{ "best_metric": 75.75802081436439, "best_model_checkpoint": "/home/jcanete/ft-data/all_results/tar/albeto_base_8/epochs_2_bs_16_lr_5e-5/checkpoint-10500", "epoch": 2.0, "global_step": 10970, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.05, "eval_exact_match": 43.973509933774835, "eval_f1": 62.94849738783248, "step": 300 }, { "epoch": 0.09, "learning_rate": 4.773473108477667e-05, "loss": 2.4433, "step": 500 }, { "epoch": 0.11, "eval_exact_match": 47.682119205298015, "eval_f1": 66.74232170504494, "step": 600 }, { "epoch": 0.16, "eval_exact_match": 49.84862819299905, "eval_f1": 67.82623236742938, "step": 900 }, { "epoch": 0.18, "learning_rate": 4.545578851412944e-05, "loss": 1.8951, "step": 1000 }, { "epoch": 0.22, "eval_exact_match": 51.7123935666982, "eval_f1": 70.35883323184615, "step": 1200 }, { "epoch": 0.27, "learning_rate": 4.3176845943482223e-05, "loss": 1.7818, "step": 1500 }, { "epoch": 0.27, "eval_exact_match": 52.69631031220435, "eval_f1": 70.10225551135818, "step": 1500 }, { "epoch": 0.33, "eval_exact_match": 52.93282876064333, "eval_f1": 70.57743501701938, "step": 1800 }, { "epoch": 0.36, "learning_rate": 4.0897903372835004e-05, "loss": 1.7351, "step": 2000 }, { "epoch": 0.38, "eval_exact_match": 53.822138126773886, "eval_f1": 72.2913967717095, "step": 2100 }, { "epoch": 0.44, "eval_exact_match": 54.15326395458846, "eval_f1": 71.8960686504451, "step": 2400 }, { "epoch": 0.46, "learning_rate": 3.8618960802187785e-05, "loss": 1.6499, "step": 2500 }, { "epoch": 0.49, "eval_exact_match": 54.91012298959319, "eval_f1": 72.36339758418018, "step": 2700 }, { "epoch": 0.55, "learning_rate": 3.6340018231540566e-05, "loss": 1.6346, "step": 3000 }, { "epoch": 0.55, "eval_exact_match": 55.2317880794702, "eval_f1": 72.83417554949612, "step": 3000 }, { "epoch": 0.6, "eval_exact_match": 56.67928098391675, "eval_f1": 73.351425055483, "step": 3300 }, { "epoch": 0.64, "learning_rate": 3.406107566089335e-05, "loss": 1.6113, "step": 3500 }, { "epoch": 0.66, "eval_exact_match": 56.20624408703879, "eval_f1": 73.5623207381341, "step": 3600 }, { "epoch": 0.71, "eval_exact_match": 56.51844843897824, "eval_f1": 73.8479893696103, "step": 3900 }, { "epoch": 0.73, "learning_rate": 3.178213309024613e-05, "loss": 1.5583, "step": 4000 }, { "epoch": 0.77, "eval_exact_match": 57.11447492904446, "eval_f1": 74.4610534306545, "step": 4200 }, { "epoch": 0.82, "learning_rate": 2.9503190519598906e-05, "loss": 1.5623, "step": 4500 }, { "epoch": 0.82, "eval_exact_match": 57.60643330179754, "eval_f1": 74.50261751014862, "step": 4500 }, { "epoch": 0.88, "eval_exact_match": 57.228003784295176, "eval_f1": 74.4736305671364, "step": 4800 }, { "epoch": 0.91, "learning_rate": 2.722880583409298e-05, "loss": 1.5367, "step": 5000 }, { "epoch": 0.93, "eval_exact_match": 57.29422894985809, "eval_f1": 74.42957742124376, "step": 5100 }, { "epoch": 0.98, "eval_exact_match": 57.93755912961211, "eval_f1": 75.26960240667054, "step": 5400 }, { "epoch": 1.0, "learning_rate": 2.4949863263445765e-05, "loss": 1.4926, "step": 5500 }, { "epoch": 1.04, "eval_exact_match": 58.438978240302745, "eval_f1": 75.16537064693637, "step": 5700 }, { "epoch": 1.09, "learning_rate": 2.2670920692798542e-05, "loss": 1.238, "step": 6000 }, { "epoch": 1.09, "eval_exact_match": 57.994323557237465, "eval_f1": 75.09801725717467, "step": 6000 }, { "epoch": 1.15, "eval_exact_match": 58.306527909176914, "eval_f1": 75.10273058477487, "step": 6300 }, { "epoch": 1.19, "learning_rate": 2.0391978122151323e-05, "loss": 1.2301, "step": 6500 }, { "epoch": 1.2, "eval_exact_match": 57.6631977294229, "eval_f1": 75.16465753346614, "step": 6600 }, { "epoch": 1.26, "eval_exact_match": 57.83349101229896, "eval_f1": 74.40142589790919, "step": 6900 }, { "epoch": 1.28, "learning_rate": 1.8113035551504104e-05, "loss": 1.2176, "step": 7000 }, { "epoch": 1.31, "eval_exact_match": 58.39167455061495, "eval_f1": 75.34980345320106, "step": 7200 }, { "epoch": 1.37, "learning_rate": 1.583409298085688e-05, "loss": 1.2076, "step": 7500 }, { "epoch": 1.37, "eval_exact_match": 58.35383159886471, "eval_f1": 75.31554814626054, "step": 7500 }, { "epoch": 1.42, "eval_exact_match": 58.00378429517502, "eval_f1": 75.15654833167893, "step": 7800 }, { "epoch": 1.46, "learning_rate": 1.3555150410209664e-05, "loss": 1.2522, "step": 8000 }, { "epoch": 1.48, "eval_exact_match": 58.42005676442763, "eval_f1": 75.42667424347809, "step": 8100 }, { "epoch": 1.53, "eval_exact_match": 58.533585619678334, "eval_f1": 75.482174465219, "step": 8400 }, { "epoch": 1.55, "learning_rate": 1.1276207839562443e-05, "loss": 1.2069, "step": 8500 }, { "epoch": 1.59, "eval_exact_match": 58.438978240302745, "eval_f1": 75.37247235785051, "step": 8700 }, { "epoch": 1.64, "learning_rate": 9.001823154056518e-06, "loss": 1.2185, "step": 9000 }, { "epoch": 1.64, "eval_exact_match": 58.80794701986755, "eval_f1": 75.55945320105242, "step": 9000 }, { "epoch": 1.7, "eval_exact_match": 58.63765373699149, "eval_f1": 75.48349982961263, "step": 9300 }, { "epoch": 1.73, "learning_rate": 6.722880583409298e-06, "loss": 1.2221, "step": 9500 }, { "epoch": 1.75, "eval_exact_match": 58.7038789025544, "eval_f1": 75.52334202995019, "step": 9600 }, { "epoch": 1.8, "eval_exact_match": 58.87417218543046, "eval_f1": 75.59476406337527, "step": 9900 }, { "epoch": 1.82, "learning_rate": 4.443938012762078e-06, "loss": 1.1926, "step": 10000 }, { "epoch": 1.86, "eval_exact_match": 58.883632923368026, "eval_f1": 75.53002560032212, "step": 10200 }, { "epoch": 1.91, "learning_rate": 2.169553327256153e-06, "loss": 1.216, "step": 10500 }, { "epoch": 1.91, "eval_exact_match": 58.80794701986755, "eval_f1": 75.75802081436439, "step": 10500 }, { "epoch": 1.97, "eval_exact_match": 58.836329233680225, "eval_f1": 75.61405946041744, "step": 10800 }, { "epoch": 2.0, "step": 10970, "total_flos": 2544773915225664.0, "train_loss": 1.4681447838476385, "train_runtime": 1862.4153, "train_samples_per_second": 94.229, "train_steps_per_second": 5.89 } ], "max_steps": 10970, "num_train_epochs": 2, "total_flos": 2544773915225664.0, "trial_name": null, "trial_params": null }