| { | |
| "best_metric": 0.3047935709180844, | |
| "best_model_checkpoint": "/content/drive/MyDrive/model_outputs/trial_3/checkpoint-584", | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 584, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.017123287671232876, | |
| "grad_norm": 11.778327941894531, | |
| "learning_rate": 9.4088732392351e-06, | |
| "loss": 2.2193, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03424657534246575, | |
| "grad_norm": 17.048717498779297, | |
| "learning_rate": 1.88177464784702e-05, | |
| "loss": 2.1721, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05136986301369863, | |
| "grad_norm": 9.865235328674316, | |
| "learning_rate": 2.1552795460563643e-05, | |
| "loss": 2.0945, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0684931506849315, | |
| "grad_norm": 8.911628723144531, | |
| "learning_rate": 2.142763404673923e-05, | |
| "loss": 2.1, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.08561643835616438, | |
| "grad_norm": 8.172422409057617, | |
| "learning_rate": 2.130247263291482e-05, | |
| "loss": 2.1, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.10273972602739725, | |
| "grad_norm": 8.123610496520996, | |
| "learning_rate": 2.117731121909041e-05, | |
| "loss": 2.0782, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11986301369863013, | |
| "grad_norm": 7.698946475982666, | |
| "learning_rate": 2.1052149805266e-05, | |
| "loss": 1.9786, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.136986301369863, | |
| "grad_norm": 8.443121910095215, | |
| "learning_rate": 2.0926988391441588e-05, | |
| "loss": 2.0303, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1541095890410959, | |
| "grad_norm": 8.679265975952148, | |
| "learning_rate": 2.0801826977617177e-05, | |
| "loss": 2.0897, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.17123287671232876, | |
| "grad_norm": 10.261368751525879, | |
| "learning_rate": 2.0676665563792766e-05, | |
| "loss": 2.0079, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18835616438356165, | |
| "grad_norm": 8.91393756866455, | |
| "learning_rate": 2.055150414996835e-05, | |
| "loss": 2.0663, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2054794520547945, | |
| "grad_norm": 10.901965141296387, | |
| "learning_rate": 2.042634273614394e-05, | |
| "loss": 1.9684, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2226027397260274, | |
| "grad_norm": 8.871338844299316, | |
| "learning_rate": 2.030118132231953e-05, | |
| "loss": 2.1153, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.23972602739726026, | |
| "grad_norm": 8.738993644714355, | |
| "learning_rate": 2.0176019908495118e-05, | |
| "loss": 2.0973, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.2568493150684932, | |
| "grad_norm": 8.016680717468262, | |
| "learning_rate": 2.0050858494670707e-05, | |
| "loss": 2.0179, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.273972602739726, | |
| "grad_norm": 7.566287994384766, | |
| "learning_rate": 1.9925697080846293e-05, | |
| "loss": 2.0837, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.2910958904109589, | |
| "grad_norm": 8.198686599731445, | |
| "learning_rate": 1.9800535667021882e-05, | |
| "loss": 2.0696, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.3082191780821918, | |
| "grad_norm": 7.653988838195801, | |
| "learning_rate": 1.967537425319747e-05, | |
| "loss": 1.9815, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3253424657534247, | |
| "grad_norm": 7.869149684906006, | |
| "learning_rate": 1.955021283937306e-05, | |
| "loss": 2.0139, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.3424657534246575, | |
| "grad_norm": 11.699186325073242, | |
| "learning_rate": 1.942505142554865e-05, | |
| "loss": 2.0046, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3595890410958904, | |
| "grad_norm": 7.273196220397949, | |
| "learning_rate": 1.9299890011724238e-05, | |
| "loss": 1.9229, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.3767123287671233, | |
| "grad_norm": 12.080012321472168, | |
| "learning_rate": 1.9174728597899827e-05, | |
| "loss": 2.0691, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.3938356164383562, | |
| "grad_norm": 10.047798156738281, | |
| "learning_rate": 1.9049567184075416e-05, | |
| "loss": 2.0008, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.410958904109589, | |
| "grad_norm": 12.511098861694336, | |
| "learning_rate": 1.8924405770251005e-05, | |
| "loss": 2.0886, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.4280821917808219, | |
| "grad_norm": 10.744943618774414, | |
| "learning_rate": 1.8799244356426594e-05, | |
| "loss": 1.9832, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.4452054794520548, | |
| "grad_norm": 10.715825080871582, | |
| "learning_rate": 1.8674082942602183e-05, | |
| "loss": 1.9835, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.4623287671232877, | |
| "grad_norm": 8.178912162780762, | |
| "learning_rate": 1.8548921528777768e-05, | |
| "loss": 2.0427, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.4794520547945205, | |
| "grad_norm": 14.410847663879395, | |
| "learning_rate": 1.8423760114953357e-05, | |
| "loss": 1.8612, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.4965753424657534, | |
| "grad_norm": 11.317428588867188, | |
| "learning_rate": 1.8298598701128946e-05, | |
| "loss": 1.9943, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.5136986301369864, | |
| "grad_norm": 8.62607192993164, | |
| "learning_rate": 1.8173437287304535e-05, | |
| "loss": 1.896, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.5308219178082192, | |
| "grad_norm": 8.02371883392334, | |
| "learning_rate": 1.8048275873480124e-05, | |
| "loss": 1.9874, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.547945205479452, | |
| "grad_norm": 9.2183198928833, | |
| "learning_rate": 1.7923114459655713e-05, | |
| "loss": 1.9973, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.565068493150685, | |
| "grad_norm": 10.167363166809082, | |
| "learning_rate": 1.7797953045831302e-05, | |
| "loss": 2.008, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.5821917808219178, | |
| "grad_norm": 8.94150161743164, | |
| "learning_rate": 1.767279163200689e-05, | |
| "loss": 1.9215, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.5993150684931506, | |
| "grad_norm": 12.496065139770508, | |
| "learning_rate": 1.754763021818248e-05, | |
| "loss": 1.9297, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.6164383561643836, | |
| "grad_norm": 8.680890083312988, | |
| "learning_rate": 1.742246880435807e-05, | |
| "loss": 1.8756, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.6335616438356164, | |
| "grad_norm": 12.159443855285645, | |
| "learning_rate": 1.7297307390533658e-05, | |
| "loss": 1.9632, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.6506849315068494, | |
| "grad_norm": 10.144058227539062, | |
| "learning_rate": 1.7172145976709244e-05, | |
| "loss": 1.8928, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.6678082191780822, | |
| "grad_norm": 9.923543930053711, | |
| "learning_rate": 1.7046984562884833e-05, | |
| "loss": 1.9076, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.684931506849315, | |
| "grad_norm": 11.46466064453125, | |
| "learning_rate": 1.6921823149060422e-05, | |
| "loss": 1.92, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.702054794520548, | |
| "grad_norm": 9.840792655944824, | |
| "learning_rate": 1.679666173523601e-05, | |
| "loss": 1.8128, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.7191780821917808, | |
| "grad_norm": 10.992412567138672, | |
| "learning_rate": 1.66715003214116e-05, | |
| "loss": 1.8548, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.7363013698630136, | |
| "grad_norm": 10.46108627319336, | |
| "learning_rate": 1.6546338907587185e-05, | |
| "loss": 1.8455, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.7534246575342466, | |
| "grad_norm": 13.512310981750488, | |
| "learning_rate": 1.6421177493762774e-05, | |
| "loss": 1.9271, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.7705479452054794, | |
| "grad_norm": 10.795140266418457, | |
| "learning_rate": 1.6296016079938363e-05, | |
| "loss": 1.9033, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.7876712328767124, | |
| "grad_norm": 11.524979591369629, | |
| "learning_rate": 1.6170854666113952e-05, | |
| "loss": 1.9248, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.8047945205479452, | |
| "grad_norm": 8.86741828918457, | |
| "learning_rate": 1.604569325228954e-05, | |
| "loss": 1.9549, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.821917808219178, | |
| "grad_norm": 12.26812744140625, | |
| "learning_rate": 1.592053183846513e-05, | |
| "loss": 1.9197, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.839041095890411, | |
| "grad_norm": 17.214059829711914, | |
| "learning_rate": 1.579537042464072e-05, | |
| "loss": 1.9561, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.8561643835616438, | |
| "grad_norm": 13.144837379455566, | |
| "learning_rate": 1.5670209010816308e-05, | |
| "loss": 1.9064, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.8732876712328768, | |
| "grad_norm": 10.606916427612305, | |
| "learning_rate": 1.5545047596991897e-05, | |
| "loss": 1.8398, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.8904109589041096, | |
| "grad_norm": 11.539923667907715, | |
| "learning_rate": 1.5419886183167483e-05, | |
| "loss": 1.9345, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.9075342465753424, | |
| "grad_norm": 14.816058158874512, | |
| "learning_rate": 1.5294724769343072e-05, | |
| "loss": 1.8451, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.9246575342465754, | |
| "grad_norm": 12.361188888549805, | |
| "learning_rate": 1.5169563355518661e-05, | |
| "loss": 1.7723, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.9417808219178082, | |
| "grad_norm": 13.830178260803223, | |
| "learning_rate": 1.504440194169425e-05, | |
| "loss": 1.9174, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.958904109589041, | |
| "grad_norm": 11.410951614379883, | |
| "learning_rate": 1.4919240527869839e-05, | |
| "loss": 1.8714, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.976027397260274, | |
| "grad_norm": 11.380172729492188, | |
| "learning_rate": 1.4794079114045428e-05, | |
| "loss": 1.8259, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.9931506849315068, | |
| "grad_norm": 17.25227165222168, | |
| "learning_rate": 1.4668917700221017e-05, | |
| "loss": 1.8291, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_classification_report": { | |
| "accuracy": 0.3175, | |
| "ar": { | |
| "f1-score": 0.2611111111111111, | |
| "precision": 0.3051948051948052, | |
| "recall": 0.22815533980582525, | |
| "support": 206.0 | |
| }, | |
| "cl": { | |
| "f1-score": 0.2444113263785395, | |
| "precision": 0.2152230971128609, | |
| "recall": 0.2827586206896552, | |
| "support": 290.0 | |
| }, | |
| "co": { | |
| "f1-score": 0.35724331926863573, | |
| "precision": 0.30238095238095236, | |
| "recall": 0.436426116838488, | |
| "support": 291.0 | |
| }, | |
| "es": { | |
| "f1-score": 0.32696390658174096, | |
| "precision": 0.4010416666666667, | |
| "recall": 0.27598566308243727, | |
| "support": 279.0 | |
| }, | |
| "macro avg": { | |
| "f1-score": 0.3047935709180844, | |
| "precision": 0.3345601053365995, | |
| "recall": 0.29767349498848117, | |
| "support": 2000.0 | |
| }, | |
| "mx": { | |
| "f1-score": 0.3294663573085847, | |
| "precision": 0.5071428571428571, | |
| "recall": 0.24398625429553264, | |
| "support": 291.0 | |
| }, | |
| "pe": { | |
| "f1-score": 0.325434439178515, | |
| "precision": 0.30116959064327486, | |
| "recall": 0.3539518900343643, | |
| "support": 291.0 | |
| }, | |
| "pr": { | |
| "f1-score": 0.6171428571428571, | |
| "precision": 0.7297297297297297, | |
| "recall": 0.5346534653465347, | |
| "support": 101.0 | |
| }, | |
| "uy": { | |
| "f1-score": 0.2813688212927757, | |
| "precision": 0.24915824915824916, | |
| "recall": 0.3231441048034934, | |
| "support": 229.0 | |
| }, | |
| "ve": { | |
| "f1-score": 0.0, | |
| "precision": 0.0, | |
| "recall": 0.0, | |
| "support": 22.0 | |
| }, | |
| "weighted avg": { | |
| "f1-score": 0.3185949649036821, | |
| "precision": 0.3455735871207114, | |
| "recall": 0.3175, | |
| "support": 2000.0 | |
| } | |
| }, | |
| "eval_f1": 0.3047935709180844, | |
| "eval_loss": 1.8013501167297363, | |
| "eval_runtime": 4.4012, | |
| "eval_samples_per_second": 454.419, | |
| "eval_steps_per_second": 56.802, | |
| "step": 584 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1752, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "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": 306938335993344.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |