| { | |
| "best_metric": 0.9376445510679586, | |
| "best_model_checkpoint": "./phobert_results_v2/checkpoint-12145", | |
| "epoch": 7.0, | |
| "eval_steps": 500, | |
| "global_step": 12145, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02881844380403458, | |
| "grad_norm": 4.723288059234619, | |
| "learning_rate": 1.994236311239193e-05, | |
| "loss": 1.9047, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05763688760806916, | |
| "grad_norm": 32.87104034423828, | |
| "learning_rate": 1.9884726224783863e-05, | |
| "loss": 1.7269, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08645533141210375, | |
| "grad_norm": 5.206538677215576, | |
| "learning_rate": 1.9827089337175795e-05, | |
| "loss": 1.6865, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.11527377521613832, | |
| "grad_norm": 52.5133056640625, | |
| "learning_rate": 1.9769452449567724e-05, | |
| "loss": 1.633, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1440922190201729, | |
| "grad_norm": 12.493169784545898, | |
| "learning_rate": 1.9711815561959656e-05, | |
| "loss": 1.553, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1729106628242075, | |
| "grad_norm": 7.675703048706055, | |
| "learning_rate": 1.9654178674351588e-05, | |
| "loss": 1.4131, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.2017291066282421, | |
| "grad_norm": 5.984454154968262, | |
| "learning_rate": 1.9596541786743517e-05, | |
| "loss": 1.3606, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.23054755043227665, | |
| "grad_norm": 9.403727531433105, | |
| "learning_rate": 1.953890489913545e-05, | |
| "loss": 1.2202, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.25936599423631124, | |
| "grad_norm": 7.890481472015381, | |
| "learning_rate": 1.9481268011527378e-05, | |
| "loss": 1.186, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.2881844380403458, | |
| "grad_norm": 12.102771759033203, | |
| "learning_rate": 1.942363112391931e-05, | |
| "loss": 1.0865, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3170028818443804, | |
| "grad_norm": 17.669885635375977, | |
| "learning_rate": 1.936599423631124e-05, | |
| "loss": 1.124, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.345821325648415, | |
| "grad_norm": 10.14146614074707, | |
| "learning_rate": 1.930835734870317e-05, | |
| "loss": 1.1307, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3746397694524496, | |
| "grad_norm": 9.363780975341797, | |
| "learning_rate": 1.9250720461095104e-05, | |
| "loss": 1.0582, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.4034582132564842, | |
| "grad_norm": 22.122907638549805, | |
| "learning_rate": 1.9193083573487033e-05, | |
| "loss": 1.007, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.4322766570605187, | |
| "grad_norm": 22.921249389648438, | |
| "learning_rate": 1.9135446685878965e-05, | |
| "loss": 1.0046, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.4610951008645533, | |
| "grad_norm": 13.880660057067871, | |
| "learning_rate": 1.9077809798270894e-05, | |
| "loss": 0.9786, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4899135446685879, | |
| "grad_norm": 13.75207805633545, | |
| "learning_rate": 1.9020172910662826e-05, | |
| "loss": 0.9166, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.5187319884726225, | |
| "grad_norm": 15.701948165893555, | |
| "learning_rate": 1.8962536023054755e-05, | |
| "loss": 0.8942, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.547550432276657, | |
| "grad_norm": 13.106974601745605, | |
| "learning_rate": 1.8904899135446687e-05, | |
| "loss": 0.8905, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.5763688760806917, | |
| "grad_norm": 13.854475021362305, | |
| "learning_rate": 1.884726224783862e-05, | |
| "loss": 0.9089, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6051873198847262, | |
| "grad_norm": 14.22427749633789, | |
| "learning_rate": 1.878962536023055e-05, | |
| "loss": 0.8688, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.6340057636887608, | |
| "grad_norm": 14.088136672973633, | |
| "learning_rate": 1.873198847262248e-05, | |
| "loss": 0.859, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.6628242074927954, | |
| "grad_norm": 15.746428489685059, | |
| "learning_rate": 1.867435158501441e-05, | |
| "loss": 0.8356, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.69164265129683, | |
| "grad_norm": 10.937832832336426, | |
| "learning_rate": 1.861671469740634e-05, | |
| "loss": 0.8699, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.7204610951008645, | |
| "grad_norm": 14.396600723266602, | |
| "learning_rate": 1.855907780979827e-05, | |
| "loss": 0.7789, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.7492795389048992, | |
| "grad_norm": 11.801300048828125, | |
| "learning_rate": 1.8501440922190203e-05, | |
| "loss": 0.8713, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.7780979827089337, | |
| "grad_norm": 10.435481071472168, | |
| "learning_rate": 1.8443804034582135e-05, | |
| "loss": 0.7961, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.8069164265129684, | |
| "grad_norm": 11.2293062210083, | |
| "learning_rate": 1.8386167146974067e-05, | |
| "loss": 0.8397, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.8357348703170029, | |
| "grad_norm": 12.809613227844238, | |
| "learning_rate": 1.8328530259365996e-05, | |
| "loss": 0.8297, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.8645533141210374, | |
| "grad_norm": 14.791847229003906, | |
| "learning_rate": 1.8270893371757928e-05, | |
| "loss": 0.7709, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.8933717579250721, | |
| "grad_norm": 14.587478637695312, | |
| "learning_rate": 1.8213256484149857e-05, | |
| "loss": 0.7552, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.9221902017291066, | |
| "grad_norm": 16.79636001586914, | |
| "learning_rate": 1.815561959654179e-05, | |
| "loss": 0.7937, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.9510086455331412, | |
| "grad_norm": 8.510680198669434, | |
| "learning_rate": 1.8097982708933718e-05, | |
| "loss": 0.7181, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.9798270893371758, | |
| "grad_norm": 20.466522216796875, | |
| "learning_rate": 1.804034582132565e-05, | |
| "loss": 0.712, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7955036748811067, | |
| "eval_f1_macro": 0.795530535937033, | |
| "eval_f1_weighted": 0.795536244762841, | |
| "eval_loss": 0.6264312267303467, | |
| "eval_precision_macro": 0.7978076489526655, | |
| "eval_precision_weighted": 0.797819457255313, | |
| "eval_recall_macro": 0.7955041514180622, | |
| "eval_recall_weighted": 0.7955036748811067, | |
| "eval_runtime": 29.7702, | |
| "eval_samples_per_second": 233.086, | |
| "eval_steps_per_second": 14.578, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 1.0086455331412103, | |
| "grad_norm": 17.28911018371582, | |
| "learning_rate": 1.7982708933717582e-05, | |
| "loss": 0.6512, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.037463976945245, | |
| "grad_norm": 28.79306411743164, | |
| "learning_rate": 1.792507204610951e-05, | |
| "loss": 0.6036, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.0662824207492796, | |
| "grad_norm": 13.444112777709961, | |
| "learning_rate": 1.7867435158501444e-05, | |
| "loss": 0.552, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.0951008645533142, | |
| "grad_norm": 22.068620681762695, | |
| "learning_rate": 1.7809798270893372e-05, | |
| "loss": 0.577, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.1239193083573487, | |
| "grad_norm": 13.942420959472656, | |
| "learning_rate": 1.7752161383285305e-05, | |
| "loss": 0.5573, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.1527377521613833, | |
| "grad_norm": 13.652227401733398, | |
| "learning_rate": 1.7694524495677234e-05, | |
| "loss": 0.5401, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.181556195965418, | |
| "grad_norm": 22.252243041992188, | |
| "learning_rate": 1.7636887608069166e-05, | |
| "loss": 0.558, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.2103746397694524, | |
| "grad_norm": 17.13372802734375, | |
| "learning_rate": 1.7579250720461095e-05, | |
| "loss": 0.514, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.239193083573487, | |
| "grad_norm": 17.164276123046875, | |
| "learning_rate": 1.7521613832853027e-05, | |
| "loss": 0.5137, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.2680115273775217, | |
| "grad_norm": 14.449616432189941, | |
| "learning_rate": 1.746397694524496e-05, | |
| "loss": 0.5, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.2968299711815563, | |
| "grad_norm": 11.61601734161377, | |
| "learning_rate": 1.7406340057636888e-05, | |
| "loss": 0.5071, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.3256484149855907, | |
| "grad_norm": 20.6407527923584, | |
| "learning_rate": 1.734870317002882e-05, | |
| "loss": 0.4803, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.3544668587896254, | |
| "grad_norm": 18.60149383544922, | |
| "learning_rate": 1.729106628242075e-05, | |
| "loss": 0.4967, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.38328530259366, | |
| "grad_norm": 22.12944793701172, | |
| "learning_rate": 1.723342939481268e-05, | |
| "loss": 0.4652, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.4121037463976944, | |
| "grad_norm": 18.460689544677734, | |
| "learning_rate": 1.717579250720461e-05, | |
| "loss": 0.4417, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.440922190201729, | |
| "grad_norm": 19.452363967895508, | |
| "learning_rate": 1.7118155619596542e-05, | |
| "loss": 0.5082, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.4697406340057637, | |
| "grad_norm": 41.46109390258789, | |
| "learning_rate": 1.7060518731988475e-05, | |
| "loss": 0.463, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.4985590778097984, | |
| "grad_norm": 25.584379196166992, | |
| "learning_rate": 1.7002881844380407e-05, | |
| "loss": 0.4788, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.527377521613833, | |
| "grad_norm": 16.92909049987793, | |
| "learning_rate": 1.6945244956772336e-05, | |
| "loss": 0.4752, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.5561959654178674, | |
| "grad_norm": 10.342531204223633, | |
| "learning_rate": 1.6887608069164268e-05, | |
| "loss": 0.4848, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.585014409221902, | |
| "grad_norm": 15.596243858337402, | |
| "learning_rate": 1.6829971181556197e-05, | |
| "loss": 0.5158, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.6138328530259365, | |
| "grad_norm": 17.031354904174805, | |
| "learning_rate": 1.6772334293948126e-05, | |
| "loss": 0.4577, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.6426512968299711, | |
| "grad_norm": 13.550045013427734, | |
| "learning_rate": 1.6714697406340058e-05, | |
| "loss": 0.4947, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.6714697406340058, | |
| "grad_norm": 10.102880477905273, | |
| "learning_rate": 1.665706051873199e-05, | |
| "loss": 0.4681, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.7002881844380404, | |
| "grad_norm": 10.968811988830566, | |
| "learning_rate": 1.6599423631123922e-05, | |
| "loss": 0.452, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.729106628242075, | |
| "grad_norm": 4.670314311981201, | |
| "learning_rate": 1.654178674351585e-05, | |
| "loss": 0.4346, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.7579250720461095, | |
| "grad_norm": 28.008899688720703, | |
| "learning_rate": 1.6484149855907783e-05, | |
| "loss": 0.4404, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.7867435158501441, | |
| "grad_norm": 27.77347183227539, | |
| "learning_rate": 1.6426512968299712e-05, | |
| "loss": 0.4547, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.8155619596541785, | |
| "grad_norm": 21.949289321899414, | |
| "learning_rate": 1.6368876080691644e-05, | |
| "loss": 0.4924, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.8443804034582132, | |
| "grad_norm": 2.8766281604766846, | |
| "learning_rate": 1.6311239193083573e-05, | |
| "loss": 0.3776, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.8731988472622478, | |
| "grad_norm": 38.812625885009766, | |
| "learning_rate": 1.6253602305475506e-05, | |
| "loss": 0.4176, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.9020172910662825, | |
| "grad_norm": 24.132482528686523, | |
| "learning_rate": 1.6195965417867438e-05, | |
| "loss": 0.4407, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.9308357348703171, | |
| "grad_norm": 19.72426414489746, | |
| "learning_rate": 1.613832853025937e-05, | |
| "loss": 0.4503, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.9596541786743515, | |
| "grad_norm": 16.077899932861328, | |
| "learning_rate": 1.60806916426513e-05, | |
| "loss": 0.4244, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.9884726224783862, | |
| "grad_norm": 21.258529663085938, | |
| "learning_rate": 1.6023054755043228e-05, | |
| "loss": 0.4161, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.87101887880098, | |
| "eval_f1_macro": 0.8705600204306078, | |
| "eval_f1_weighted": 0.8705702821418305, | |
| "eval_loss": 0.43536442518234253, | |
| "eval_precision_macro": 0.8749786652431168, | |
| "eval_precision_weighted": 0.8749977247467932, | |
| "eval_recall_macro": 0.8710178763247105, | |
| "eval_recall_weighted": 0.87101887880098, | |
| "eval_runtime": 29.9836, | |
| "eval_samples_per_second": 231.426, | |
| "eval_steps_per_second": 14.475, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 2.0172910662824206, | |
| "grad_norm": 13.94642448425293, | |
| "learning_rate": 1.596541786743516e-05, | |
| "loss": 0.321, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.0461095100864553, | |
| "grad_norm": 17.44521713256836, | |
| "learning_rate": 1.590778097982709e-05, | |
| "loss": 0.2707, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.07492795389049, | |
| "grad_norm": 12.439199447631836, | |
| "learning_rate": 1.585014409221902e-05, | |
| "loss": 0.2606, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.1037463976945245, | |
| "grad_norm": 10.950318336486816, | |
| "learning_rate": 1.5792507204610953e-05, | |
| "loss": 0.2601, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 2.132564841498559, | |
| "grad_norm": 20.581911087036133, | |
| "learning_rate": 1.5734870317002882e-05, | |
| "loss": 0.265, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.161383285302594, | |
| "grad_norm": 9.320990562438965, | |
| "learning_rate": 1.5677233429394814e-05, | |
| "loss": 0.297, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.1902017291066285, | |
| "grad_norm": 20.260753631591797, | |
| "learning_rate": 1.5619596541786747e-05, | |
| "loss": 0.3121, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.2190201729106627, | |
| "grad_norm": 55.92860412597656, | |
| "learning_rate": 1.5561959654178675e-05, | |
| "loss": 0.2295, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 2.2478386167146973, | |
| "grad_norm": 9.637039184570312, | |
| "learning_rate": 1.5504322766570608e-05, | |
| "loss": 0.2558, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.276657060518732, | |
| "grad_norm": 13.105185508728027, | |
| "learning_rate": 1.5446685878962537e-05, | |
| "loss": 0.2432, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 2.3054755043227666, | |
| "grad_norm": 17.233076095581055, | |
| "learning_rate": 1.538904899135447e-05, | |
| "loss": 0.2817, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.3342939481268012, | |
| "grad_norm": 14.211281776428223, | |
| "learning_rate": 1.5331412103746398e-05, | |
| "loss": 0.2663, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.363112391930836, | |
| "grad_norm": 30.84634780883789, | |
| "learning_rate": 1.527377521613833e-05, | |
| "loss": 0.3084, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.39193083573487, | |
| "grad_norm": 29.224945068359375, | |
| "learning_rate": 1.521613832853026e-05, | |
| "loss": 0.3109, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.4207492795389047, | |
| "grad_norm": 3.674872398376465, | |
| "learning_rate": 1.5158501440922191e-05, | |
| "loss": 0.2346, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.4495677233429394, | |
| "grad_norm": 5.084238052368164, | |
| "learning_rate": 1.5100864553314123e-05, | |
| "loss": 0.2227, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.478386167146974, | |
| "grad_norm": 21.61268424987793, | |
| "learning_rate": 1.5043227665706052e-05, | |
| "loss": 0.2426, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.5072046109510087, | |
| "grad_norm": 12.604729652404785, | |
| "learning_rate": 1.4985590778097984e-05, | |
| "loss": 0.325, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.5360230547550433, | |
| "grad_norm": 25.42458152770996, | |
| "learning_rate": 1.4927953890489915e-05, | |
| "loss": 0.3394, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.564841498559078, | |
| "grad_norm": 9.900081634521484, | |
| "learning_rate": 1.4870317002881847e-05, | |
| "loss": 0.2565, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.5936599423631126, | |
| "grad_norm": 14.56777572631836, | |
| "learning_rate": 1.4812680115273776e-05, | |
| "loss": 0.2542, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.6224783861671472, | |
| "grad_norm": 27.645551681518555, | |
| "learning_rate": 1.4755043227665706e-05, | |
| "loss": 0.2753, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.6512968299711814, | |
| "grad_norm": 24.675256729125977, | |
| "learning_rate": 1.4697406340057639e-05, | |
| "loss": 0.2648, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.680115273775216, | |
| "grad_norm": 19.5800838470459, | |
| "learning_rate": 1.4639769452449568e-05, | |
| "loss": 0.2441, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.7089337175792507, | |
| "grad_norm": 32.76830291748047, | |
| "learning_rate": 1.45821325648415e-05, | |
| "loss": 0.2748, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.7377521613832854, | |
| "grad_norm": 9.661020278930664, | |
| "learning_rate": 1.452449567723343e-05, | |
| "loss": 0.2721, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.76657060518732, | |
| "grad_norm": 25.79545021057129, | |
| "learning_rate": 1.4466858789625363e-05, | |
| "loss": 0.2743, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.795389048991354, | |
| "grad_norm": 30.077226638793945, | |
| "learning_rate": 1.4409221902017291e-05, | |
| "loss": 0.2314, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.824207492795389, | |
| "grad_norm": 11.516585350036621, | |
| "learning_rate": 1.4351585014409224e-05, | |
| "loss": 0.2098, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.8530259365994235, | |
| "grad_norm": 12.071518898010254, | |
| "learning_rate": 1.4293948126801154e-05, | |
| "loss": 0.2447, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.881844380403458, | |
| "grad_norm": 1.786498785018921, | |
| "learning_rate": 1.4236311239193086e-05, | |
| "loss": 0.2374, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.910662824207493, | |
| "grad_norm": 31.069623947143555, | |
| "learning_rate": 1.4178674351585015e-05, | |
| "loss": 0.2667, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 2.9394812680115274, | |
| "grad_norm": 2.980510950088501, | |
| "learning_rate": 1.4121037463976946e-05, | |
| "loss": 0.28, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.968299711815562, | |
| "grad_norm": 18.496265411376953, | |
| "learning_rate": 1.4063400576368878e-05, | |
| "loss": 0.1831, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.9971181556195967, | |
| "grad_norm": 33.23817825317383, | |
| "learning_rate": 1.4005763688760807e-05, | |
| "loss": 0.3026, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9050295431618389, | |
| "eval_f1_macro": 0.9049667283676557, | |
| "eval_f1_weighted": 0.9049779714917217, | |
| "eval_loss": 0.3607948124408722, | |
| "eval_precision_macro": 0.9065337153272169, | |
| "eval_precision_weighted": 0.9065450713204785, | |
| "eval_recall_macro": 0.9050188737810804, | |
| "eval_recall_weighted": 0.9050295431618389, | |
| "eval_runtime": 29.9421, | |
| "eval_samples_per_second": 231.747, | |
| "eval_steps_per_second": 14.495, | |
| "step": 5205 | |
| }, | |
| { | |
| "epoch": 3.025936599423631, | |
| "grad_norm": 14.13509750366211, | |
| "learning_rate": 1.3948126801152739e-05, | |
| "loss": 0.1617, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 3.0547550432276656, | |
| "grad_norm": 0.5927883982658386, | |
| "learning_rate": 1.389048991354467e-05, | |
| "loss": 0.1693, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 3.0835734870317, | |
| "grad_norm": 14.065008163452148, | |
| "learning_rate": 1.3832853025936602e-05, | |
| "loss": 0.128, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 3.112391930835735, | |
| "grad_norm": 40.97966003417969, | |
| "learning_rate": 1.377521613832853e-05, | |
| "loss": 0.1558, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 3.1412103746397695, | |
| "grad_norm": 10.36765193939209, | |
| "learning_rate": 1.3717579250720463e-05, | |
| "loss": 0.207, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 3.170028818443804, | |
| "grad_norm": 0.5478718876838684, | |
| "learning_rate": 1.3659942363112394e-05, | |
| "loss": 0.1802, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 3.1988472622478388, | |
| "grad_norm": 0.6302068829536438, | |
| "learning_rate": 1.3602305475504324e-05, | |
| "loss": 0.14, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 3.227665706051873, | |
| "grad_norm": 2.4789652824401855, | |
| "learning_rate": 1.3544668587896255e-05, | |
| "loss": 0.1137, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 3.2564841498559076, | |
| "grad_norm": 22.161422729492188, | |
| "learning_rate": 1.3487031700288185e-05, | |
| "loss": 0.2182, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 3.2853025936599423, | |
| "grad_norm": 28.67848014831543, | |
| "learning_rate": 1.3429394812680117e-05, | |
| "loss": 0.1741, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 3.314121037463977, | |
| "grad_norm": 13.24758243560791, | |
| "learning_rate": 1.3371757925072046e-05, | |
| "loss": 0.2205, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 3.3429394812680115, | |
| "grad_norm": 0.6006250977516174, | |
| "learning_rate": 1.3314121037463979e-05, | |
| "loss": 0.1419, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 3.371757925072046, | |
| "grad_norm": 4.931090831756592, | |
| "learning_rate": 1.3256484149855909e-05, | |
| "loss": 0.1466, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 3.400576368876081, | |
| "grad_norm": 17.918506622314453, | |
| "learning_rate": 1.319884726224784e-05, | |
| "loss": 0.1437, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 3.4293948126801155, | |
| "grad_norm": 18.44112205505371, | |
| "learning_rate": 1.314121037463977e-05, | |
| "loss": 0.1923, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 3.4582132564841497, | |
| "grad_norm": 36.937828063964844, | |
| "learning_rate": 1.3083573487031702e-05, | |
| "loss": 0.2021, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.4870317002881843, | |
| "grad_norm": 0.9251816868782043, | |
| "learning_rate": 1.3025936599423631e-05, | |
| "loss": 0.2166, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 3.515850144092219, | |
| "grad_norm": 29.732206344604492, | |
| "learning_rate": 1.2968299711815563e-05, | |
| "loss": 0.1772, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 3.5446685878962536, | |
| "grad_norm": 8.258246421813965, | |
| "learning_rate": 1.2910662824207494e-05, | |
| "loss": 0.2043, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 3.5734870317002883, | |
| "grad_norm": 0.8275717496871948, | |
| "learning_rate": 1.2853025936599423e-05, | |
| "loss": 0.1502, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 3.602305475504323, | |
| "grad_norm": 27.115209579467773, | |
| "learning_rate": 1.2795389048991355e-05, | |
| "loss": 0.1733, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 3.631123919308357, | |
| "grad_norm": 0.7643899321556091, | |
| "learning_rate": 1.2737752161383286e-05, | |
| "loss": 0.1349, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 3.6599423631123917, | |
| "grad_norm": 33.05510330200195, | |
| "learning_rate": 1.2680115273775218e-05, | |
| "loss": 0.1526, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 3.6887608069164264, | |
| "grad_norm": 13.939464569091797, | |
| "learning_rate": 1.2622478386167147e-05, | |
| "loss": 0.1616, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 3.717579250720461, | |
| "grad_norm": 2.4451797008514404, | |
| "learning_rate": 1.2564841498559079e-05, | |
| "loss": 0.1751, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 3.7463976945244957, | |
| "grad_norm": 17.99618148803711, | |
| "learning_rate": 1.250720461095101e-05, | |
| "loss": 0.2061, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 3.7752161383285303, | |
| "grad_norm": 19.82054328918457, | |
| "learning_rate": 1.2449567723342942e-05, | |
| "loss": 0.2168, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 3.804034582132565, | |
| "grad_norm": 17.200483322143555, | |
| "learning_rate": 1.239193083573487e-05, | |
| "loss": 0.1849, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 3.8328530259365996, | |
| "grad_norm": 9.288723945617676, | |
| "learning_rate": 1.2334293948126803e-05, | |
| "loss": 0.1902, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 3.8616714697406342, | |
| "grad_norm": 52.95684051513672, | |
| "learning_rate": 1.2276657060518733e-05, | |
| "loss": 0.1566, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 3.8904899135446684, | |
| "grad_norm": 22.813720703125, | |
| "learning_rate": 1.2219020172910662e-05, | |
| "loss": 0.1212, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 3.919308357348703, | |
| "grad_norm": 3.377829074859619, | |
| "learning_rate": 1.2161383285302594e-05, | |
| "loss": 0.1147, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 3.9481268011527377, | |
| "grad_norm": 0.6791939735412598, | |
| "learning_rate": 1.2103746397694525e-05, | |
| "loss": 0.1678, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 3.9769452449567724, | |
| "grad_norm": 12.582691192626953, | |
| "learning_rate": 1.2046109510086457e-05, | |
| "loss": 0.2679, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.9279435077100446, | |
| "eval_f1_macro": 0.927584474052213, | |
| "eval_f1_weighted": 0.927592833052565, | |
| "eval_loss": 0.35773107409477234, | |
| "eval_precision_macro": 0.9276877579070416, | |
| "eval_precision_weighted": 0.9276960791160526, | |
| "eval_recall_macro": 0.9279352304379973, | |
| "eval_recall_weighted": 0.9279435077100446, | |
| "eval_runtime": 30.2758, | |
| "eval_samples_per_second": 229.193, | |
| "eval_steps_per_second": 14.335, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 4.005763688760807, | |
| "grad_norm": 0.19622278213500977, | |
| "learning_rate": 1.1988472622478386e-05, | |
| "loss": 0.1063, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 4.034582132564841, | |
| "grad_norm": 2.3306796550750732, | |
| "learning_rate": 1.1930835734870318e-05, | |
| "loss": 0.1007, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 4.063400576368876, | |
| "grad_norm": 36.256927490234375, | |
| "learning_rate": 1.1873198847262249e-05, | |
| "loss": 0.0781, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 4.0922190201729105, | |
| "grad_norm": 13.904011726379395, | |
| "learning_rate": 1.1815561959654181e-05, | |
| "loss": 0.1152, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 4.121037463976945, | |
| "grad_norm": 2.325575828552246, | |
| "learning_rate": 1.175792507204611e-05, | |
| "loss": 0.1033, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 4.14985590778098, | |
| "grad_norm": 0.9815200567245483, | |
| "learning_rate": 1.1700288184438042e-05, | |
| "loss": 0.0964, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 4.178674351585014, | |
| "grad_norm": 32.72802734375, | |
| "learning_rate": 1.1642651296829973e-05, | |
| "loss": 0.1008, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 4.207492795389049, | |
| "grad_norm": 19.83048439025879, | |
| "learning_rate": 1.1585014409221902e-05, | |
| "loss": 0.1044, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 4.236311239193084, | |
| "grad_norm": 22.580406188964844, | |
| "learning_rate": 1.1527377521613834e-05, | |
| "loss": 0.1528, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 4.265129682997118, | |
| "grad_norm": 2.180345296859741, | |
| "learning_rate": 1.1469740634005764e-05, | |
| "loss": 0.1219, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 4.293948126801153, | |
| "grad_norm": 8.74466609954834, | |
| "learning_rate": 1.1412103746397697e-05, | |
| "loss": 0.0763, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 4.322766570605188, | |
| "grad_norm": 0.5323246717453003, | |
| "learning_rate": 1.1354466858789625e-05, | |
| "loss": 0.0875, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 4.351585014409222, | |
| "grad_norm": 0.2620614171028137, | |
| "learning_rate": 1.1296829971181558e-05, | |
| "loss": 0.1046, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 4.380403458213257, | |
| "grad_norm": 7.299178123474121, | |
| "learning_rate": 1.1239193083573488e-05, | |
| "loss": 0.1218, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 4.409221902017291, | |
| "grad_norm": 12.042703628540039, | |
| "learning_rate": 1.1181556195965419e-05, | |
| "loss": 0.1083, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 4.438040345821325, | |
| "grad_norm": 0.6515465378761292, | |
| "learning_rate": 1.112391930835735e-05, | |
| "loss": 0.0987, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 4.46685878962536, | |
| "grad_norm": 4.6855316162109375, | |
| "learning_rate": 1.1066282420749282e-05, | |
| "loss": 0.0983, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 4.495677233429395, | |
| "grad_norm": 23.7063045501709, | |
| "learning_rate": 1.100864553314121e-05, | |
| "loss": 0.0989, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 4.524495677233429, | |
| "grad_norm": 1.5553531646728516, | |
| "learning_rate": 1.0951008645533141e-05, | |
| "loss": 0.1309, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 4.553314121037464, | |
| "grad_norm": 32.84361267089844, | |
| "learning_rate": 1.0893371757925073e-05, | |
| "loss": 0.1097, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 4.582132564841499, | |
| "grad_norm": 70.86231231689453, | |
| "learning_rate": 1.0835734870317004e-05, | |
| "loss": 0.1408, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 4.610951008645533, | |
| "grad_norm": 0.31812503933906555, | |
| "learning_rate": 1.0778097982708934e-05, | |
| "loss": 0.1226, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 4.639769452449568, | |
| "grad_norm": 39.376468658447266, | |
| "learning_rate": 1.0720461095100865e-05, | |
| "loss": 0.1727, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 4.6685878962536025, | |
| "grad_norm": 28.209671020507812, | |
| "learning_rate": 1.0662824207492797e-05, | |
| "loss": 0.1426, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 4.697406340057637, | |
| "grad_norm": 30.422863006591797, | |
| "learning_rate": 1.0605187319884726e-05, | |
| "loss": 0.1224, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 4.726224783861672, | |
| "grad_norm": 0.9933204054832458, | |
| "learning_rate": 1.0547550432276658e-05, | |
| "loss": 0.0922, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 4.755043227665706, | |
| "grad_norm": 0.3882824182510376, | |
| "learning_rate": 1.0489913544668589e-05, | |
| "loss": 0.1163, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 4.78386167146974, | |
| "grad_norm": 0.04565083980560303, | |
| "learning_rate": 1.0432276657060521e-05, | |
| "loss": 0.1135, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 4.812680115273775, | |
| "grad_norm": 34.51498794555664, | |
| "learning_rate": 1.037463976945245e-05, | |
| "loss": 0.0817, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 4.8414985590778095, | |
| "grad_norm": 23.297779083251953, | |
| "learning_rate": 1.031700288184438e-05, | |
| "loss": 0.1033, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 4.870317002881844, | |
| "grad_norm": 0.41480186581611633, | |
| "learning_rate": 1.0259365994236313e-05, | |
| "loss": 0.107, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 4.899135446685879, | |
| "grad_norm": 49.01393508911133, | |
| "learning_rate": 1.0201729106628241e-05, | |
| "loss": 0.1095, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 4.927953890489913, | |
| "grad_norm": 0.21958515048027039, | |
| "learning_rate": 1.0144092219020174e-05, | |
| "loss": 0.0695, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 4.956772334293948, | |
| "grad_norm": 0.6442630290985107, | |
| "learning_rate": 1.0086455331412104e-05, | |
| "loss": 0.0979, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 4.985590778097983, | |
| "grad_norm": 0.04263289272785187, | |
| "learning_rate": 1.0028818443804036e-05, | |
| "loss": 0.1149, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.933852140077821, | |
| "eval_f1_macro": 0.9335300711231832, | |
| "eval_f1_weighted": 0.9335363129664682, | |
| "eval_loss": 0.3687053918838501, | |
| "eval_precision_macro": 0.9336924382488527, | |
| "eval_precision_weighted": 0.9336961352032257, | |
| "eval_recall_macro": 0.9338433865620074, | |
| "eval_recall_weighted": 0.933852140077821, | |
| "eval_runtime": 29.8558, | |
| "eval_samples_per_second": 232.417, | |
| "eval_steps_per_second": 14.537, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 5.014409221902017, | |
| "grad_norm": 4.232170581817627, | |
| "learning_rate": 9.971181556195965e-06, | |
| "loss": 0.1073, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 5.043227665706052, | |
| "grad_norm": 83.36505126953125, | |
| "learning_rate": 9.913544668587897e-06, | |
| "loss": 0.0736, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 5.072046109510087, | |
| "grad_norm": 0.28978821635246277, | |
| "learning_rate": 9.855907780979828e-06, | |
| "loss": 0.0804, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 5.100864553314121, | |
| "grad_norm": 11.236791610717773, | |
| "learning_rate": 9.798270893371759e-06, | |
| "loss": 0.0989, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 5.129682997118156, | |
| "grad_norm": 0.07707870006561279, | |
| "learning_rate": 9.740634005763689e-06, | |
| "loss": 0.0519, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 5.1585014409221905, | |
| "grad_norm": 0.45296791195869446, | |
| "learning_rate": 9.68299711815562e-06, | |
| "loss": 0.0403, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 5.187319884726225, | |
| "grad_norm": 46.82713317871094, | |
| "learning_rate": 9.625360230547552e-06, | |
| "loss": 0.0895, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 5.216138328530259, | |
| "grad_norm": 6.646805286407471, | |
| "learning_rate": 9.567723342939482e-06, | |
| "loss": 0.0419, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 5.244956772334294, | |
| "grad_norm": 21.99289321899414, | |
| "learning_rate": 9.510086455331413e-06, | |
| "loss": 0.0573, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 5.273775216138328, | |
| "grad_norm": 39.594390869140625, | |
| "learning_rate": 9.452449567723344e-06, | |
| "loss": 0.0585, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 5.302593659942363, | |
| "grad_norm": 18.201231002807617, | |
| "learning_rate": 9.394812680115276e-06, | |
| "loss": 0.1113, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 5.3314121037463975, | |
| "grad_norm": 30.270816802978516, | |
| "learning_rate": 9.337175792507205e-06, | |
| "loss": 0.0531, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 5.360230547550432, | |
| "grad_norm": 42.12540817260742, | |
| "learning_rate": 9.279538904899135e-06, | |
| "loss": 0.0975, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 5.389048991354467, | |
| "grad_norm": 0.4828750193119049, | |
| "learning_rate": 9.221902017291067e-06, | |
| "loss": 0.0636, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 5.417867435158501, | |
| "grad_norm": 21.690710067749023, | |
| "learning_rate": 9.164265129682998e-06, | |
| "loss": 0.042, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 5.446685878962536, | |
| "grad_norm": 43.906761169433594, | |
| "learning_rate": 9.106628242074928e-06, | |
| "loss": 0.0887, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 5.475504322766571, | |
| "grad_norm": 37.99407958984375, | |
| "learning_rate": 9.048991354466859e-06, | |
| "loss": 0.0444, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 5.504322766570605, | |
| "grad_norm": 0.9924225807189941, | |
| "learning_rate": 8.991354466858791e-06, | |
| "loss": 0.0893, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 5.53314121037464, | |
| "grad_norm": 0.024848056957125664, | |
| "learning_rate": 8.933717579250722e-06, | |
| "loss": 0.0701, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 5.561959654178675, | |
| "grad_norm": 0.013018026947975159, | |
| "learning_rate": 8.876080691642652e-06, | |
| "loss": 0.1031, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 5.590778097982709, | |
| "grad_norm": 0.36339619755744934, | |
| "learning_rate": 8.818443804034583e-06, | |
| "loss": 0.0422, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 5.619596541786743, | |
| "grad_norm": 0.5731251835823059, | |
| "learning_rate": 8.760806916426513e-06, | |
| "loss": 0.0666, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 5.648414985590778, | |
| "grad_norm": 3.643000841140747, | |
| "learning_rate": 8.703170028818444e-06, | |
| "loss": 0.0817, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 5.677233429394812, | |
| "grad_norm": 0.039307739585638046, | |
| "learning_rate": 8.645533141210375e-06, | |
| "loss": 0.1151, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 5.706051873198847, | |
| "grad_norm": 0.2764396667480469, | |
| "learning_rate": 8.587896253602305e-06, | |
| "loss": 0.1021, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 5.734870317002882, | |
| "grad_norm": 46.3472785949707, | |
| "learning_rate": 8.530259365994237e-06, | |
| "loss": 0.0848, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 5.763688760806916, | |
| "grad_norm": 0.2657397389411926, | |
| "learning_rate": 8.472622478386168e-06, | |
| "loss": 0.0904, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 5.792507204610951, | |
| "grad_norm": 30.730506896972656, | |
| "learning_rate": 8.414985590778098e-06, | |
| "loss": 0.0715, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 5.821325648414986, | |
| "grad_norm": 2.388108253479004, | |
| "learning_rate": 8.357348703170029e-06, | |
| "loss": 0.0474, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 5.85014409221902, | |
| "grad_norm": 0.06589208543300629, | |
| "learning_rate": 8.299711815561961e-06, | |
| "loss": 0.0718, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 5.878962536023055, | |
| "grad_norm": 0.017373552545905113, | |
| "learning_rate": 8.242074927953892e-06, | |
| "loss": 0.0851, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 5.9077809798270895, | |
| "grad_norm": 1.8382971286773682, | |
| "learning_rate": 8.184438040345822e-06, | |
| "loss": 0.0771, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 5.936599423631124, | |
| "grad_norm": 0.5171680450439453, | |
| "learning_rate": 8.126801152737753e-06, | |
| "loss": 0.0352, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 5.965417867435159, | |
| "grad_norm": 0.052714597433805466, | |
| "learning_rate": 8.069164265129685e-06, | |
| "loss": 0.0541, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 5.994236311239193, | |
| "grad_norm": 55.4918327331543, | |
| "learning_rate": 8.011527377521614e-06, | |
| "loss": 0.0491, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.9348609309698804, | |
| "eval_f1_macro": 0.9339829958075209, | |
| "eval_f1_weighted": 0.9339902153482186, | |
| "eval_loss": 0.4194980561733246, | |
| "eval_precision_macro": 0.9352763632278608, | |
| "eval_precision_weighted": 0.9352802017792717, | |
| "eval_recall_macro": 0.9348508698098554, | |
| "eval_recall_weighted": 0.9348609309698804, | |
| "eval_runtime": 29.9694, | |
| "eval_samples_per_second": 231.536, | |
| "eval_steps_per_second": 14.481, | |
| "step": 10410 | |
| }, | |
| { | |
| "epoch": 6.023054755043228, | |
| "grad_norm": 0.13180404901504517, | |
| "learning_rate": 7.953890489913544e-06, | |
| "loss": 0.0896, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 6.051873198847262, | |
| "grad_norm": 14.752634048461914, | |
| "learning_rate": 7.896253602305477e-06, | |
| "loss": 0.0528, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 6.0806916426512965, | |
| "grad_norm": 0.12015581876039505, | |
| "learning_rate": 7.838616714697407e-06, | |
| "loss": 0.0762, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 6.109510086455331, | |
| "grad_norm": 0.8792430758476257, | |
| "learning_rate": 7.780979827089338e-06, | |
| "loss": 0.0199, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 6.138328530259366, | |
| "grad_norm": 0.15485620498657227, | |
| "learning_rate": 7.723342939481268e-06, | |
| "loss": 0.0448, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 6.1671469740634, | |
| "grad_norm": 24.48517417907715, | |
| "learning_rate": 7.665706051873199e-06, | |
| "loss": 0.0477, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 6.195965417867435, | |
| "grad_norm": 0.01565726287662983, | |
| "learning_rate": 7.60806916426513e-06, | |
| "loss": 0.0477, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 6.22478386167147, | |
| "grad_norm": 0.06371276825666428, | |
| "learning_rate": 7.550432276657062e-06, | |
| "loss": 0.072, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 6.253602305475504, | |
| "grad_norm": 0.036789000034332275, | |
| "learning_rate": 7.492795389048992e-06, | |
| "loss": 0.024, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 6.282420749279539, | |
| "grad_norm": 10.600634574890137, | |
| "learning_rate": 7.4351585014409235e-06, | |
| "loss": 0.0583, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 6.311239193083574, | |
| "grad_norm": 0.00421316921710968, | |
| "learning_rate": 7.377521613832853e-06, | |
| "loss": 0.0416, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 6.340057636887608, | |
| "grad_norm": 43.42197799682617, | |
| "learning_rate": 7.319884726224784e-06, | |
| "loss": 0.0642, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 6.368876080691643, | |
| "grad_norm": 0.05901940539479256, | |
| "learning_rate": 7.262247838616715e-06, | |
| "loss": 0.0511, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 6.3976945244956775, | |
| "grad_norm": 0.022665705531835556, | |
| "learning_rate": 7.204610951008646e-06, | |
| "loss": 0.0076, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 6.426512968299712, | |
| "grad_norm": 1.6211119890213013, | |
| "learning_rate": 7.146974063400577e-06, | |
| "loss": 0.0401, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 6.455331412103746, | |
| "grad_norm": 0.006085489876568317, | |
| "learning_rate": 7.089337175792508e-06, | |
| "loss": 0.0732, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 6.484149855907781, | |
| "grad_norm": 44.77919006347656, | |
| "learning_rate": 7.031700288184439e-06, | |
| "loss": 0.0634, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 6.512968299711815, | |
| "grad_norm": 13.231744766235352, | |
| "learning_rate": 6.9740634005763696e-06, | |
| "loss": 0.0429, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 6.54178674351585, | |
| "grad_norm": 0.025533217936754227, | |
| "learning_rate": 6.916426512968301e-06, | |
| "loss": 0.0301, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 6.5706051873198845, | |
| "grad_norm": 0.014192778617143631, | |
| "learning_rate": 6.8587896253602315e-06, | |
| "loss": 0.0588, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 6.599423631123919, | |
| "grad_norm": 0.12580570578575134, | |
| "learning_rate": 6.801152737752162e-06, | |
| "loss": 0.0463, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 6.628242074927954, | |
| "grad_norm": 27.74668312072754, | |
| "learning_rate": 6.743515850144093e-06, | |
| "loss": 0.0801, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 6.6570605187319885, | |
| "grad_norm": 0.016943486407399178, | |
| "learning_rate": 6.685878962536023e-06, | |
| "loss": 0.0563, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 6.685878962536023, | |
| "grad_norm": 0.41584333777427673, | |
| "learning_rate": 6.6282420749279545e-06, | |
| "loss": 0.043, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 6.714697406340058, | |
| "grad_norm": 0.15631648898124695, | |
| "learning_rate": 6.570605187319885e-06, | |
| "loss": 0.0479, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 6.743515850144092, | |
| "grad_norm": 0.560581386089325, | |
| "learning_rate": 6.512968299711816e-06, | |
| "loss": 0.0374, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 6.772334293948127, | |
| "grad_norm": 0.1623823046684265, | |
| "learning_rate": 6.455331412103747e-06, | |
| "loss": 0.0402, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 6.801152737752162, | |
| "grad_norm": 0.0034744683653116226, | |
| "learning_rate": 6.3976945244956775e-06, | |
| "loss": 0.0653, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 6.829971181556196, | |
| "grad_norm": 11.196998596191406, | |
| "learning_rate": 6.340057636887609e-06, | |
| "loss": 0.0453, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 6.858789625360231, | |
| "grad_norm": 11.29255199432373, | |
| "learning_rate": 6.2824207492795395e-06, | |
| "loss": 0.0774, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 6.887608069164266, | |
| "grad_norm": 0.0062851207330822945, | |
| "learning_rate": 6.224783861671471e-06, | |
| "loss": 0.0354, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 6.916426512968299, | |
| "grad_norm": 0.030905550345778465, | |
| "learning_rate": 6.167146974063401e-06, | |
| "loss": 0.0353, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 6.945244956772334, | |
| "grad_norm": 0.2567192018032074, | |
| "learning_rate": 6.109510086455331e-06, | |
| "loss": 0.0248, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 6.974063400576369, | |
| "grad_norm": 0.03146115690469742, | |
| "learning_rate": 6.0518731988472625e-06, | |
| "loss": 0.052, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.9383196425997983, | |
| "eval_f1_macro": 0.9376382344287736, | |
| "eval_f1_weighted": 0.9376445510679586, | |
| "eval_loss": 0.4325847625732422, | |
| "eval_precision_macro": 0.938292936865987, | |
| "eval_precision_weighted": 0.9382956472792613, | |
| "eval_recall_macro": 0.9383099973494167, | |
| "eval_recall_weighted": 0.9383196425997983, | |
| "eval_runtime": 30.0017, | |
| "eval_samples_per_second": 231.287, | |
| "eval_steps_per_second": 14.466, | |
| "step": 12145 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 17350, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 2.55585712669056e+16, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |