| { | |
| "best_metric": 0.272621791856563, | |
| "best_model_checkpoint": "/content/drive/MyDrive/model_outputs/trial_7/checkpoint-882", | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 882, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.011337868480725623, | |
| "grad_norm": 10.448426246643066, | |
| "learning_rate": 3.2522795772612694e-07, | |
| "loss": 2.2378, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.022675736961451247, | |
| "grad_norm": 9.789140701293945, | |
| "learning_rate": 6.504559154522539e-07, | |
| "loss": 2.2313, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.034013605442176874, | |
| "grad_norm": 10.550893783569336, | |
| "learning_rate": 9.756838731783808e-07, | |
| "loss": 2.2025, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.045351473922902494, | |
| "grad_norm": 11.21814250946045, | |
| "learning_rate": 1.3009118309045078e-06, | |
| "loss": 2.2219, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05668934240362812, | |
| "grad_norm": 9.440560340881348, | |
| "learning_rate": 1.6261397886306348e-06, | |
| "loss": 2.1929, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06802721088435375, | |
| "grad_norm": 9.947296142578125, | |
| "learning_rate": 1.9513677463567616e-06, | |
| "loss": 2.1799, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07936507936507936, | |
| "grad_norm": 9.87543773651123, | |
| "learning_rate": 2.2765957040828886e-06, | |
| "loss": 2.1658, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09070294784580499, | |
| "grad_norm": 8.421579360961914, | |
| "learning_rate": 2.6018236618090155e-06, | |
| "loss": 2.1612, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.10204081632653061, | |
| "grad_norm": 11.202038764953613, | |
| "learning_rate": 2.9270516195351425e-06, | |
| "loss": 2.1247, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11337868480725624, | |
| "grad_norm": 12.861761093139648, | |
| "learning_rate": 3.2522795772612695e-06, | |
| "loss": 2.1237, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12471655328798185, | |
| "grad_norm": 11.995753288269043, | |
| "learning_rate": 3.5775075349873965e-06, | |
| "loss": 2.0857, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.1360544217687075, | |
| "grad_norm": 10.376130104064941, | |
| "learning_rate": 3.902735492713523e-06, | |
| "loss": 2.0663, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.1473922902494331, | |
| "grad_norm": 11.153587341308594, | |
| "learning_rate": 4.227963450439651e-06, | |
| "loss": 2.0499, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15873015873015872, | |
| "grad_norm": 10.43906307220459, | |
| "learning_rate": 4.553191408165777e-06, | |
| "loss": 2.0444, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.17006802721088435, | |
| "grad_norm": 10.224808692932129, | |
| "learning_rate": 4.878419365891904e-06, | |
| "loss": 2.0634, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.18140589569160998, | |
| "grad_norm": 11.265706062316895, | |
| "learning_rate": 5.203647323618031e-06, | |
| "loss": 1.9827, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.1927437641723356, | |
| "grad_norm": 9.84843921661377, | |
| "learning_rate": 5.528875281344158e-06, | |
| "loss": 2.0075, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.20408163265306123, | |
| "grad_norm": 8.700265884399414, | |
| "learning_rate": 5.854103239070285e-06, | |
| "loss": 2.0327, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.21541950113378686, | |
| "grad_norm": 9.785911560058594, | |
| "learning_rate": 6.179331196796412e-06, | |
| "loss": 2.0588, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.22675736961451248, | |
| "grad_norm": 11.363122940063477, | |
| "learning_rate": 6.504559154522539e-06, | |
| "loss": 2.1046, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.23809523809523808, | |
| "grad_norm": 8.90456485748291, | |
| "learning_rate": 6.829787112248665e-06, | |
| "loss": 2.0855, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.2494331065759637, | |
| "grad_norm": 9.236421585083008, | |
| "learning_rate": 7.155015069974793e-06, | |
| "loss": 2.0381, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.26077097505668934, | |
| "grad_norm": 7.954260349273682, | |
| "learning_rate": 7.48024302770092e-06, | |
| "loss": 2.049, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.272108843537415, | |
| "grad_norm": 8.307881355285645, | |
| "learning_rate": 7.805470985427046e-06, | |
| "loss": 1.969, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.2834467120181406, | |
| "grad_norm": 10.518112182617188, | |
| "learning_rate": 8.130698943153173e-06, | |
| "loss": 1.9447, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.2947845804988662, | |
| "grad_norm": 9.017768859863281, | |
| "learning_rate": 8.455926900879302e-06, | |
| "loss": 2.0961, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.30612244897959184, | |
| "grad_norm": 8.073211669921875, | |
| "learning_rate": 8.781154858605427e-06, | |
| "loss": 1.9331, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.31746031746031744, | |
| "grad_norm": 9.508843421936035, | |
| "learning_rate": 9.106382816331554e-06, | |
| "loss": 1.9698, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.3287981859410431, | |
| "grad_norm": 9.191576957702637, | |
| "learning_rate": 9.431610774057681e-06, | |
| "loss": 1.9649, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.3401360544217687, | |
| "grad_norm": 10.584879875183105, | |
| "learning_rate": 9.756838731783808e-06, | |
| "loss": 1.9858, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.35147392290249435, | |
| "grad_norm": 10.487820625305176, | |
| "learning_rate": 1.0082066689509935e-05, | |
| "loss": 1.9865, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.36281179138321995, | |
| "grad_norm": 11.283440589904785, | |
| "learning_rate": 1.0407294647236062e-05, | |
| "loss": 1.9821, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.3741496598639456, | |
| "grad_norm": 8.875638008117676, | |
| "learning_rate": 1.073252260496219e-05, | |
| "loss": 1.929, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.3854875283446712, | |
| "grad_norm": 10.396716117858887, | |
| "learning_rate": 1.1057750562688316e-05, | |
| "loss": 2.026, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.3968253968253968, | |
| "grad_norm": 7.329042911529541, | |
| "learning_rate": 1.1382978520414443e-05, | |
| "loss": 1.9425, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.40816326530612246, | |
| "grad_norm": 11.754528999328613, | |
| "learning_rate": 1.170820647814057e-05, | |
| "loss": 1.8786, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.41950113378684806, | |
| "grad_norm": 8.387690544128418, | |
| "learning_rate": 1.1732014759459226e-05, | |
| "loss": 1.9279, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.4308390022675737, | |
| "grad_norm": 10.3031005859375, | |
| "learning_rate": 1.1680468121676012e-05, | |
| "loss": 1.9839, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.4421768707482993, | |
| "grad_norm": 14.319021224975586, | |
| "learning_rate": 1.16289214838928e-05, | |
| "loss": 2.0223, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.45351473922902497, | |
| "grad_norm": 9.42290210723877, | |
| "learning_rate": 1.1577374846109587e-05, | |
| "loss": 1.9165, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.46485260770975056, | |
| "grad_norm": 11.768155097961426, | |
| "learning_rate": 1.1525828208326375e-05, | |
| "loss": 2.0116, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.47619047619047616, | |
| "grad_norm": 8.139991760253906, | |
| "learning_rate": 1.1474281570543161e-05, | |
| "loss": 1.8628, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.4875283446712018, | |
| "grad_norm": 10.308381080627441, | |
| "learning_rate": 1.1422734932759949e-05, | |
| "loss": 2.017, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.4988662131519274, | |
| "grad_norm": 9.104990005493164, | |
| "learning_rate": 1.1371188294976736e-05, | |
| "loss": 2.0109, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.5102040816326531, | |
| "grad_norm": 9.941882133483887, | |
| "learning_rate": 1.1319641657193522e-05, | |
| "loss": 1.9498, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.5215419501133787, | |
| "grad_norm": 10.337907791137695, | |
| "learning_rate": 1.126809501941031e-05, | |
| "loss": 1.9946, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5328798185941043, | |
| "grad_norm": 9.354191780090332, | |
| "learning_rate": 1.1216548381627098e-05, | |
| "loss": 1.8953, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.54421768707483, | |
| "grad_norm": 10.24344539642334, | |
| "learning_rate": 1.1165001743843885e-05, | |
| "loss": 1.889, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.5555555555555556, | |
| "grad_norm": 8.560635566711426, | |
| "learning_rate": 1.1113455106060671e-05, | |
| "loss": 1.9698, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.5668934240362812, | |
| "grad_norm": 9.784881591796875, | |
| "learning_rate": 1.1061908468277459e-05, | |
| "loss": 1.9718, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.5782312925170068, | |
| "grad_norm": 11.199634552001953, | |
| "learning_rate": 1.1010361830494247e-05, | |
| "loss": 1.9762, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.5895691609977324, | |
| "grad_norm": 10.915511131286621, | |
| "learning_rate": 1.0958815192711034e-05, | |
| "loss": 1.9347, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.6009070294784581, | |
| "grad_norm": 13.744818687438965, | |
| "learning_rate": 1.090726855492782e-05, | |
| "loss": 2.0184, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.6122448979591837, | |
| "grad_norm": 11.262948036193848, | |
| "learning_rate": 1.0855721917144608e-05, | |
| "loss": 1.9323, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.6235827664399093, | |
| "grad_norm": 8.698514938354492, | |
| "learning_rate": 1.0804175279361396e-05, | |
| "loss": 1.917, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6349206349206349, | |
| "grad_norm": 15.39905834197998, | |
| "learning_rate": 1.0752628641578182e-05, | |
| "loss": 1.9203, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.6462585034013606, | |
| "grad_norm": 9.898249626159668, | |
| "learning_rate": 1.070108200379497e-05, | |
| "loss": 1.9317, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.6575963718820862, | |
| "grad_norm": 9.889189720153809, | |
| "learning_rate": 1.0649535366011757e-05, | |
| "loss": 1.9203, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.6689342403628118, | |
| "grad_norm": 9.887190818786621, | |
| "learning_rate": 1.0597988728228545e-05, | |
| "loss": 1.8459, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.6802721088435374, | |
| "grad_norm": 9.046891212463379, | |
| "learning_rate": 1.054644209044533e-05, | |
| "loss": 1.983, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.691609977324263, | |
| "grad_norm": 11.797575950622559, | |
| "learning_rate": 1.0494895452662118e-05, | |
| "loss": 1.9428, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.7029478458049887, | |
| "grad_norm": 13.745688438415527, | |
| "learning_rate": 1.0443348814878906e-05, | |
| "loss": 1.9652, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.7142857142857143, | |
| "grad_norm": 11.441299438476562, | |
| "learning_rate": 1.0391802177095692e-05, | |
| "loss": 1.9636, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.7256235827664399, | |
| "grad_norm": 11.024886131286621, | |
| "learning_rate": 1.034025553931248e-05, | |
| "loss": 1.9634, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.7369614512471655, | |
| "grad_norm": 18.117177963256836, | |
| "learning_rate": 1.0288708901529269e-05, | |
| "loss": 1.9059, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7482993197278912, | |
| "grad_norm": 13.534732818603516, | |
| "learning_rate": 1.0237162263746055e-05, | |
| "loss": 1.8964, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.7596371882086168, | |
| "grad_norm": 11.099019050598145, | |
| "learning_rate": 1.0185615625962842e-05, | |
| "loss": 1.7963, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.7709750566893424, | |
| "grad_norm": 26.388456344604492, | |
| "learning_rate": 1.013406898817963e-05, | |
| "loss": 1.9809, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.782312925170068, | |
| "grad_norm": 12.548821449279785, | |
| "learning_rate": 1.0082522350396418e-05, | |
| "loss": 1.914, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.7936507936507936, | |
| "grad_norm": 11.160512924194336, | |
| "learning_rate": 1.0030975712613204e-05, | |
| "loss": 1.82, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.8049886621315193, | |
| "grad_norm": 13.189496994018555, | |
| "learning_rate": 9.979429074829991e-06, | |
| "loss": 1.835, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.8163265306122449, | |
| "grad_norm": 14.305792808532715, | |
| "learning_rate": 9.927882437046779e-06, | |
| "loss": 1.7968, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.8276643990929705, | |
| "grad_norm": 15.657683372497559, | |
| "learning_rate": 9.876335799263565e-06, | |
| "loss": 1.8773, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.8390022675736961, | |
| "grad_norm": 13.894340515136719, | |
| "learning_rate": 9.824789161480353e-06, | |
| "loss": 1.848, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8503401360544217, | |
| "grad_norm": 18.09587860107422, | |
| "learning_rate": 9.77324252369714e-06, | |
| "loss": 1.9382, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.8616780045351474, | |
| "grad_norm": 14.405220985412598, | |
| "learning_rate": 9.721695885913928e-06, | |
| "loss": 1.973, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.873015873015873, | |
| "grad_norm": 12.452035903930664, | |
| "learning_rate": 9.670149248130714e-06, | |
| "loss": 1.83, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.8843537414965986, | |
| "grad_norm": 17.113231658935547, | |
| "learning_rate": 9.618602610347502e-06, | |
| "loss": 1.9733, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.8956916099773242, | |
| "grad_norm": 11.744027137756348, | |
| "learning_rate": 9.56705597256429e-06, | |
| "loss": 1.7322, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.9070294784580499, | |
| "grad_norm": 12.800016403198242, | |
| "learning_rate": 9.515509334781077e-06, | |
| "loss": 1.7837, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.9183673469387755, | |
| "grad_norm": 12.709762573242188, | |
| "learning_rate": 9.463962696997863e-06, | |
| "loss": 1.8111, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.9297052154195011, | |
| "grad_norm": 12.49414348602295, | |
| "learning_rate": 9.41241605921465e-06, | |
| "loss": 1.8696, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.9410430839002267, | |
| "grad_norm": 14.208767890930176, | |
| "learning_rate": 9.360869421431438e-06, | |
| "loss": 1.9039, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.9523809523809523, | |
| "grad_norm": 12.151473999023438, | |
| "learning_rate": 9.309322783648224e-06, | |
| "loss": 1.9351, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.963718820861678, | |
| "grad_norm": 15.71224594116211, | |
| "learning_rate": 9.257776145865012e-06, | |
| "loss": 1.7715, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.9750566893424036, | |
| "grad_norm": 12.460648536682129, | |
| "learning_rate": 9.2062295080818e-06, | |
| "loss": 1.8283, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.9863945578231292, | |
| "grad_norm": 14.45302963256836, | |
| "learning_rate": 9.154682870298587e-06, | |
| "loss": 1.8793, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.9977324263038548, | |
| "grad_norm": 16.95499610900879, | |
| "learning_rate": 9.103136232515373e-06, | |
| "loss": 1.7859, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_classification_report": { | |
| "accuracy": 0.31624214356599406, | |
| "ar": { | |
| "f1-score": 0.34932349323493234, | |
| "precision": 0.32494279176201374, | |
| "recall": 0.3776595744680851, | |
| "support": 376.0 | |
| }, | |
| "cl": { | |
| "f1-score": 0.2806361085126286, | |
| "precision": 0.30425963488843816, | |
| "recall": 0.2604166666666667, | |
| "support": 576.0 | |
| }, | |
| "co": { | |
| "f1-score": 0.29607250755287007, | |
| "precision": 0.3081761006289308, | |
| "recall": 0.28488372093023256, | |
| "support": 344.0 | |
| }, | |
| "es": { | |
| "f1-score": 0.3728115345005149, | |
| "precision": 0.42289719626168226, | |
| "recall": 0.3333333333333333, | |
| "support": 543.0 | |
| }, | |
| "macro avg": { | |
| "f1-score": 0.272621791856563, | |
| "precision": 0.30099582851554657, | |
| "recall": 0.2856216737927651, | |
| "support": 3023.0 | |
| }, | |
| "mx": { | |
| "f1-score": 0.35931307793923384, | |
| "precision": 0.26433430515063167, | |
| "recall": 0.5608247422680412, | |
| "support": 485.0 | |
| }, | |
| "pe": { | |
| "f1-score": 0.16205533596837945, | |
| "precision": 0.25949367088607594, | |
| "recall": 0.11781609195402298, | |
| "support": 348.0 | |
| }, | |
| "pr": { | |
| "f1-score": 0.5296803652968036, | |
| "precision": 0.4915254237288136, | |
| "recall": 0.5742574257425742, | |
| "support": 101.0 | |
| }, | |
| "uy": { | |
| "f1-score": 0.1037037037037037, | |
| "precision": 0.3333333333333333, | |
| "recall": 0.06140350877192982, | |
| "support": 228.0 | |
| }, | |
| "ve": { | |
| "f1-score": 0.0, | |
| "precision": 0.0, | |
| "recall": 0.0, | |
| "support": 22.0 | |
| }, | |
| "weighted avg": { | |
| "f1-score": 0.29939857719104257, | |
| "precision": 0.32326424043920743, | |
| "recall": 0.31624214356599406, | |
| "support": 3023.0 | |
| } | |
| }, | |
| "eval_f1": 0.272621791856563, | |
| "eval_loss": 1.826755166053772, | |
| "eval_runtime": 4.3231, | |
| "eval_samples_per_second": 699.273, | |
| "eval_steps_per_second": 87.438, | |
| "step": 882 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2646, | |
| "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": 115989931620288.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |