| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 1.0, |
| "eval_steps": 79, |
| "global_step": 782, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0, |
| "eval_cosine_accuracy@1": 0.2495, |
| "eval_cosine_accuracy@10": 0.4625, |
| "eval_cosine_accuracy@3": 0.344, |
| "eval_cosine_accuracy@5": 0.3965, |
| "eval_cosine_map@100": 0.1559092484894483, |
| "eval_cosine_mrr@10": 0.3108763888888889, |
| "eval_cosine_ndcg@10": 0.2096322787377784, |
| "eval_cosine_precision@1": 0.2495, |
| "eval_cosine_precision@10": 0.1072, |
| "eval_cosine_precision@3": 0.18666666666666665, |
| "eval_cosine_precision@5": 0.15159999999999998, |
| "eval_cosine_recall@1": 0.0643912531049299, |
| "eval_cosine_recall@10": 0.20289694590062918, |
| "eval_cosine_recall@3": 0.12590059739815648, |
| "eval_cosine_recall@5": 0.15806433844030476, |
| "eval_loss": 3.089846134185791, |
| "eval_runtime": 271.6862, |
| "eval_samples_per_second": 7.361, |
| "eval_steps_per_second": 0.118, |
| "step": 0 |
| }, |
| { |
| "epoch": 0.10112, |
| "grad_norm": 6.488596439361572, |
| "learning_rate": 4.936708860759494e-05, |
| "loss": 1.1095, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.10112, |
| "eval_cosine_accuracy@1": 0.559, |
| "eval_cosine_accuracy@10": 0.8265, |
| "eval_cosine_accuracy@3": 0.698, |
| "eval_cosine_accuracy@5": 0.7555, |
| "eval_cosine_map@100": 0.4500740959017874, |
| "eval_cosine_mrr@10": 0.6431708333333328, |
| "eval_cosine_ndcg@10": 0.5303275584101319, |
| "eval_cosine_precision@1": 0.559, |
| "eval_cosine_precision@10": 0.24805000000000002, |
| "eval_cosine_precision@3": 0.44783333333333336, |
| "eval_cosine_precision@5": 0.36169999999999997, |
| "eval_cosine_recall@1": 0.17592097493397865, |
| "eval_cosine_recall@10": 0.5413604201504909, |
| "eval_cosine_recall@3": 0.3544624337101741, |
| "eval_cosine_recall@5": 0.43704739404031684, |
| "eval_loss": 0.4636768698692322, |
| "eval_runtime": 279.9611, |
| "eval_samples_per_second": 7.144, |
| "eval_steps_per_second": 0.114, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.20224, |
| "grad_norm": 4.111133575439453, |
| "learning_rate": 4.4452347083926033e-05, |
| "loss": 0.369, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.20224, |
| "eval_cosine_accuracy@1": 0.628, |
| "eval_cosine_accuracy@10": 0.879, |
| "eval_cosine_accuracy@3": 0.7655, |
| "eval_cosine_accuracy@5": 0.817, |
| "eval_cosine_map@100": 0.5197308098624411, |
| "eval_cosine_mrr@10": 0.7098900793650795, |
| "eval_cosine_ndcg@10": 0.6001941938931266, |
| "eval_cosine_precision@1": 0.628, |
| "eval_cosine_precision@10": 0.27375, |
| "eval_cosine_precision@3": 0.49133333333333334, |
| "eval_cosine_precision@5": 0.3966, |
| "eval_cosine_recall@1": 0.2136060048255348, |
| "eval_cosine_recall@10": 0.607901124413225, |
| "eval_cosine_recall@3": 0.4115900630337399, |
| "eval_cosine_recall@5": 0.4973507574392067, |
| "eval_loss": 0.29928770661354065, |
| "eval_runtime": 278.2977, |
| "eval_samples_per_second": 7.187, |
| "eval_steps_per_second": 0.115, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.30336, |
| "grad_norm": 2.7525439262390137, |
| "learning_rate": 3.883357041251778e-05, |
| "loss": 0.2779, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.30336, |
| "eval_cosine_accuracy@1": 0.637, |
| "eval_cosine_accuracy@10": 0.8895, |
| "eval_cosine_accuracy@3": 0.779, |
| "eval_cosine_accuracy@5": 0.829, |
| "eval_cosine_map@100": 0.5359616957068315, |
| "eval_cosine_mrr@10": 0.7192515873015861, |
| "eval_cosine_ndcg@10": 0.6154242844350049, |
| "eval_cosine_precision@1": 0.637, |
| "eval_cosine_precision@10": 0.28485000000000005, |
| "eval_cosine_precision@3": 0.5083333333333333, |
| "eval_cosine_precision@5": 0.4151, |
| "eval_cosine_recall@1": 0.21074726147683806, |
| "eval_cosine_recall@10": 0.6277595587315691, |
| "eval_cosine_recall@3": 0.4188129487357172, |
| "eval_cosine_recall@5": 0.5121328631227711, |
| "eval_loss": 0.25021883845329285, |
| "eval_runtime": 277.4835, |
| "eval_samples_per_second": 7.208, |
| "eval_steps_per_second": 0.115, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.40448, |
| "grad_norm": 2.8394055366516113, |
| "learning_rate": 3.321479374110953e-05, |
| "loss": 0.224, |
| "step": 316 |
| }, |
| { |
| "epoch": 0.40448, |
| "eval_cosine_accuracy@1": 0.668, |
| "eval_cosine_accuracy@10": 0.914, |
| "eval_cosine_accuracy@3": 0.8115, |
| "eval_cosine_accuracy@5": 0.865, |
| "eval_cosine_map@100": 0.5721924042351211, |
| "eval_cosine_mrr@10": 0.7500142857142853, |
| "eval_cosine_ndcg@10": 0.6495973268729938, |
| "eval_cosine_precision@1": 0.668, |
| "eval_cosine_precision@10": 0.2962, |
| "eval_cosine_precision@3": 0.5318333333333333, |
| "eval_cosine_precision@5": 0.4289000000000001, |
| "eval_cosine_recall@1": 0.23326866628390464, |
| "eval_cosine_recall@10": 0.6563189809120494, |
| "eval_cosine_recall@3": 0.45311345100760037, |
| "eval_cosine_recall@5": 0.5465839152881188, |
| "eval_loss": 0.21039120852947235, |
| "eval_runtime": 277.7261, |
| "eval_samples_per_second": 7.201, |
| "eval_steps_per_second": 0.115, |
| "step": 316 |
| }, |
| { |
| "epoch": 0.5056, |
| "grad_norm": 3.2892227172851562, |
| "learning_rate": 2.759601706970128e-05, |
| "loss": 0.1912, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.5056, |
| "eval_cosine_accuracy@1": 0.6665, |
| "eval_cosine_accuracy@10": 0.9255, |
| "eval_cosine_accuracy@3": 0.8095, |
| "eval_cosine_accuracy@5": 0.8635, |
| "eval_cosine_map@100": 0.5806000470832352, |
| "eval_cosine_mrr@10": 0.7500138888888884, |
| "eval_cosine_ndcg@10": 0.6572566525532144, |
| "eval_cosine_precision@1": 0.6665, |
| "eval_cosine_precision@10": 0.30145000000000005, |
| "eval_cosine_precision@3": 0.534, |
| "eval_cosine_precision@5": 0.43610000000000004, |
| "eval_cosine_recall@1": 0.23175847262295138, |
| "eval_cosine_recall@10": 0.6703179486558828, |
| "eval_cosine_recall@3": 0.44944904156805865, |
| "eval_cosine_recall@5": 0.554136889607487, |
| "eval_loss": 0.16795697808265686, |
| "eval_runtime": 277.2687, |
| "eval_samples_per_second": 7.213, |
| "eval_steps_per_second": 0.115, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.60672, |
| "grad_norm": 2.89556622505188, |
| "learning_rate": 2.197724039829303e-05, |
| "loss": 0.1624, |
| "step": 474 |
| }, |
| { |
| "epoch": 0.60672, |
| "eval_cosine_accuracy@1": 0.6975, |
| "eval_cosine_accuracy@10": 0.9355, |
| "eval_cosine_accuracy@3": 0.8375, |
| "eval_cosine_accuracy@5": 0.887, |
| "eval_cosine_map@100": 0.6093090009951994, |
| "eval_cosine_mrr@10": 0.7765077380952375, |
| "eval_cosine_ndcg@10": 0.6851203967143871, |
| "eval_cosine_precision@1": 0.6975, |
| "eval_cosine_precision@10": 0.3129, |
| "eval_cosine_precision@3": 0.5613333333333332, |
| "eval_cosine_precision@5": 0.4528000000000001, |
| "eval_cosine_recall@1": 0.2445300283996011, |
| "eval_cosine_recall@10": 0.6928329252080592, |
| "eval_cosine_recall@3": 0.4812278629980086, |
| "eval_cosine_recall@5": 0.5788771761374806, |
| "eval_loss": 0.15388070046901703, |
| "eval_runtime": 276.9338, |
| "eval_samples_per_second": 7.222, |
| "eval_steps_per_second": 0.116, |
| "step": 474 |
| }, |
| { |
| "epoch": 0.70784, |
| "grad_norm": 1.8849042654037476, |
| "learning_rate": 1.6358463726884778e-05, |
| "loss": 0.144, |
| "step": 553 |
| }, |
| { |
| "epoch": 0.70784, |
| "eval_cosine_accuracy@1": 0.7125, |
| "eval_cosine_accuracy@10": 0.942, |
| "eval_cosine_accuracy@3": 0.8455, |
| "eval_cosine_accuracy@5": 0.8935, |
| "eval_cosine_map@100": 0.6286298958450689, |
| "eval_cosine_mrr@10": 0.7904275793650791, |
| "eval_cosine_ndcg@10": 0.7012467907285034, |
| "eval_cosine_precision@1": 0.7125, |
| "eval_cosine_precision@10": 0.31815, |
| "eval_cosine_precision@3": 0.5643333333333334, |
| "eval_cosine_precision@5": 0.46270000000000006, |
| "eval_cosine_recall@1": 0.2580733056498445, |
| "eval_cosine_recall@10": 0.7046707449914186, |
| "eval_cosine_recall@3": 0.4866446364412588, |
| "eval_cosine_recall@5": 0.589748134227783, |
| "eval_loss": 0.14380718767642975, |
| "eval_runtime": 276.988, |
| "eval_samples_per_second": 7.221, |
| "eval_steps_per_second": 0.116, |
| "step": 553 |
| }, |
| { |
| "epoch": 0.80896, |
| "grad_norm": 1.348805546760559, |
| "learning_rate": 1.073968705547653e-05, |
| "loss": 0.1189, |
| "step": 632 |
| }, |
| { |
| "epoch": 0.80896, |
| "eval_cosine_accuracy@1": 0.695, |
| "eval_cosine_accuracy@10": 0.947, |
| "eval_cosine_accuracy@3": 0.843, |
| "eval_cosine_accuracy@5": 0.896, |
| "eval_cosine_map@100": 0.6235293375316939, |
| "eval_cosine_mrr@10": 0.779907539682539, |
| "eval_cosine_ndcg@10": 0.69591191846525, |
| "eval_cosine_precision@1": 0.695, |
| "eval_cosine_precision@10": 0.31725, |
| "eval_cosine_precision@3": 0.5648333333333333, |
| "eval_cosine_precision@5": 0.463, |
| "eval_cosine_recall@1": 0.24691620939983008, |
| "eval_cosine_recall@10": 0.706279289199774, |
| "eval_cosine_recall@3": 0.48786519068806705, |
| "eval_cosine_recall@5": 0.5948458444043049, |
| "eval_loss": 0.12694863975048065, |
| "eval_runtime": 276.686, |
| "eval_samples_per_second": 7.228, |
| "eval_steps_per_second": 0.116, |
| "step": 632 |
| }, |
| { |
| "epoch": 0.91008, |
| "grad_norm": 1.5586862564086914, |
| "learning_rate": 5.120910384068279e-06, |
| "loss": 0.1048, |
| "step": 711 |
| }, |
| { |
| "epoch": 0.91008, |
| "eval_cosine_accuracy@1": 0.7235, |
| "eval_cosine_accuracy@10": 0.9485, |
| "eval_cosine_accuracy@3": 0.86, |
| "eval_cosine_accuracy@5": 0.901, |
| "eval_cosine_map@100": 0.6422912711695729, |
| "eval_cosine_mrr@10": 0.8008688492063484, |
| "eval_cosine_ndcg@10": 0.712959315512067, |
| "eval_cosine_precision@1": 0.7235, |
| "eval_cosine_precision@10": 0.32180000000000003, |
| "eval_cosine_precision@3": 0.5791666666666666, |
| "eval_cosine_precision@5": 0.4691, |
| "eval_cosine_recall@1": 0.2616417105499789, |
| "eval_cosine_recall@10": 0.715349774186521, |
| "eval_cosine_recall@3": 0.505012640408982, |
| "eval_cosine_recall@5": 0.6035736332324141, |
| "eval_loss": 0.11470390856266022, |
| "eval_runtime": 276.7545, |
| "eval_samples_per_second": 7.227, |
| "eval_steps_per_second": 0.116, |
| "step": 711 |
| } |
| ], |
| "logging_steps": 79, |
| "max_steps": 782, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 79, |
| "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": 64, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
