| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.09235315847801995, | |
| "eval_steps": 100, | |
| "global_step": 500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01847063169560399, | |
| "grad_norm": 17.696088790893555, | |
| "learning_rate": 1.980139427847242e-05, | |
| "loss": 0.4835, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01847063169560399, | |
| "eval_cosine_accuracy@1": 0.8868686868686869, | |
| "eval_cosine_accuracy@10": 0.9579124579124579, | |
| "eval_cosine_accuracy@3": 0.9343434343434344, | |
| "eval_cosine_accuracy@5": 0.9454545454545454, | |
| "eval_cosine_map@100": 0.9137674463447905, | |
| "eval_cosine_mrr@10": 0.9124690021912245, | |
| "eval_cosine_ndcg@10": 0.9236137131767355, | |
| "eval_cosine_precision@1": 0.8868686868686869, | |
| "eval_cosine_precision@10": 0.09579124579124577, | |
| "eval_cosine_precision@3": 0.3114478114478115, | |
| "eval_cosine_precision@5": 0.18909090909090906, | |
| "eval_cosine_recall@1": 0.8868686868686869, | |
| "eval_cosine_recall@10": 0.9579124579124579, | |
| "eval_cosine_recall@3": 0.9343434343434344, | |
| "eval_cosine_recall@5": 0.9454545454545454, | |
| "eval_loss": 0.07506837695837021, | |
| "eval_runtime": 49.7303, | |
| "eval_samples_per_second": 140.759, | |
| "eval_sequential_score": 0.9137674463447905, | |
| "eval_steps_per_second": 8.808, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03694126339120798, | |
| "grad_norm": 2.3271963596343994, | |
| "learning_rate": 1.917211301505453e-05, | |
| "loss": 0.0646, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.03694126339120798, | |
| "eval_cosine_accuracy@1": 0.9195286195286195, | |
| "eval_cosine_accuracy@10": 0.967003367003367, | |
| "eval_cosine_accuracy@3": 0.9518518518518518, | |
| "eval_cosine_accuracy@5": 0.9612794612794613, | |
| "eval_cosine_map@100": 0.9383862806127067, | |
| "eval_cosine_mrr@10": 0.937248810859922, | |
| "eval_cosine_ndcg@10": 0.9445895366693552, | |
| "eval_cosine_precision@1": 0.9195286195286195, | |
| "eval_cosine_precision@10": 0.0967003367003367, | |
| "eval_cosine_precision@3": 0.317283950617284, | |
| "eval_cosine_precision@5": 0.19225589225589226, | |
| "eval_cosine_recall@1": 0.9195286195286195, | |
| "eval_cosine_recall@10": 0.967003367003367, | |
| "eval_cosine_recall@3": 0.9518518518518518, | |
| "eval_cosine_recall@5": 0.9612794612794613, | |
| "eval_loss": 0.05896875262260437, | |
| "eval_runtime": 51.3489, | |
| "eval_samples_per_second": 136.322, | |
| "eval_sequential_score": 0.9383862806127067, | |
| "eval_steps_per_second": 8.53, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.05541189508681197, | |
| "grad_norm": 3.7376925945281982, | |
| "learning_rate": 1.8139290433532415e-05, | |
| "loss": 0.0594, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.05541189508681197, | |
| "eval_cosine_accuracy@1": 0.9303030303030303, | |
| "eval_cosine_accuracy@10": 0.9737373737373738, | |
| "eval_cosine_accuracy@3": 0.9579124579124579, | |
| "eval_cosine_accuracy@5": 0.9656565656565657, | |
| "eval_cosine_map@100": 0.9462389210939364, | |
| "eval_cosine_mrr@10": 0.9454570840681954, | |
| "eval_cosine_ndcg@10": 0.9523521034308455, | |
| "eval_cosine_precision@1": 0.9303030303030303, | |
| "eval_cosine_precision@10": 0.09737373737373735, | |
| "eval_cosine_precision@3": 0.31930415263748596, | |
| "eval_cosine_precision@5": 0.1931313131313131, | |
| "eval_cosine_recall@1": 0.9303030303030303, | |
| "eval_cosine_recall@10": 0.9737373737373738, | |
| "eval_cosine_recall@3": 0.9579124579124579, | |
| "eval_cosine_recall@5": 0.9656565656565657, | |
| "eval_loss": 0.051894593983888626, | |
| "eval_runtime": 49.496, | |
| "eval_samples_per_second": 141.426, | |
| "eval_sequential_score": 0.9462389210939364, | |
| "eval_steps_per_second": 8.849, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07388252678241596, | |
| "grad_norm": 0.3877984583377838, | |
| "learning_rate": 1.6748367163042577e-05, | |
| "loss": 0.0471, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.07388252678241596, | |
| "eval_cosine_accuracy@1": 0.9367003367003367, | |
| "eval_cosine_accuracy@10": 0.9750841750841751, | |
| "eval_cosine_accuracy@3": 0.9612794612794613, | |
| "eval_cosine_accuracy@5": 0.969023569023569, | |
| "eval_cosine_map@100": 0.9513700816079773, | |
| "eval_cosine_mrr@10": 0.9505351130351131, | |
| "eval_cosine_ndcg@10": 0.9565510675566292, | |
| "eval_cosine_precision@1": 0.9367003367003367, | |
| "eval_cosine_precision@10": 0.09750841750841752, | |
| "eval_cosine_precision@3": 0.3204264870931538, | |
| "eval_cosine_precision@5": 0.19380471380471379, | |
| "eval_cosine_recall@1": 0.9367003367003367, | |
| "eval_cosine_recall@10": 0.9750841750841751, | |
| "eval_cosine_recall@3": 0.9612794612794613, | |
| "eval_cosine_recall@5": 0.969023569023569, | |
| "eval_loss": 0.04832224175333977, | |
| "eval_runtime": 49.2695, | |
| "eval_samples_per_second": 142.076, | |
| "eval_sequential_score": 0.9513700816079773, | |
| "eval_steps_per_second": 8.89, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09235315847801995, | |
| "grad_norm": 0.9424126744270325, | |
| "learning_rate": 1.5060539027168317e-05, | |
| "loss": 0.0524, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.09235315847801995, | |
| "eval_cosine_accuracy@1": 0.9387205387205387, | |
| "eval_cosine_accuracy@10": 0.9787878787878788, | |
| "eval_cosine_accuracy@3": 0.9622895622895623, | |
| "eval_cosine_accuracy@5": 0.9703703703703703, | |
| "eval_cosine_map@100": 0.9530933506069861, | |
| "eval_cosine_mrr@10": 0.9525124258457593, | |
| "eval_cosine_ndcg@10": 0.9588799906525647, | |
| "eval_cosine_precision@1": 0.9387205387205387, | |
| "eval_cosine_precision@10": 0.09787878787878787, | |
| "eval_cosine_precision@3": 0.3207631874298541, | |
| "eval_cosine_precision@5": 0.19407407407407404, | |
| "eval_cosine_recall@1": 0.9387205387205387, | |
| "eval_cosine_recall@10": 0.9787878787878788, | |
| "eval_cosine_recall@3": 0.9622895622895623, | |
| "eval_cosine_recall@5": 0.9703703703703703, | |
| "eval_loss": 0.04548173025250435, | |
| "eval_runtime": 49.8784, | |
| "eval_samples_per_second": 140.341, | |
| "eval_sequential_score": 0.9530933506069861, | |
| "eval_steps_per_second": 8.781, | |
| "step": 500 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 1500, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |