| { | |
| "best_metric": 0.33959150175639824, | |
| "best_model_checkpoint": "/content/drive/MyDrive/model_outputs/trial_6/checkpoint-1168", | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 1168, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.017123287671232876, | |
| "grad_norm": 10.533243179321289, | |
| "learning_rate": 2.2368737828601498e-06, | |
| "loss": 2.2235, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03424657534246575, | |
| "grad_norm": 13.400142669677734, | |
| "learning_rate": 4.4737475657202996e-06, | |
| "loss": 2.203, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05136986301369863, | |
| "grad_norm": 11.283329010009766, | |
| "learning_rate": 6.71062134858045e-06, | |
| "loss": 2.202, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0684931506849315, | |
| "grad_norm": 9.742809295654297, | |
| "learning_rate": 8.947495131440599e-06, | |
| "loss": 2.1295, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.08561643835616438, | |
| "grad_norm": 9.277532577514648, | |
| "learning_rate": 1.1184368914300748e-05, | |
| "loss": 2.1058, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.10273972602739725, | |
| "grad_norm": 9.417570114135742, | |
| "learning_rate": 1.34212426971609e-05, | |
| "loss": 2.0999, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11986301369863013, | |
| "grad_norm": 8.50729751586914, | |
| "learning_rate": 1.5658116480021048e-05, | |
| "loss": 2.0348, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.136986301369863, | |
| "grad_norm": 11.117788314819336, | |
| "learning_rate": 1.7894990262881198e-05, | |
| "loss": 2.0573, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1541095890410959, | |
| "grad_norm": 8.794454574584961, | |
| "learning_rate": 2.013186404574135e-05, | |
| "loss": 2.1162, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.17123287671232876, | |
| "grad_norm": 10.189006805419922, | |
| "learning_rate": 2.2368737828601496e-05, | |
| "loss": 2.0447, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18835616438356165, | |
| "grad_norm": 8.304252624511719, | |
| "learning_rate": 2.4605611611461647e-05, | |
| "loss": 2.0826, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2054794520547945, | |
| "grad_norm": 11.60432243347168, | |
| "learning_rate": 2.5884293984157745e-05, | |
| "loss": 2.0481, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2226027397260274, | |
| "grad_norm": 11.781548500061035, | |
| "learning_rate": 2.5725689241607756e-05, | |
| "loss": 2.1976, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.23972602739726026, | |
| "grad_norm": 8.521300315856934, | |
| "learning_rate": 2.5567084499057773e-05, | |
| "loss": 2.1029, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.2568493150684932, | |
| "grad_norm": 9.029247283935547, | |
| "learning_rate": 2.5408479756507783e-05, | |
| "loss": 2.0236, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.273972602739726, | |
| "grad_norm": 8.05414867401123, | |
| "learning_rate": 2.52498750139578e-05, | |
| "loss": 2.1098, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.2910958904109589, | |
| "grad_norm": 8.391693115234375, | |
| "learning_rate": 2.509127027140781e-05, | |
| "loss": 2.0854, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.3082191780821918, | |
| "grad_norm": 8.5321044921875, | |
| "learning_rate": 2.4932665528857828e-05, | |
| "loss": 2.0164, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3253424657534247, | |
| "grad_norm": 8.022624969482422, | |
| "learning_rate": 2.477406078630784e-05, | |
| "loss": 2.0136, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.3424657534246575, | |
| "grad_norm": 16.765403747558594, | |
| "learning_rate": 2.4615456043757856e-05, | |
| "loss": 2.0038, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3595890410958904, | |
| "grad_norm": 8.68701171875, | |
| "learning_rate": 2.4456851301207866e-05, | |
| "loss": 1.9602, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.3767123287671233, | |
| "grad_norm": 8.106729507446289, | |
| "learning_rate": 2.4298246558657883e-05, | |
| "loss": 2.0734, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.3938356164383562, | |
| "grad_norm": 9.488205909729004, | |
| "learning_rate": 2.4139641816107894e-05, | |
| "loss": 2.0728, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.410958904109589, | |
| "grad_norm": 10.983114242553711, | |
| "learning_rate": 2.398103707355791e-05, | |
| "loss": 2.0433, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.4280821917808219, | |
| "grad_norm": 10.720239639282227, | |
| "learning_rate": 2.382243233100792e-05, | |
| "loss": 1.9801, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.4452054794520548, | |
| "grad_norm": 9.009136199951172, | |
| "learning_rate": 2.3663827588457938e-05, | |
| "loss": 2.0108, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.4623287671232877, | |
| "grad_norm": 8.843775749206543, | |
| "learning_rate": 2.350522284590795e-05, | |
| "loss": 2.072, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.4794520547945205, | |
| "grad_norm": 10.325736999511719, | |
| "learning_rate": 2.3346618103357966e-05, | |
| "loss": 1.8827, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.4965753424657534, | |
| "grad_norm": 12.433212280273438, | |
| "learning_rate": 2.318801336080798e-05, | |
| "loss": 2.0127, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.5136986301369864, | |
| "grad_norm": 8.865859985351562, | |
| "learning_rate": 2.3029408618257993e-05, | |
| "loss": 1.92, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.5308219178082192, | |
| "grad_norm": 8.225539207458496, | |
| "learning_rate": 2.2870803875708007e-05, | |
| "loss": 2.0065, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.547945205479452, | |
| "grad_norm": 9.395380973815918, | |
| "learning_rate": 2.271219913315802e-05, | |
| "loss": 2.018, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.565068493150685, | |
| "grad_norm": 11.355223655700684, | |
| "learning_rate": 2.2553594390608035e-05, | |
| "loss": 2.0069, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.5821917808219178, | |
| "grad_norm": 9.485352516174316, | |
| "learning_rate": 2.239498964805805e-05, | |
| "loss": 1.8434, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.5993150684931506, | |
| "grad_norm": 9.75179386138916, | |
| "learning_rate": 2.2236384905508062e-05, | |
| "loss": 1.9366, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.6164383561643836, | |
| "grad_norm": 8.961932182312012, | |
| "learning_rate": 2.2077780162958076e-05, | |
| "loss": 1.8823, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.6335616438356164, | |
| "grad_norm": 11.1987886428833, | |
| "learning_rate": 2.191917542040809e-05, | |
| "loss": 2.0209, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.6506849315068494, | |
| "grad_norm": 9.715045928955078, | |
| "learning_rate": 2.1760570677858104e-05, | |
| "loss": 2.001, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.6678082191780822, | |
| "grad_norm": 12.430977821350098, | |
| "learning_rate": 2.1601965935308117e-05, | |
| "loss": 1.9342, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.684931506849315, | |
| "grad_norm": 13.551401138305664, | |
| "learning_rate": 2.144336119275813e-05, | |
| "loss": 1.9697, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.702054794520548, | |
| "grad_norm": 10.96045970916748, | |
| "learning_rate": 2.1284756450208145e-05, | |
| "loss": 1.8875, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.7191780821917808, | |
| "grad_norm": 10.306199073791504, | |
| "learning_rate": 2.112615170765816e-05, | |
| "loss": 1.9113, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.7363013698630136, | |
| "grad_norm": 8.820138931274414, | |
| "learning_rate": 2.0967546965108172e-05, | |
| "loss": 1.8304, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.7534246575342466, | |
| "grad_norm": 11.025798797607422, | |
| "learning_rate": 2.0808942222558186e-05, | |
| "loss": 1.9428, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.7705479452054794, | |
| "grad_norm": 10.171765327453613, | |
| "learning_rate": 2.06503374800082e-05, | |
| "loss": 1.9606, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.7876712328767124, | |
| "grad_norm": 12.033662796020508, | |
| "learning_rate": 2.0491732737458214e-05, | |
| "loss": 1.9102, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.8047945205479452, | |
| "grad_norm": 10.383848190307617, | |
| "learning_rate": 2.0333127994908228e-05, | |
| "loss": 1.9791, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.821917808219178, | |
| "grad_norm": 12.919981002807617, | |
| "learning_rate": 2.017452325235824e-05, | |
| "loss": 1.9671, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.839041095890411, | |
| "grad_norm": 12.617621421813965, | |
| "learning_rate": 2.0015918509808255e-05, | |
| "loss": 1.9002, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.8561643835616438, | |
| "grad_norm": 10.720662117004395, | |
| "learning_rate": 1.985731376725827e-05, | |
| "loss": 1.925, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.8732876712328768, | |
| "grad_norm": 10.810115814208984, | |
| "learning_rate": 1.9698709024708283e-05, | |
| "loss": 1.8476, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.8904109589041096, | |
| "grad_norm": 11.600144386291504, | |
| "learning_rate": 1.9540104282158296e-05, | |
| "loss": 2.0233, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.9075342465753424, | |
| "grad_norm": 19.464778900146484, | |
| "learning_rate": 1.938149953960831e-05, | |
| "loss": 1.8731, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.9246575342465754, | |
| "grad_norm": 14.501326560974121, | |
| "learning_rate": 1.9222894797058324e-05, | |
| "loss": 1.8296, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.9417808219178082, | |
| "grad_norm": 12.78044605255127, | |
| "learning_rate": 1.9064290054508338e-05, | |
| "loss": 1.9029, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.958904109589041, | |
| "grad_norm": 11.88862133026123, | |
| "learning_rate": 1.890568531195835e-05, | |
| "loss": 1.9337, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.976027397260274, | |
| "grad_norm": 12.300003051757812, | |
| "learning_rate": 1.8747080569408365e-05, | |
| "loss": 1.8796, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.9931506849315068, | |
| "grad_norm": 15.312654495239258, | |
| "learning_rate": 1.858847582685838e-05, | |
| "loss": 1.8259, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_classification_report": { | |
| "accuracy": 0.3195, | |
| "ar": { | |
| "f1-score": 0.23384615384615384, | |
| "precision": 0.31932773109243695, | |
| "recall": 0.18446601941747573, | |
| "support": 206.0 | |
| }, | |
| "cl": { | |
| "f1-score": 0.25039123630672927, | |
| "precision": 0.22922636103151864, | |
| "recall": 0.27586206896551724, | |
| "support": 290.0 | |
| }, | |
| "co": { | |
| "f1-score": 0.37142857142857144, | |
| "precision": 0.2985386221294363, | |
| "recall": 0.49140893470790376, | |
| "support": 291.0 | |
| }, | |
| "es": { | |
| "f1-score": 0.36607142857142855, | |
| "precision": 0.48520710059171596, | |
| "recall": 0.2939068100358423, | |
| "support": 279.0 | |
| }, | |
| "macro avg": { | |
| "f1-score": 0.30612372170181307, | |
| "precision": 0.3459220621957859, | |
| "recall": 0.29992450825787437, | |
| "support": 2000.0 | |
| }, | |
| "mx": { | |
| "f1-score": 0.3076923076923077, | |
| "precision": 0.4782608695652174, | |
| "recall": 0.2268041237113402, | |
| "support": 291.0 | |
| }, | |
| "pe": { | |
| "f1-score": 0.3013698630136986, | |
| "precision": 0.3003412969283277, | |
| "recall": 0.3024054982817869, | |
| "support": 291.0 | |
| }, | |
| "pr": { | |
| "f1-score": 0.6395348837209303, | |
| "precision": 0.7746478873239436, | |
| "recall": 0.5445544554455446, | |
| "support": 101.0 | |
| }, | |
| "uy": { | |
| "f1-score": 0.28477905073649756, | |
| "precision": 0.22774869109947643, | |
| "recall": 0.3799126637554585, | |
| "support": 229.0 | |
| }, | |
| "ve": { | |
| "f1-score": 0.0, | |
| "precision": 0.0, | |
| "recall": 0.0, | |
| "support": 22.0 | |
| }, | |
| "weighted avg": { | |
| "f1-score": 0.3190249633141609, | |
| "precision": 0.3557358973700286, | |
| "recall": 0.3195, | |
| "support": 2000.0 | |
| } | |
| }, | |
| "eval_f1": 0.30612372170181307, | |
| "eval_loss": 1.8373411893844604, | |
| "eval_runtime": 2.9167, | |
| "eval_samples_per_second": 685.713, | |
| "eval_steps_per_second": 85.714, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.0102739726027397, | |
| "grad_norm": 9.993717193603516, | |
| "learning_rate": 1.8429871084308393e-05, | |
| "loss": 1.7138, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.0273972602739727, | |
| "grad_norm": 16.997873306274414, | |
| "learning_rate": 1.8271266341758407e-05, | |
| "loss": 1.6405, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.0445205479452055, | |
| "grad_norm": 14.189482688903809, | |
| "learning_rate": 1.811266159920842e-05, | |
| "loss": 1.6928, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.0616438356164384, | |
| "grad_norm": 11.937070846557617, | |
| "learning_rate": 1.7954056856658434e-05, | |
| "loss": 1.606, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.0787671232876712, | |
| "grad_norm": 15.003684997558594, | |
| "learning_rate": 1.7795452114108448e-05, | |
| "loss": 1.532, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.095890410958904, | |
| "grad_norm": 12.880982398986816, | |
| "learning_rate": 1.7636847371558462e-05, | |
| "loss": 1.6335, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.1130136986301369, | |
| "grad_norm": 27.958925247192383, | |
| "learning_rate": 1.7478242629008476e-05, | |
| "loss": 1.7617, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.13013698630137, | |
| "grad_norm": 20.256528854370117, | |
| "learning_rate": 1.731963788645849e-05, | |
| "loss": 1.6595, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.1472602739726028, | |
| "grad_norm": 11.43181037902832, | |
| "learning_rate": 1.7161033143908503e-05, | |
| "loss": 1.4586, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.1643835616438356, | |
| "grad_norm": 16.205936431884766, | |
| "learning_rate": 1.7002428401358517e-05, | |
| "loss": 1.6602, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.1815068493150684, | |
| "grad_norm": 10.263736724853516, | |
| "learning_rate": 1.684382365880853e-05, | |
| "loss": 1.614, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.1986301369863013, | |
| "grad_norm": 15.32834529876709, | |
| "learning_rate": 1.6685218916258544e-05, | |
| "loss": 1.6591, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.2157534246575343, | |
| "grad_norm": 19.10625648498535, | |
| "learning_rate": 1.6526614173708558e-05, | |
| "loss": 1.5541, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.2328767123287672, | |
| "grad_norm": 16.795921325683594, | |
| "learning_rate": 1.6368009431158572e-05, | |
| "loss": 1.6756, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 15.281452178955078, | |
| "learning_rate": 1.6209404688608586e-05, | |
| "loss": 1.6431, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.2671232876712328, | |
| "grad_norm": 20.04804039001465, | |
| "learning_rate": 1.60507999460586e-05, | |
| "loss": 1.6474, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.2842465753424657, | |
| "grad_norm": 14.96975040435791, | |
| "learning_rate": 1.5892195203508613e-05, | |
| "loss": 1.696, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.3013698630136985, | |
| "grad_norm": 24.126489639282227, | |
| "learning_rate": 1.5733590460958627e-05, | |
| "loss": 1.7574, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.3184931506849316, | |
| "grad_norm": 18.405784606933594, | |
| "learning_rate": 1.557498571840864e-05, | |
| "loss": 1.4954, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.3356164383561644, | |
| "grad_norm": 14.699495315551758, | |
| "learning_rate": 1.5416380975858658e-05, | |
| "loss": 1.5321, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.3527397260273972, | |
| "grad_norm": 31.16180419921875, | |
| "learning_rate": 1.5257776233308668e-05, | |
| "loss": 1.6808, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.36986301369863, | |
| "grad_norm": 18.2718448638916, | |
| "learning_rate": 1.5099171490758684e-05, | |
| "loss": 1.5726, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.3869863013698631, | |
| "grad_norm": 17.004480361938477, | |
| "learning_rate": 1.4940566748208696e-05, | |
| "loss": 1.5544, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.404109589041096, | |
| "grad_norm": 21.022958755493164, | |
| "learning_rate": 1.4781962005658711e-05, | |
| "loss": 1.674, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.4212328767123288, | |
| "grad_norm": 15.981225967407227, | |
| "learning_rate": 1.4623357263108724e-05, | |
| "loss": 1.7799, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.4383561643835616, | |
| "grad_norm": 21.790754318237305, | |
| "learning_rate": 1.4464752520558739e-05, | |
| "loss": 1.6517, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.4554794520547945, | |
| "grad_norm": 28.04483413696289, | |
| "learning_rate": 1.4306147778008751e-05, | |
| "loss": 1.5257, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.4726027397260273, | |
| "grad_norm": 11.142111778259277, | |
| "learning_rate": 1.4147543035458767e-05, | |
| "loss": 1.692, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.4897260273972603, | |
| "grad_norm": 18.453630447387695, | |
| "learning_rate": 1.3988938292908779e-05, | |
| "loss": 1.7486, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.5068493150684932, | |
| "grad_norm": 18.69659423828125, | |
| "learning_rate": 1.3830333550358794e-05, | |
| "loss": 1.5123, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.523972602739726, | |
| "grad_norm": 17.997787475585938, | |
| "learning_rate": 1.3671728807808806e-05, | |
| "loss": 1.5542, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.541095890410959, | |
| "grad_norm": 23.682682037353516, | |
| "learning_rate": 1.3513124065258822e-05, | |
| "loss": 1.5797, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.558219178082192, | |
| "grad_norm": 15.725004196166992, | |
| "learning_rate": 1.3354519322708834e-05, | |
| "loss": 1.4599, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.5753424657534247, | |
| "grad_norm": 27.747419357299805, | |
| "learning_rate": 1.319591458015885e-05, | |
| "loss": 1.575, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.5924657534246576, | |
| "grad_norm": 24.641860961914062, | |
| "learning_rate": 1.3037309837608863e-05, | |
| "loss": 1.5311, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.6095890410958904, | |
| "grad_norm": 21.082124710083008, | |
| "learning_rate": 1.2878705095058877e-05, | |
| "loss": 1.6095, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.6267123287671232, | |
| "grad_norm": 20.251115798950195, | |
| "learning_rate": 1.272010035250889e-05, | |
| "loss": 1.4753, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.643835616438356, | |
| "grad_norm": 28.846712112426758, | |
| "learning_rate": 1.2561495609958904e-05, | |
| "loss": 1.5826, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.660958904109589, | |
| "grad_norm": 16.3935489654541, | |
| "learning_rate": 1.2402890867408918e-05, | |
| "loss": 1.5531, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.678082191780822, | |
| "grad_norm": 26.24190330505371, | |
| "learning_rate": 1.2244286124858932e-05, | |
| "loss": 1.6214, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.6952054794520548, | |
| "grad_norm": 25.269094467163086, | |
| "learning_rate": 1.2085681382308946e-05, | |
| "loss": 1.4609, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.7123287671232876, | |
| "grad_norm": 27.003097534179688, | |
| "learning_rate": 1.1927076639758961e-05, | |
| "loss": 1.5031, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.7294520547945207, | |
| "grad_norm": 24.793109893798828, | |
| "learning_rate": 1.1768471897208975e-05, | |
| "loss": 1.5054, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.7465753424657535, | |
| "grad_norm": 19.835695266723633, | |
| "learning_rate": 1.1609867154658989e-05, | |
| "loss": 1.5893, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.7636986301369864, | |
| "grad_norm": 20.677410125732422, | |
| "learning_rate": 1.1451262412109003e-05, | |
| "loss": 1.4415, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.7808219178082192, | |
| "grad_norm": 12.999378204345703, | |
| "learning_rate": 1.1292657669559016e-05, | |
| "loss": 1.6227, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.797945205479452, | |
| "grad_norm": 24.370025634765625, | |
| "learning_rate": 1.113405292700903e-05, | |
| "loss": 1.6393, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.8150684931506849, | |
| "grad_norm": 21.418380737304688, | |
| "learning_rate": 1.0975448184459044e-05, | |
| "loss": 1.4325, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.8321917808219177, | |
| "grad_norm": 16.800621032714844, | |
| "learning_rate": 1.0816843441909058e-05, | |
| "loss": 1.5552, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.8493150684931505, | |
| "grad_norm": 25.122791290283203, | |
| "learning_rate": 1.0658238699359071e-05, | |
| "loss": 1.4473, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.8664383561643836, | |
| "grad_norm": 30.148611068725586, | |
| "learning_rate": 1.0499633956809085e-05, | |
| "loss": 1.4344, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.8835616438356164, | |
| "grad_norm": 23.485918045043945, | |
| "learning_rate": 1.0341029214259099e-05, | |
| "loss": 1.5767, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.9006849315068495, | |
| "grad_norm": 17.657108306884766, | |
| "learning_rate": 1.0182424471709113e-05, | |
| "loss": 1.5423, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.9178082191780823, | |
| "grad_norm": 22.166349411010742, | |
| "learning_rate": 1.0023819729159127e-05, | |
| "loss": 1.5175, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.9349315068493151, | |
| "grad_norm": 15.70091438293457, | |
| "learning_rate": 9.86521498660914e-06, | |
| "loss": 1.335, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.952054794520548, | |
| "grad_norm": 17.611722946166992, | |
| "learning_rate": 9.706610244059154e-06, | |
| "loss": 1.3978, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.9691780821917808, | |
| "grad_norm": 22.696762084960938, | |
| "learning_rate": 9.548005501509168e-06, | |
| "loss": 1.5925, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.9863013698630136, | |
| "grad_norm": 19.230512619018555, | |
| "learning_rate": 9.389400758959182e-06, | |
| "loss": 1.4233, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_classification_report": { | |
| "accuracy": 0.361, | |
| "ar": { | |
| "f1-score": 0.32941176470588235, | |
| "precision": 0.417910447761194, | |
| "recall": 0.27184466019417475, | |
| "support": 206.0 | |
| }, | |
| "cl": { | |
| "f1-score": 0.25830258302583026, | |
| "precision": 0.2777777777777778, | |
| "recall": 0.2413793103448276, | |
| "support": 290.0 | |
| }, | |
| "co": { | |
| "f1-score": 0.3806970509383378, | |
| "precision": 0.3120879120879121, | |
| "recall": 0.4879725085910653, | |
| "support": 291.0 | |
| }, | |
| "es": { | |
| "f1-score": 0.3992467043314501, | |
| "precision": 0.42063492063492064, | |
| "recall": 0.37992831541218636, | |
| "support": 279.0 | |
| }, | |
| "macro avg": { | |
| "f1-score": 0.33959150175639824, | |
| "precision": 0.34685147408513256, | |
| "recall": 0.34303341981862456, | |
| "support": 2000.0 | |
| }, | |
| "mx": { | |
| "f1-score": 0.3804143126177024, | |
| "precision": 0.42083333333333334, | |
| "recall": 0.3470790378006873, | |
| "support": 291.0 | |
| }, | |
| "pe": { | |
| "f1-score": 0.3188854489164087, | |
| "precision": 0.29014084507042254, | |
| "recall": 0.3539518900343643, | |
| "support": 291.0 | |
| }, | |
| "pr": { | |
| "f1-score": 0.6325581395348837, | |
| "precision": 0.5964912280701754, | |
| "recall": 0.6732673267326733, | |
| "support": 101.0 | |
| }, | |
| "uy": { | |
| "f1-score": 0.3568075117370892, | |
| "precision": 0.38578680203045684, | |
| "recall": 0.3318777292576419, | |
| "support": 229.0 | |
| }, | |
| "ve": { | |
| "f1-score": 0.0, | |
| "precision": 0.0, | |
| "recall": 0.0, | |
| "support": 22.0 | |
| }, | |
| "weighted avg": { | |
| "f1-score": 0.3570163839128382, | |
| "precision": 0.3651520553423211, | |
| "recall": 0.361, | |
| "support": 2000.0 | |
| } | |
| }, | |
| "eval_f1": 0.33959150175639824, | |
| "eval_loss": 1.7848442792892456, | |
| "eval_runtime": 2.9769, | |
| "eval_samples_per_second": 671.834, | |
| "eval_steps_per_second": 83.979, | |
| "step": 1168 | |
| } | |
| ], | |
| "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 | |
| } | |