|
{ |
|
"best_metric": 0.5188340210302143, |
|
"best_model_checkpoint": "models/Fit_HardNeg_with_bm25_unique_positive_shuffled_bge-m3-TripletLoss-triplet-legal-finetune/checkpoint-207", |
|
"epoch": 1.0, |
|
"eval_steps": 500, |
|
"global_step": 207, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.04830917874396135, |
|
"grad_norm": 2.6675727367401123, |
|
"learning_rate": 2.666666666666667e-07, |
|
"loss": 9.7628, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.0966183574879227, |
|
"grad_norm": 2.8036043643951416, |
|
"learning_rate": 5.333333333333335e-07, |
|
"loss": 9.7395, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.14492753623188406, |
|
"grad_norm": 2.622026205062866, |
|
"learning_rate": 8.000000000000001e-07, |
|
"loss": 9.7537, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.1932367149758454, |
|
"grad_norm": 2.709652900695801, |
|
"learning_rate": 1.066666666666667e-06, |
|
"loss": 9.7429, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.24154589371980675, |
|
"grad_norm": 2.856275796890259, |
|
"learning_rate": 1.3333333333333334e-06, |
|
"loss": 9.732, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.2898550724637681, |
|
"grad_norm": 3.1299498081207275, |
|
"learning_rate": 1.6000000000000001e-06, |
|
"loss": 9.6906, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.33816425120772947, |
|
"grad_norm": 2.6750693321228027, |
|
"learning_rate": 1.8666666666666669e-06, |
|
"loss": 9.6831, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.3864734299516908, |
|
"grad_norm": 3.187295436859131, |
|
"learning_rate": 2.133333333333334e-06, |
|
"loss": 9.6572, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.43478260869565216, |
|
"grad_norm": 2.98656964302063, |
|
"learning_rate": 2.4000000000000003e-06, |
|
"loss": 9.6315, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.4830917874396135, |
|
"grad_norm": 3.327542304992676, |
|
"learning_rate": 2.666666666666667e-06, |
|
"loss": 9.6043, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.5314009661835749, |
|
"grad_norm": 3.8943748474121094, |
|
"learning_rate": 2.9333333333333338e-06, |
|
"loss": 9.5636, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.5797101449275363, |
|
"grad_norm": 4.195869445800781, |
|
"learning_rate": 3.2000000000000003e-06, |
|
"loss": 9.5688, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.6280193236714976, |
|
"grad_norm": 4.08402156829834, |
|
"learning_rate": 3.4666666666666672e-06, |
|
"loss": 9.5443, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.6763285024154589, |
|
"grad_norm": 4.850657939910889, |
|
"learning_rate": 3.7333333333333337e-06, |
|
"loss": 9.5055, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.7246376811594203, |
|
"grad_norm": 4.306090354919434, |
|
"learning_rate": 4.000000000000001e-06, |
|
"loss": 9.4979, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.7729468599033816, |
|
"grad_norm": 4.171666622161865, |
|
"learning_rate": 4.266666666666668e-06, |
|
"loss": 9.4396, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.821256038647343, |
|
"grad_norm": 8.503725051879883, |
|
"learning_rate": 4.533333333333334e-06, |
|
"loss": 9.4772, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.8695652173913043, |
|
"grad_norm": 5.773899555206299, |
|
"learning_rate": 4.800000000000001e-06, |
|
"loss": 9.4261, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.9178743961352657, |
|
"grad_norm": 4.697412014007568, |
|
"learning_rate": 5.0666666666666676e-06, |
|
"loss": 9.4023, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.966183574879227, |
|
"grad_norm": 5.804080963134766, |
|
"learning_rate": 5.333333333333334e-06, |
|
"loss": 9.3668, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_accuracy@1": 0.3158525222946892, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_accuracy@10": 0.587381871422867, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_accuracy@3": 0.4582723279648609, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_accuracy@5": 0.5188340210302143, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_map@100": 0.40935732242081835, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_mrr@10": 0.40233201603991975, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_ndcg@10": 0.44675579037173485, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_precision@1": 0.3158525222946892, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_precision@10": 0.058738187142286694, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_precision@3": 0.1527574426549536, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_precision@5": 0.10376680420604285, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_recall@1": 0.3158525222946892, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_recall@10": 0.587381871422867, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_recall@3": 0.4582723279648609, |
|
"eval_Legal_IR_Evaluation_Bench_cosine_recall@5": 0.5188340210302143, |
|
"eval_loss": 4.606808185577393, |
|
"eval_runtime": 49.209, |
|
"eval_samples_per_second": 29.873, |
|
"eval_steps_per_second": 0.935, |
|
"step": 207 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 828, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 4, |
|
"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": 32, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|