| { | |
| "best_metric": 20.9461, | |
| "best_model_checkpoint": "/content/gdrive/MyDrive/FYP/Models/t5-base-bt0/checkpoint-6000", | |
| "epoch": 1.5783540022547915, | |
| "eval_steps": 1000, | |
| "global_step": 7000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 0.237862229347229, | |
| "learning_rate": 2.7749718151071025e-05, | |
| "loss": 0.162, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "eval_gen_len": 8.6155, | |
| "eval_loss": 0.14371667802333832, | |
| "eval_rouge1": 19.4503, | |
| "eval_rouge2": 5.4909, | |
| "eval_rougeL": 19.2468, | |
| "eval_rougeLsum": 19.2536, | |
| "eval_runtime": 278.2764, | |
| "eval_samples_per_second": 10.393, | |
| "eval_steps_per_second": 1.301, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 0.19609051942825317, | |
| "learning_rate": 2.5501691093573846e-05, | |
| "loss": 0.1496, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "eval_gen_len": 8.2244, | |
| "eval_loss": 0.14119617640972137, | |
| "eval_rouge1": 19.988, | |
| "eval_rouge2": 5.6128, | |
| "eval_rougeL": 19.6697, | |
| "eval_rougeLsum": 19.6661, | |
| "eval_runtime": 274.1608, | |
| "eval_samples_per_second": 10.549, | |
| "eval_steps_per_second": 1.32, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.1622067540884018, | |
| "learning_rate": 2.3255918827508456e-05, | |
| "loss": 0.1482, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "eval_gen_len": 8.3513, | |
| "eval_loss": 0.13958066701889038, | |
| "eval_rouge1": 20.4724, | |
| "eval_rouge2": 5.918, | |
| "eval_rougeL": 20.1806, | |
| "eval_rougeLsum": 20.1879, | |
| "eval_runtime": 273.1755, | |
| "eval_samples_per_second": 10.587, | |
| "eval_steps_per_second": 1.325, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.22670583426952362, | |
| "learning_rate": 2.1003382187147688e-05, | |
| "loss": 0.1477, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "eval_gen_len": 8.5363, | |
| "eval_loss": 0.1383427083492279, | |
| "eval_rouge1": 20.0921, | |
| "eval_rouge2": 5.7649, | |
| "eval_rougeL": 19.8454, | |
| "eval_rougeLsum": 19.8533, | |
| "eval_runtime": 273.0415, | |
| "eval_samples_per_second": 10.592, | |
| "eval_steps_per_second": 1.326, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "grad_norm": 0.2830670475959778, | |
| "learning_rate": 1.8753100338218713e-05, | |
| "loss": 0.1432, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "eval_gen_len": 8.3544, | |
| "eval_loss": 0.13800442218780518, | |
| "eval_rouge1": 20.5887, | |
| "eval_rouge2": 6.0275, | |
| "eval_rougeL": 20.2746, | |
| "eval_rougeLsum": 20.2963, | |
| "eval_runtime": 274.4284, | |
| "eval_samples_per_second": 10.538, | |
| "eval_steps_per_second": 1.319, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "grad_norm": 0.3157794177532196, | |
| "learning_rate": 1.6507328072153326e-05, | |
| "loss": 0.1389, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "eval_gen_len": 8.4136, | |
| "eval_loss": 0.13757741451263428, | |
| "eval_rouge1": 20.9461, | |
| "eval_rouge2": 6.3797, | |
| "eval_rougeL": 20.7292, | |
| "eval_rougeLsum": 20.7592, | |
| "eval_runtime": 272.1466, | |
| "eval_samples_per_second": 10.627, | |
| "eval_steps_per_second": 1.33, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "grad_norm": 0.2532835900783539, | |
| "learning_rate": 1.4259301014656145e-05, | |
| "loss": 0.1393, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "eval_gen_len": 8.2206, | |
| "eval_loss": 0.1370321363210678, | |
| "eval_rouge1": 20.2115, | |
| "eval_rouge2": 5.8953, | |
| "eval_rougeL": 19.9895, | |
| "eval_rougeLsum": 20.0005, | |
| "eval_runtime": 269.4813, | |
| "eval_samples_per_second": 10.732, | |
| "eval_steps_per_second": 1.343, | |
| "step": 7000 | |
| } | |
| ], | |
| "logging_steps": 1000, | |
| "max_steps": 13305, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 1000, | |
| "total_flos": 3.409981499768832e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |