| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.9963369963369964, |
| "eval_steps": 100, |
| "global_step": 1227, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.02442002442002442, |
| "grad_norm": 10.498289108276367, |
| "learning_rate": 4.0000000000000003e-07, |
| "loss": 0.7042, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.04884004884004884, |
| "grad_norm": 9.02344799041748, |
| "learning_rate": 8.000000000000001e-07, |
| "loss": 0.7274, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.07326007326007326, |
| "grad_norm": 12.760648727416992, |
| "learning_rate": 1.2000000000000002e-06, |
| "loss": 0.7136, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.09768009768009768, |
| "grad_norm": 11.814812660217285, |
| "learning_rate": 1.6000000000000001e-06, |
| "loss": 0.7063, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.1221001221001221, |
| "grad_norm": 10.731100082397461, |
| "learning_rate": 2.0000000000000003e-06, |
| "loss": 0.6947, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.14652014652014653, |
| "grad_norm": 9.641815185546875, |
| "learning_rate": 2.4000000000000003e-06, |
| "loss": 0.6883, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.17094017094017094, |
| "grad_norm": 10.62743854522705, |
| "learning_rate": 2.8000000000000003e-06, |
| "loss": 0.6673, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.19536019536019536, |
| "grad_norm": 15.131406784057617, |
| "learning_rate": 3.2000000000000003e-06, |
| "loss": 0.6422, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.21978021978021978, |
| "grad_norm": 11.530914306640625, |
| "learning_rate": 3.6000000000000003e-06, |
| "loss": 0.6155, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.2442002442002442, |
| "grad_norm": 9.192888259887695, |
| "learning_rate": 4.000000000000001e-06, |
| "loss": 0.5848, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.2442002442002442, |
| "eval_accuracy": 0.7782529016493586, |
| "eval_accuracy_label_negative": 0.8547979797979798, |
| "eval_accuracy_label_positive": 0.7065088757396449, |
| "eval_f1": 0.7773703489129234, |
| "eval_loss": 0.5668291449546814, |
| "eval_precision": 0.7869130340900089, |
| "eval_recall": 0.7782529016493586, |
| "eval_runtime": 6.1736, |
| "eval_samples_per_second": 265.16, |
| "eval_steps_per_second": 16.684, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.2686202686202686, |
| "grad_norm": 11.79243278503418, |
| "learning_rate": 4.4e-06, |
| "loss": 0.5391, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.29304029304029305, |
| "grad_norm": 13.897492408752441, |
| "learning_rate": 4.800000000000001e-06, |
| "loss": 0.4971, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.31746031746031744, |
| "grad_norm": 18.756803512573242, |
| "learning_rate": 5.2e-06, |
| "loss": 0.4148, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.3418803418803419, |
| "grad_norm": 18.242826461791992, |
| "learning_rate": 5.600000000000001e-06, |
| "loss": 0.3547, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.3663003663003663, |
| "grad_norm": 7.90754508972168, |
| "learning_rate": 6e-06, |
| "loss": 0.3547, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.3907203907203907, |
| "grad_norm": 11.013474464416504, |
| "learning_rate": 6.4000000000000006e-06, |
| "loss": 0.309, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.41514041514041516, |
| "grad_norm": 24.09418487548828, |
| "learning_rate": 6.800000000000001e-06, |
| "loss": 0.3168, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.43956043956043955, |
| "grad_norm": 12.913431167602539, |
| "learning_rate": 7.2000000000000005e-06, |
| "loss": 0.2891, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.463980463980464, |
| "grad_norm": 38.170555114746094, |
| "learning_rate": 7.600000000000001e-06, |
| "loss": 0.277, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.4884004884004884, |
| "grad_norm": 19.653949737548828, |
| "learning_rate": 8.000000000000001e-06, |
| "loss": 0.2761, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.4884004884004884, |
| "eval_accuracy": 0.8912645082467929, |
| "eval_accuracy_label_negative": 0.9318181818181818, |
| "eval_accuracy_label_positive": 0.8532544378698225, |
| "eval_f1": 0.8912162170726895, |
| "eval_loss": 0.2858136296272278, |
| "eval_precision": 0.8944362774670781, |
| "eval_recall": 0.8912645082467929, |
| "eval_runtime": 6.345, |
| "eval_samples_per_second": 257.999, |
| "eval_steps_per_second": 16.233, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.5128205128205128, |
| "grad_norm": 38.47966003417969, |
| "learning_rate": 8.400000000000001e-06, |
| "loss": 0.228, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.5372405372405372, |
| "grad_norm": 37.67184829711914, |
| "learning_rate": 8.8e-06, |
| "loss": 0.204, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.5616605616605617, |
| "grad_norm": 9.530702590942383, |
| "learning_rate": 9.200000000000002e-06, |
| "loss": 0.2728, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.5860805860805861, |
| "grad_norm": 24.261260986328125, |
| "learning_rate": 9.600000000000001e-06, |
| "loss": 0.2755, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.6105006105006106, |
| "grad_norm": 25.7033634185791, |
| "learning_rate": 1e-05, |
| "loss": 0.2787, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.6349206349206349, |
| "grad_norm": 21.64457893371582, |
| "learning_rate": 1.04e-05, |
| "loss": 0.2357, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.6593406593406593, |
| "grad_norm": 23.77570915222168, |
| "learning_rate": 1.0800000000000002e-05, |
| "loss": 0.2388, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.6837606837606838, |
| "grad_norm": 20.548864364624023, |
| "learning_rate": 1.1200000000000001e-05, |
| "loss": 0.2058, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.7081807081807082, |
| "grad_norm": 14.279947280883789, |
| "learning_rate": 1.16e-05, |
| "loss": 0.2169, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.7326007326007326, |
| "grad_norm": 20.323959350585938, |
| "learning_rate": 1.2e-05, |
| "loss": 0.2099, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.7326007326007326, |
| "eval_accuracy": 0.9114233353695785, |
| "eval_accuracy_label_negative": 0.8964646464646465, |
| "eval_accuracy_label_positive": 0.9254437869822485, |
| "eval_f1": 0.9113780300690499, |
| "eval_loss": 0.2412199079990387, |
| "eval_precision": 0.9115789422424927, |
| "eval_recall": 0.9114233353695785, |
| "eval_runtime": 6.209, |
| "eval_samples_per_second": 263.65, |
| "eval_steps_per_second": 16.589, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.757020757020757, |
| "grad_norm": 13.649131774902344, |
| "learning_rate": 1.2400000000000002e-05, |
| "loss": 0.2337, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.7814407814407814, |
| "grad_norm": 18.731950759887695, |
| "learning_rate": 1.2800000000000001e-05, |
| "loss": 0.202, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.8058608058608059, |
| "grad_norm": 2.1580898761749268, |
| "learning_rate": 1.3200000000000002e-05, |
| "loss": 0.1882, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.8302808302808303, |
| "grad_norm": 17.960113525390625, |
| "learning_rate": 1.3600000000000002e-05, |
| "loss": 0.2176, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.8547008547008547, |
| "grad_norm": 19.708494186401367, |
| "learning_rate": 1.4e-05, |
| "loss": 0.2565, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.8791208791208791, |
| "grad_norm": 15.667474746704102, |
| "learning_rate": 1.4400000000000001e-05, |
| "loss": 0.2572, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.9035409035409036, |
| "grad_norm": 38.89118194580078, |
| "learning_rate": 1.48e-05, |
| "loss": 0.2842, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.927960927960928, |
| "grad_norm": 25.795814514160156, |
| "learning_rate": 1.5200000000000002e-05, |
| "loss": 0.1919, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.9523809523809523, |
| "grad_norm": 57.172943115234375, |
| "learning_rate": 1.5600000000000003e-05, |
| "loss": 0.1728, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.9768009768009768, |
| "grad_norm": 12.05911922454834, |
| "learning_rate": 1.6000000000000003e-05, |
| "loss": 0.2717, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.9768009768009768, |
| "eval_accuracy": 0.9132559560171044, |
| "eval_accuracy_label_negative": 0.9318181818181818, |
| "eval_accuracy_label_positive": 0.8958579881656805, |
| "eval_f1": 0.9132768698355349, |
| "eval_loss": 0.25323960185050964, |
| "eval_precision": 0.9140857692521688, |
| "eval_recall": 0.9132559560171044, |
| "eval_runtime": 6.6576, |
| "eval_samples_per_second": 245.883, |
| "eval_steps_per_second": 15.471, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.0012210012210012, |
| "grad_norm": 175.93984985351562, |
| "learning_rate": 1.64e-05, |
| "loss": 0.2158, |
| "step": 410 |
| }, |
| { |
| "epoch": 1.0256410256410255, |
| "grad_norm": 4.0570220947265625, |
| "learning_rate": 1.6800000000000002e-05, |
| "loss": 0.1614, |
| "step": 420 |
| }, |
| { |
| "epoch": 1.05006105006105, |
| "grad_norm": 20.152868270874023, |
| "learning_rate": 1.72e-05, |
| "loss": 0.1849, |
| "step": 430 |
| }, |
| { |
| "epoch": 1.0744810744810744, |
| "grad_norm": 15.713109970092773, |
| "learning_rate": 1.76e-05, |
| "loss": 0.1128, |
| "step": 440 |
| }, |
| { |
| "epoch": 1.098901098901099, |
| "grad_norm": 6.342801094055176, |
| "learning_rate": 1.8e-05, |
| "loss": 0.1914, |
| "step": 450 |
| }, |
| { |
| "epoch": 1.1233211233211233, |
| "grad_norm": 35.65519714355469, |
| "learning_rate": 1.8400000000000003e-05, |
| "loss": 0.2143, |
| "step": 460 |
| }, |
| { |
| "epoch": 1.1477411477411477, |
| "grad_norm": 61.438838958740234, |
| "learning_rate": 1.88e-05, |
| "loss": 0.2332, |
| "step": 470 |
| }, |
| { |
| "epoch": 1.1721611721611722, |
| "grad_norm": 21.36277961730957, |
| "learning_rate": 1.9200000000000003e-05, |
| "loss": 0.2433, |
| "step": 480 |
| }, |
| { |
| "epoch": 1.1965811965811965, |
| "grad_norm": 6.238389015197754, |
| "learning_rate": 1.9600000000000002e-05, |
| "loss": 0.131, |
| "step": 490 |
| }, |
| { |
| "epoch": 1.221001221001221, |
| "grad_norm": 44.72506332397461, |
| "learning_rate": 2e-05, |
| "loss": 0.2076, |
| "step": 500 |
| }, |
| { |
| "epoch": 1.221001221001221, |
| "eval_accuracy": 0.9083689676237019, |
| "eval_accuracy_label_negative": 0.9457070707070707, |
| "eval_accuracy_label_positive": 0.8733727810650888, |
| "eval_f1": 0.9083448942204814, |
| "eval_loss": 0.25875234603881836, |
| "eval_precision": 0.9111034284789092, |
| "eval_recall": 0.9083689676237019, |
| "eval_runtime": 6.2625, |
| "eval_samples_per_second": 261.397, |
| "eval_steps_per_second": 16.447, |
| "step": 500 |
| }, |
| { |
| "epoch": 1.2454212454212454, |
| "grad_norm": 24.117422103881836, |
| "learning_rate": 1.9724896836313617e-05, |
| "loss": 0.217, |
| "step": 510 |
| }, |
| { |
| "epoch": 1.2698412698412698, |
| "grad_norm": 15.197860717773438, |
| "learning_rate": 1.9449793672627236e-05, |
| "loss": 0.2137, |
| "step": 520 |
| }, |
| { |
| "epoch": 1.2942612942612943, |
| "grad_norm": 15.89400577545166, |
| "learning_rate": 1.9174690508940855e-05, |
| "loss": 0.2309, |
| "step": 530 |
| }, |
| { |
| "epoch": 1.3186813186813187, |
| "grad_norm": 15.406135559082031, |
| "learning_rate": 1.889958734525447e-05, |
| "loss": 0.2548, |
| "step": 540 |
| }, |
| { |
| "epoch": 1.3431013431013432, |
| "grad_norm": 20.590370178222656, |
| "learning_rate": 1.862448418156809e-05, |
| "loss": 0.1266, |
| "step": 550 |
| }, |
| { |
| "epoch": 1.3675213675213675, |
| "grad_norm": 19.7510929107666, |
| "learning_rate": 1.8349381017881708e-05, |
| "loss": 0.3163, |
| "step": 560 |
| }, |
| { |
| "epoch": 1.3919413919413919, |
| "grad_norm": 43.13203048706055, |
| "learning_rate": 1.8074277854195327e-05, |
| "loss": 0.19, |
| "step": 570 |
| }, |
| { |
| "epoch": 1.4163614163614164, |
| "grad_norm": 9.15944766998291, |
| "learning_rate": 1.7799174690508942e-05, |
| "loss": 0.2279, |
| "step": 580 |
| }, |
| { |
| "epoch": 1.4407814407814408, |
| "grad_norm": 8.235884666442871, |
| "learning_rate": 1.7524071526822558e-05, |
| "loss": 0.1793, |
| "step": 590 |
| }, |
| { |
| "epoch": 1.4652014652014653, |
| "grad_norm": 15.345870018005371, |
| "learning_rate": 1.7248968363136177e-05, |
| "loss": 0.1745, |
| "step": 600 |
| }, |
| { |
| "epoch": 1.4652014652014653, |
| "eval_accuracy": 0.9132559560171044, |
| "eval_accuracy_label_negative": 0.9027777777777778, |
| "eval_accuracy_label_positive": 0.9230769230769231, |
| "eval_f1": 0.9132306675694815, |
| "eval_loss": 0.2216513752937317, |
| "eval_precision": 0.9133037139258499, |
| "eval_recall": 0.9132559560171044, |
| "eval_runtime": 6.2687, |
| "eval_samples_per_second": 261.138, |
| "eval_steps_per_second": 16.431, |
| "step": 600 |
| }, |
| { |
| "epoch": 1.4896214896214897, |
| "grad_norm": 11.133660316467285, |
| "learning_rate": 1.6973865199449796e-05, |
| "loss": 0.1678, |
| "step": 610 |
| }, |
| { |
| "epoch": 1.514041514041514, |
| "grad_norm": 19.868358612060547, |
| "learning_rate": 1.669876203576341e-05, |
| "loss": 0.1957, |
| "step": 620 |
| }, |
| { |
| "epoch": 1.5384615384615383, |
| "grad_norm": 12.538798332214355, |
| "learning_rate": 1.642365887207703e-05, |
| "loss": 0.1615, |
| "step": 630 |
| }, |
| { |
| "epoch": 1.5628815628815629, |
| "grad_norm": 44.59722900390625, |
| "learning_rate": 1.614855570839065e-05, |
| "loss": 0.239, |
| "step": 640 |
| }, |
| { |
| "epoch": 1.5873015873015874, |
| "grad_norm": 21.872129440307617, |
| "learning_rate": 1.5873452544704268e-05, |
| "loss": 0.2026, |
| "step": 650 |
| }, |
| { |
| "epoch": 1.6117216117216118, |
| "grad_norm": 69.74796295166016, |
| "learning_rate": 1.5598349381017883e-05, |
| "loss": 0.1606, |
| "step": 660 |
| }, |
| { |
| "epoch": 1.636141636141636, |
| "grad_norm": 18.729034423828125, |
| "learning_rate": 1.53232462173315e-05, |
| "loss": 0.1394, |
| "step": 670 |
| }, |
| { |
| "epoch": 1.6605616605616604, |
| "grad_norm": 34.83360290527344, |
| "learning_rate": 1.5048143053645119e-05, |
| "loss": 0.2067, |
| "step": 680 |
| }, |
| { |
| "epoch": 1.684981684981685, |
| "grad_norm": 26.961673736572266, |
| "learning_rate": 1.4773039889958736e-05, |
| "loss": 0.2053, |
| "step": 690 |
| }, |
| { |
| "epoch": 1.7094017094017095, |
| "grad_norm": 11.258790969848633, |
| "learning_rate": 1.4497936726272352e-05, |
| "loss": 0.21, |
| "step": 700 |
| }, |
| { |
| "epoch": 1.7094017094017095, |
| "eval_accuracy": 0.9156994502138057, |
| "eval_accuracy_label_negative": 0.9078282828282829, |
| "eval_accuracy_label_positive": 0.9230769230769231, |
| "eval_f1": 0.9156840772951227, |
| "eval_loss": 0.21606621146202087, |
| "eval_precision": 0.9157125089385034, |
| "eval_recall": 0.9156994502138057, |
| "eval_runtime": 6.2326, |
| "eval_samples_per_second": 262.649, |
| "eval_steps_per_second": 16.526, |
| "step": 700 |
| }, |
| { |
| "epoch": 1.7338217338217339, |
| "grad_norm": 11.965163230895996, |
| "learning_rate": 1.4222833562585972e-05, |
| "loss": 0.1675, |
| "step": 710 |
| }, |
| { |
| "epoch": 1.7582417582417582, |
| "grad_norm": 9.539685249328613, |
| "learning_rate": 1.3947730398899588e-05, |
| "loss": 0.1617, |
| "step": 720 |
| }, |
| { |
| "epoch": 1.7826617826617825, |
| "grad_norm": 13.261054039001465, |
| "learning_rate": 1.3672627235213205e-05, |
| "loss": 0.1469, |
| "step": 730 |
| }, |
| { |
| "epoch": 1.807081807081807, |
| "grad_norm": 9.91487979888916, |
| "learning_rate": 1.3397524071526824e-05, |
| "loss": 0.1807, |
| "step": 740 |
| }, |
| { |
| "epoch": 1.8315018315018317, |
| "grad_norm": 4.809123992919922, |
| "learning_rate": 1.3122420907840441e-05, |
| "loss": 0.1937, |
| "step": 750 |
| }, |
| { |
| "epoch": 1.855921855921856, |
| "grad_norm": 10.304144859313965, |
| "learning_rate": 1.284731774415406e-05, |
| "loss": 0.1712, |
| "step": 760 |
| }, |
| { |
| "epoch": 1.8803418803418803, |
| "grad_norm": 9.22486400604248, |
| "learning_rate": 1.2572214580467677e-05, |
| "loss": 0.1429, |
| "step": 770 |
| }, |
| { |
| "epoch": 1.9047619047619047, |
| "grad_norm": 25.931554794311523, |
| "learning_rate": 1.2297111416781292e-05, |
| "loss": 0.1961, |
| "step": 780 |
| }, |
| { |
| "epoch": 1.9291819291819292, |
| "grad_norm": 21.258495330810547, |
| "learning_rate": 1.2022008253094913e-05, |
| "loss": 0.2019, |
| "step": 790 |
| }, |
| { |
| "epoch": 1.9536019536019538, |
| "grad_norm": 6.851562023162842, |
| "learning_rate": 1.1746905089408529e-05, |
| "loss": 0.1349, |
| "step": 800 |
| }, |
| { |
| "epoch": 1.9536019536019538, |
| "eval_accuracy": 0.9242516799022602, |
| "eval_accuracy_label_negative": 0.9078282828282829, |
| "eval_accuracy_label_positive": 0.9396449704142011, |
| "eval_f1": 0.9242049416831661, |
| "eval_loss": 0.2092478722333908, |
| "eval_precision": 0.9244931103347949, |
| "eval_recall": 0.9242516799022602, |
| "eval_runtime": 6.2401, |
| "eval_samples_per_second": 262.338, |
| "eval_steps_per_second": 16.506, |
| "step": 800 |
| }, |
| { |
| "epoch": 1.978021978021978, |
| "grad_norm": 27.77749252319336, |
| "learning_rate": 1.1471801925722146e-05, |
| "loss": 0.2208, |
| "step": 810 |
| }, |
| { |
| "epoch": 2.0024420024420024, |
| "grad_norm": 9.1483154296875, |
| "learning_rate": 1.1196698762035765e-05, |
| "loss": 0.1494, |
| "step": 820 |
| }, |
| { |
| "epoch": 2.0268620268620268, |
| "grad_norm": 7.372486591339111, |
| "learning_rate": 1.0921595598349382e-05, |
| "loss": 0.1161, |
| "step": 830 |
| }, |
| { |
| "epoch": 2.051282051282051, |
| "grad_norm": 9.255379676818848, |
| "learning_rate": 1.0646492434663e-05, |
| "loss": 0.1054, |
| "step": 840 |
| }, |
| { |
| "epoch": 2.075702075702076, |
| "grad_norm": 14.69979476928711, |
| "learning_rate": 1.0371389270976618e-05, |
| "loss": 0.127, |
| "step": 850 |
| }, |
| { |
| "epoch": 2.1001221001221, |
| "grad_norm": 5.8154473304748535, |
| "learning_rate": 1.0096286107290235e-05, |
| "loss": 0.1329, |
| "step": 860 |
| }, |
| { |
| "epoch": 2.1245421245421245, |
| "grad_norm": 8.349048614501953, |
| "learning_rate": 9.821182943603852e-06, |
| "loss": 0.0872, |
| "step": 870 |
| }, |
| { |
| "epoch": 2.148962148962149, |
| "grad_norm": 7.116409778594971, |
| "learning_rate": 9.54607977991747e-06, |
| "loss": 0.0743, |
| "step": 880 |
| }, |
| { |
| "epoch": 2.173382173382173, |
| "grad_norm": 2.7027432918548584, |
| "learning_rate": 9.270976616231088e-06, |
| "loss": 0.1437, |
| "step": 890 |
| }, |
| { |
| "epoch": 2.197802197802198, |
| "grad_norm": 13.962987899780273, |
| "learning_rate": 8.995873452544705e-06, |
| "loss": 0.1795, |
| "step": 900 |
| }, |
| { |
| "epoch": 2.197802197802198, |
| "eval_accuracy": 0.9175320708613317, |
| "eval_accuracy_label_negative": 0.9431818181818182, |
| "eval_accuracy_label_positive": 0.893491124260355, |
| "eval_f1": 0.9175431498623747, |
| "eval_loss": 0.24922694265842438, |
| "eval_precision": 0.918941636944056, |
| "eval_recall": 0.9175320708613317, |
| "eval_runtime": 6.239, |
| "eval_samples_per_second": 262.38, |
| "eval_steps_per_second": 16.509, |
| "step": 900 |
| }, |
| { |
| "epoch": 2.2222222222222223, |
| "grad_norm": 1.1736209392547607, |
| "learning_rate": 8.720770288858322e-06, |
| "loss": 0.1549, |
| "step": 910 |
| }, |
| { |
| "epoch": 2.2466422466422467, |
| "grad_norm": 9.85667610168457, |
| "learning_rate": 8.44566712517194e-06, |
| "loss": 0.1192, |
| "step": 920 |
| }, |
| { |
| "epoch": 2.271062271062271, |
| "grad_norm": 22.14609146118164, |
| "learning_rate": 8.170563961485558e-06, |
| "loss": 0.144, |
| "step": 930 |
| }, |
| { |
| "epoch": 2.2954822954822953, |
| "grad_norm": 1.3833993673324585, |
| "learning_rate": 7.895460797799176e-06, |
| "loss": 0.1299, |
| "step": 940 |
| }, |
| { |
| "epoch": 2.3199023199023197, |
| "grad_norm": 18.047609329223633, |
| "learning_rate": 7.620357634112793e-06, |
| "loss": 0.1161, |
| "step": 950 |
| }, |
| { |
| "epoch": 2.3443223443223444, |
| "grad_norm": 26.69508171081543, |
| "learning_rate": 7.345254470426411e-06, |
| "loss": 0.0901, |
| "step": 960 |
| }, |
| { |
| "epoch": 2.3687423687423688, |
| "grad_norm": 2.519742965698242, |
| "learning_rate": 7.070151306740029e-06, |
| "loss": 0.0839, |
| "step": 970 |
| }, |
| { |
| "epoch": 2.393162393162393, |
| "grad_norm": 51.8133659362793, |
| "learning_rate": 6.795048143053645e-06, |
| "loss": 0.156, |
| "step": 980 |
| }, |
| { |
| "epoch": 2.4175824175824174, |
| "grad_norm": 31.978530883789062, |
| "learning_rate": 6.519944979367263e-06, |
| "loss": 0.1026, |
| "step": 990 |
| }, |
| { |
| "epoch": 2.442002442002442, |
| "grad_norm": 35.09687042236328, |
| "learning_rate": 6.244841815680881e-06, |
| "loss": 0.107, |
| "step": 1000 |
| }, |
| { |
| "epoch": 2.442002442002442, |
| "eval_accuracy": 0.9120342089187539, |
| "eval_accuracy_label_negative": 0.9595959595959596, |
| "eval_accuracy_label_positive": 0.8674556213017751, |
| "eval_f1": 0.911963285711659, |
| "eval_loss": 0.274323433637619, |
| "eval_precision": 0.9162668443060722, |
| "eval_recall": 0.9120342089187539, |
| "eval_runtime": 6.2476, |
| "eval_samples_per_second": 262.022, |
| "eval_steps_per_second": 16.486, |
| "step": 1000 |
| }, |
| { |
| "epoch": 2.4664224664224665, |
| "grad_norm": 63.01274108886719, |
| "learning_rate": 5.969738651994499e-06, |
| "loss": 0.1913, |
| "step": 1010 |
| }, |
| { |
| "epoch": 2.490842490842491, |
| "grad_norm": 59.207305908203125, |
| "learning_rate": 5.6946354883081155e-06, |
| "loss": 0.1108, |
| "step": 1020 |
| }, |
| { |
| "epoch": 2.515262515262515, |
| "grad_norm": 11.869065284729004, |
| "learning_rate": 5.4195323246217335e-06, |
| "loss": 0.0884, |
| "step": 1030 |
| }, |
| { |
| "epoch": 2.5396825396825395, |
| "grad_norm": 12.766225814819336, |
| "learning_rate": 5.1444291609353515e-06, |
| "loss": 0.1491, |
| "step": 1040 |
| }, |
| { |
| "epoch": 2.564102564102564, |
| "grad_norm": 5.824493408203125, |
| "learning_rate": 4.869325997248969e-06, |
| "loss": 0.1633, |
| "step": 1050 |
| }, |
| { |
| "epoch": 2.5885225885225887, |
| "grad_norm": 6.6170477867126465, |
| "learning_rate": 4.594222833562587e-06, |
| "loss": 0.1276, |
| "step": 1060 |
| }, |
| { |
| "epoch": 2.612942612942613, |
| "grad_norm": 9.731184959411621, |
| "learning_rate": 4.319119669876204e-06, |
| "loss": 0.1023, |
| "step": 1070 |
| }, |
| { |
| "epoch": 2.6373626373626373, |
| "grad_norm": 3.0301451683044434, |
| "learning_rate": 4.044016506189822e-06, |
| "loss": 0.1082, |
| "step": 1080 |
| }, |
| { |
| "epoch": 2.6617826617826617, |
| "grad_norm": 3.955033779144287, |
| "learning_rate": 3.768913342503439e-06, |
| "loss": 0.1763, |
| "step": 1090 |
| }, |
| { |
| "epoch": 2.6862026862026864, |
| "grad_norm": 46.200225830078125, |
| "learning_rate": 3.4938101788170562e-06, |
| "loss": 0.08, |
| "step": 1100 |
| }, |
| { |
| "epoch": 2.6862026862026864, |
| "eval_accuracy": 0.9187538179596824, |
| "eval_accuracy_label_negative": 0.9431818181818182, |
| "eval_accuracy_label_positive": 0.8958579881656805, |
| "eval_f1": 0.9187668553379099, |
| "eval_loss": 0.2606331408023834, |
| "eval_precision": 0.9200482702030894, |
| "eval_recall": 0.9187538179596824, |
| "eval_runtime": 6.2203, |
| "eval_samples_per_second": 263.172, |
| "eval_steps_per_second": 16.559, |
| "step": 1100 |
| }, |
| { |
| "epoch": 2.7106227106227108, |
| "grad_norm": 42.084537506103516, |
| "learning_rate": 3.2187070151306742e-06, |
| "loss": 0.1241, |
| "step": 1110 |
| }, |
| { |
| "epoch": 2.735042735042735, |
| "grad_norm": 11.397581100463867, |
| "learning_rate": 2.943603851444292e-06, |
| "loss": 0.0858, |
| "step": 1120 |
| }, |
| { |
| "epoch": 2.7594627594627594, |
| "grad_norm": 58.64878845214844, |
| "learning_rate": 2.6685006877579094e-06, |
| "loss": 0.1114, |
| "step": 1130 |
| }, |
| { |
| "epoch": 2.7838827838827838, |
| "grad_norm": 43.395687103271484, |
| "learning_rate": 2.393397524071527e-06, |
| "loss": 0.1682, |
| "step": 1140 |
| }, |
| { |
| "epoch": 2.808302808302808, |
| "grad_norm": 17.384784698486328, |
| "learning_rate": 2.1182943603851446e-06, |
| "loss": 0.1692, |
| "step": 1150 |
| }, |
| { |
| "epoch": 2.832722832722833, |
| "grad_norm": 24.320571899414062, |
| "learning_rate": 1.8431911966987622e-06, |
| "loss": 0.1473, |
| "step": 1160 |
| }, |
| { |
| "epoch": 2.857142857142857, |
| "grad_norm": 2.8611717224121094, |
| "learning_rate": 1.56808803301238e-06, |
| "loss": 0.1429, |
| "step": 1170 |
| }, |
| { |
| "epoch": 2.8815628815628815, |
| "grad_norm": 16.274402618408203, |
| "learning_rate": 1.2929848693259976e-06, |
| "loss": 0.0589, |
| "step": 1180 |
| }, |
| { |
| "epoch": 2.905982905982906, |
| "grad_norm": 12.312983512878418, |
| "learning_rate": 1.017881705639615e-06, |
| "loss": 0.113, |
| "step": 1190 |
| }, |
| { |
| "epoch": 2.9304029304029307, |
| "grad_norm": 2.3755383491516113, |
| "learning_rate": 7.427785419532325e-07, |
| "loss": 0.1275, |
| "step": 1200 |
| }, |
| { |
| "epoch": 2.9304029304029307, |
| "eval_accuracy": 0.9254734270006109, |
| "eval_accuracy_label_negative": 0.9166666666666666, |
| "eval_accuracy_label_positive": 0.9337278106508876, |
| "eval_f1": 0.9254558801955471, |
| "eval_loss": 0.2550033628940582, |
| "eval_precision": 0.925507524308242, |
| "eval_recall": 0.9254734270006109, |
| "eval_runtime": 6.2179, |
| "eval_samples_per_second": 263.27, |
| "eval_steps_per_second": 16.565, |
| "step": 1200 |
| }, |
| { |
| "epoch": 2.954822954822955, |
| "grad_norm": 11.446648597717285, |
| "learning_rate": 4.6767537826685014e-07, |
| "loss": 0.0984, |
| "step": 1210 |
| }, |
| { |
| "epoch": 2.9792429792429793, |
| "grad_norm": 6.963866710662842, |
| "learning_rate": 1.925722145804677e-07, |
| "loss": 0.1507, |
| "step": 1220 |
| }, |
| { |
| "epoch": 2.9963369963369964, |
| "step": 1227, |
| "total_flos": 157212915612480.0, |
| "train_loss": 0.22916113538477803, |
| "train_runtime": 681.0988, |
| "train_samples_per_second": 57.657, |
| "train_steps_per_second": 1.802 |
| }, |
| { |
| "epoch": 2.9963369963369964, |
| "eval_accuracy": 0.9260843005497862, |
| "eval_accuracy_label_negative": 0.9255050505050505, |
| "eval_accuracy_label_positive": 0.9266272189349113, |
| "eval_f1": 0.9260884418426176, |
| "eval_loss": 0.25100183486938477, |
| "eval_precision": 0.9260988078725894, |
| "eval_recall": 0.9260843005497862, |
| "eval_runtime": 6.2023, |
| "eval_samples_per_second": 263.933, |
| "eval_steps_per_second": 16.607, |
| "step": 1227 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 1227, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 1000, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 157212915612480.0, |
| "train_batch_size": 16, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
