| { | |
| "best_metric": 80.53869377255532, | |
| "best_model_checkpoint": "/root/turkic_qa/en_uzn_models/en_uzn_xlm_roberta_large_squad_model/checkpoint-2760", | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 3450, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "step": 690, | |
| "train_exact_match": 69.83016983016984, | |
| "train_f1": 83.62002651699788, | |
| "train_runtime": 28.9765, | |
| "train_samples_per_second": 43.553, | |
| "train_steps_per_second": 1.587 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 4.982762813568115, | |
| "learning_rate": 1e-05, | |
| "loss": 1.164, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_exact_match": 64.75, | |
| "eval_f1": 78.91171401645128, | |
| "eval_runtime": 92.299, | |
| "eval_samples_per_second": 43.901, | |
| "eval_steps_per_second": 1.571, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 1380, | |
| "train_exact_match": 77.22277722277722, | |
| "train_f1": 89.40203728575936, | |
| "train_runtime": 29.1697, | |
| "train_samples_per_second": 43.641, | |
| "train_steps_per_second": 1.577 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 21.06043243408203, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.7976, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_exact_match": 66.375, | |
| "eval_f1": 79.72828200772302, | |
| "eval_runtime": 92.6779, | |
| "eval_samples_per_second": 43.721, | |
| "eval_steps_per_second": 1.565, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 2070, | |
| "train_exact_match": 82.71728271728271, | |
| "train_f1": 92.4650831936441, | |
| "train_runtime": 29.7118, | |
| "train_samples_per_second": 43.585, | |
| "train_steps_per_second": 1.582 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 33.353965759277344, | |
| "learning_rate": 5e-06, | |
| "loss": 0.5401, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_exact_match": 66.46875, | |
| "eval_f1": 80.21293859314422, | |
| "eval_runtime": 92.4601, | |
| "eval_samples_per_second": 43.824, | |
| "eval_steps_per_second": 1.568, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 2760, | |
| "train_exact_match": 87.61238761238761, | |
| "train_f1": 95.04445785242362, | |
| "train_runtime": 29.8676, | |
| "train_samples_per_second": 43.358, | |
| "train_steps_per_second": 1.574 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 337.98419189453125, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.38, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_exact_match": 67.125, | |
| "eval_f1": 80.53869377255532, | |
| "eval_runtime": 92.9481, | |
| "eval_samples_per_second": 43.594, | |
| "eval_steps_per_second": 1.56, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 3450, | |
| "train_exact_match": 89.01098901098901, | |
| "train_f1": 95.70481727554154, | |
| "train_runtime": 28.4706, | |
| "train_samples_per_second": 43.519, | |
| "train_steps_per_second": 1.581 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 27.529285430908203, | |
| "learning_rate": 0.0, | |
| "loss": 0.2769, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_exact_match": 66.5625, | |
| "eval_f1": 80.1141164891758, | |
| "eval_runtime": 92.4845, | |
| "eval_samples_per_second": 43.813, | |
| "eval_steps_per_second": 1.568, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 3450, | |
| "total_flos": 6.719078999672064e+16, | |
| "train_loss": 0.6317114832781363, | |
| "train_runtime": 6178.5968, | |
| "train_samples_per_second": 15.613, | |
| "train_steps_per_second": 0.558 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 3450, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "total_flos": 6.719078999672064e+16, | |
| "train_batch_size": 28, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |