| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 15.0, | |
| "eval_steps": 500, | |
| "global_step": 2700, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 2.7777777777777777, | |
| "grad_norm": 0.20568367838859558, | |
| "learning_rate": 9.980000000000001e-06, | |
| "loss": 0.1444, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "eval_runtime": 0.9226, | |
| "eval_samples_per_second": 0.0, | |
| "eval_steps_per_second": 0.0, | |
| "eval_test-eval_cosine_accuracy@1": 0.8393782383419689, | |
| "eval_test-eval_cosine_accuracy@10": 0.9948186528497409, | |
| "eval_test-eval_cosine_accuracy@3": 0.9637305699481865, | |
| "eval_test-eval_cosine_accuracy@5": 0.9844559585492227, | |
| "eval_test-eval_cosine_map@100": 0.9030430134057077, | |
| "eval_test-eval_cosine_mrr@10": 0.9026112344765196, | |
| "eval_test-eval_cosine_ndcg@10": 0.9257511988328408, | |
| "eval_test-eval_cosine_precision@1": 0.8393782383419689, | |
| "eval_test-eval_cosine_precision@10": 0.09948186528497409, | |
| "eval_test-eval_cosine_precision@3": 0.32124352331606215, | |
| "eval_test-eval_cosine_precision@5": 0.19689119170984457, | |
| "eval_test-eval_cosine_recall@1": 0.8393782383419689, | |
| "eval_test-eval_cosine_recall@10": 0.9948186528497409, | |
| "eval_test-eval_cosine_recall@3": 0.9637305699481865, | |
| "eval_test-eval_cosine_recall@5": 0.9844559585492227, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 5.555555555555555, | |
| "grad_norm": 0.7912865281105042, | |
| "learning_rate": 1.9980000000000002e-05, | |
| "loss": 0.0916, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.555555555555555, | |
| "eval_runtime": 0.7556, | |
| "eval_samples_per_second": 0.0, | |
| "eval_steps_per_second": 0.0, | |
| "eval_test-eval_cosine_accuracy@1": 0.8601036269430051, | |
| "eval_test-eval_cosine_accuracy@10": 1.0, | |
| "eval_test-eval_cosine_accuracy@3": 0.9792746113989638, | |
| "eval_test-eval_cosine_accuracy@5": 1.0, | |
| "eval_test-eval_cosine_map@100": 0.9189119170984456, | |
| "eval_test-eval_cosine_mrr@10": 0.9189119170984456, | |
| "eval_test-eval_cosine_ndcg@10": 0.9394665325932218, | |
| "eval_test-eval_cosine_precision@1": 0.8601036269430051, | |
| "eval_test-eval_cosine_precision@10": 0.09999999999999999, | |
| "eval_test-eval_cosine_precision@3": 0.32642487046632124, | |
| "eval_test-eval_cosine_precision@5": 0.19999999999999998, | |
| "eval_test-eval_cosine_recall@1": 0.8601036269430051, | |
| "eval_test-eval_cosine_recall@10": 1.0, | |
| "eval_test-eval_cosine_recall@3": 0.9792746113989638, | |
| "eval_test-eval_cosine_recall@5": 1.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 8.333333333333334, | |
| "grad_norm": 0.019018512219190598, | |
| "learning_rate": 0.0, | |
| "loss": 0.0842, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 8.333333333333334, | |
| "eval_runtime": 0.8035, | |
| "eval_samples_per_second": 0.0, | |
| "eval_steps_per_second": 0.0, | |
| "eval_test-eval_cosine_accuracy@1": 0.8601036269430051, | |
| "eval_test-eval_cosine_accuracy@10": 1.0, | |
| "eval_test-eval_cosine_accuracy@3": 0.9792746113989638, | |
| "eval_test-eval_cosine_accuracy@5": 1.0, | |
| "eval_test-eval_cosine_map@100": 0.9189119170984456, | |
| "eval_test-eval_cosine_mrr@10": 0.9189119170984456, | |
| "eval_test-eval_cosine_ndcg@10": 0.9394665325932218, | |
| "eval_test-eval_cosine_precision@1": 0.8601036269430051, | |
| "eval_test-eval_cosine_precision@10": 0.09999999999999999, | |
| "eval_test-eval_cosine_precision@3": 0.32642487046632124, | |
| "eval_test-eval_cosine_precision@5": 0.19999999999999998, | |
| "eval_test-eval_cosine_recall@1": 0.8601036269430051, | |
| "eval_test-eval_cosine_recall@10": 1.0, | |
| "eval_test-eval_cosine_recall@3": 0.9792746113989638, | |
| "eval_test-eval_cosine_recall@5": 1.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 11.11111111111111, | |
| "grad_norm": 0.14690840244293213, | |
| "learning_rate": 0.0, | |
| "loss": 0.0877, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 11.11111111111111, | |
| "eval_runtime": 0.9091, | |
| "eval_samples_per_second": 0.0, | |
| "eval_steps_per_second": 0.0, | |
| "eval_test-eval_cosine_accuracy@1": 0.8601036269430051, | |
| "eval_test-eval_cosine_accuracy@10": 1.0, | |
| "eval_test-eval_cosine_accuracy@3": 0.9792746113989638, | |
| "eval_test-eval_cosine_accuracy@5": 1.0, | |
| "eval_test-eval_cosine_map@100": 0.9189119170984456, | |
| "eval_test-eval_cosine_mrr@10": 0.9189119170984456, | |
| "eval_test-eval_cosine_ndcg@10": 0.9394665325932218, | |
| "eval_test-eval_cosine_precision@1": 0.8601036269430051, | |
| "eval_test-eval_cosine_precision@10": 0.09999999999999999, | |
| "eval_test-eval_cosine_precision@3": 0.32642487046632124, | |
| "eval_test-eval_cosine_precision@5": 0.19999999999999998, | |
| "eval_test-eval_cosine_recall@1": 0.8601036269430051, | |
| "eval_test-eval_cosine_recall@10": 1.0, | |
| "eval_test-eval_cosine_recall@3": 0.9792746113989638, | |
| "eval_test-eval_cosine_recall@5": 1.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 13.88888888888889, | |
| "grad_norm": 1.6923232078552246, | |
| "learning_rate": 0.0, | |
| "loss": 0.0827, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 13.88888888888889, | |
| "eval_runtime": 1.0313, | |
| "eval_samples_per_second": 0.0, | |
| "eval_steps_per_second": 0.0, | |
| "eval_test-eval_cosine_accuracy@1": 0.8601036269430051, | |
| "eval_test-eval_cosine_accuracy@10": 1.0, | |
| "eval_test-eval_cosine_accuracy@3": 0.9792746113989638, | |
| "eval_test-eval_cosine_accuracy@5": 1.0, | |
| "eval_test-eval_cosine_map@100": 0.9189119170984456, | |
| "eval_test-eval_cosine_mrr@10": 0.9189119170984456, | |
| "eval_test-eval_cosine_ndcg@10": 0.9394665325932218, | |
| "eval_test-eval_cosine_precision@1": 0.8601036269430051, | |
| "eval_test-eval_cosine_precision@10": 0.09999999999999999, | |
| "eval_test-eval_cosine_precision@3": 0.32642487046632124, | |
| "eval_test-eval_cosine_precision@5": 0.19999999999999998, | |
| "eval_test-eval_cosine_recall@1": 0.8601036269430051, | |
| "eval_test-eval_cosine_recall@10": 1.0, | |
| "eval_test-eval_cosine_recall@3": 0.9792746113989638, | |
| "eval_test-eval_cosine_recall@5": 1.0, | |
| "step": 2500 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 2700, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 15, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |