eval_exact_match": 30.4, | |
| "eval_f1": 49.683785743427386, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "eval_exact_match": 29.2, | |
| "eval_f1": 49.43557912883076, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 3.0517644765061416e-05, | |
| "loss": 1.9246, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "eval_exact_match": 28.6, | |
| "eval_f1": 49.88678537045562, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "eval_exact_match": 29.6, | |
| "eval_f1": 49.07639865144071, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.9301520764281537e-05, | |
| "loss": 1.9307, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "eval_exact_match": 28.8, | |
| "eval_f1": 48.23767439357753, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.8082959641255607e-05, | |
| "loss": 1.9349, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "eval_exact_match": 31.2, | |
| "eval_f1": 50.781472465546585, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "eval_exact_match": 30.0, | |
| "eval_f1": 50.23351257981621, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.6864398518229676e-05, | |
| "loss": 1.9576, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "eval_exact_match": 31.0, | |
| "eval_f1": 51.829229828981525, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "eval_exact_match": 29.2, | |
| "eval_f1": 49.095681311256506, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.5645837395203742e-05, | |
| "loss": 1.9053, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "eval_exact_match": 28.8, | |
| "eval_f1": 48.97215068029405, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.4429713394423863e-05, | |
| "loss": 1.7791, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "eval_exact_match": 28.0, | |
| "eval_f1": 48.382507801772945, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "eval_exact_match": 30.2, | |
| "eval_f1": 50.230880304338086, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2.3211152271397936e-05, | |
| "loss": 1.6007, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "eval_exact_match": 29.6, | |
| "eval_f1": 50.00092829756792, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "eval_exact_match": 29.0, | |
| "eval_f1": 49.579059588205716, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 2.1992591148372002e-05, | |
| "loss": 1.6511, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "eval_exact_match": 28.6, | |
| "eval_f1": 49.2543102064297, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 2.0774030025346072e-05, | |
| "loss": 1.6304, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "eval_exact_match": 30.0, | |
| "eval_f1": 51.309331388015764, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "eval_exact_match": 29.0, | |
| "eval_f1": 49.1048900660549, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 1.9557906024566196e-05, | |
| "loss": 1.6205, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "eval_exact_match": 30.2, | |
| "eval_f1": 50.32681060468914, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "eval_exact_match": 29.6, | |
| "eval_f1": 48.99278230695513, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 1.8339344901540262e-05, | |
| "loss": 1.6519, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "eval_exact_match": 28.8, | |
| "eval_f1": 49.41627592032968, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 1.7123220900760383e-05, | |
| "loss": 1.6335, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "eval_exact_match": 30.6, | |
| "eval_f1": 50.570542671457055, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "eval_exact_match": 27.8, | |
| "eval_f1": 48.140510336166955, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 1.5904659777734453e-05, | |
| "loss": 1.639, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "eval_exact_match": 31.6, | |
| "eval_f1": 50.75177934290833, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "eval_exact_match": 30.8, | |
| "eval_f1": 50.33659610449098, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1.4686098654708522e-05, | |
| "loss": 1.6372, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "eval_exact_match": 30.4, | |
| "eval_f1": 49.89566233626405, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.3467537531682588e-05, | |
| "loss": 1.6279, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "eval_exact_match": 30.0, | |
| "eval_f1": 48.826938690847044, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "eval_exact_match": 29.0, | |
| "eval_f1": 48.18167854101267, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.225141353090271e-05, | |
| "loss": 1.5829, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "eval_exact_match": 28.6, | |
| "eval_f1": 47.470440396461385, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "eval_exact_match": 27.2, | |
| "eval_f1": 46.74933772498741, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 1.103285240787678e-05, | |
| "loss": 1.403, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "eval_exact_match": 27.8, | |
| "eval_f1": 47.008373124089864, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 9.814291284850848e-06, | |
| "loss": 1.4313, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "eval_exact_match": 28.8, | |
| "eval_f1": 47.451193025700725, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "eval_exact_match": 30.0, | |
| "eval_f1": 48.87142197992581, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 8.595730161824918e-06, | |
| "loss": 1.4254, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "eval_exact_match": 28.8, | |
| "eval_f1": 48.37616004379432, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "eval_exact_match": 28.8, | |
| "eval_f1": 48.3576189811418, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 7.377169038798987e-06, | |
| "loss": 1.3888, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "eval_exact_match": 28.0, | |
| "eval_f1": 47.60739154944984, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 6.1610450380191074e-06, | |
| "loss": 1.3774, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "eval_exact_match": 28.6, | |
| "eval_f1": 47.69237616167284, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "eval_exact_match": 27.8, | |
| "eval_f1": 47.57014705223912, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 4.942483914993176e-06, | |
| "loss": 1.404, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "eval_exact_match": 28.4, | |
| "eval_f1": 47.985261787062825, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "eval_exact_match": 28.2, | |
| "eval_f1": 48.037524314173936, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.723922791967245e-06, | |
| "loss": 1.4078, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "eval_exact_match": 28.0, | |
| "eval_f1": 47.65747670612322, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 2.505361668941314e-06, | |
| "loss": 1.4156, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "eval_exact_match": 28.4, | |
| "eval_f1": 47.348293727766865, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "eval_exact_match": 28.2, | |
| "eval_f1": 47.41990249421838, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 1.289237668161435e-06, | |
| "loss": 1.4043, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "eval_exact_match": 28.0, | |
| "eval_f1": 47.442334275454485, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "eval_exact_match": 27.6, | |
| "eval_f1": 47.41643572955593, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 7.0676545135504e-08, | |
| "loss": 1.4053, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 20516, | |
| "total_flos": 846250173819072.0, | |
| "train_loss": 1.8602011191215075, | |
| "train_runtime": 2243.5347, | |
| "train_samples_per_second": 146.289, | |
| "train_steps_per_second": 9.144 | |
| } | |
| ], | |
| "max_steps": 20516, | |
| "num_train_epochs": 4, | |
| "total_flos": 846250173819072.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |