Edit model card

2_5e-3_10_0.1

This model is a fine-tuned version of bert-large-uncased on the super_glue dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6847
  • Accuracy: 0.7226

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 0.005
  • train_batch_size: 16
  • eval_batch_size: 8
  • seed: 11
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 60.0

Training results

Training Loss Epoch Step Validation Loss Accuracy
1.2209 1.0 590 0.9564 0.6162
1.1661 2.0 1180 0.9456 0.5817
1.1103 3.0 1770 0.9574 0.6214
1.0789 4.0 2360 0.9671 0.6217
1.0422 5.0 2950 1.0276 0.4997
0.9949 6.0 3540 0.8934 0.6312
0.99 7.0 4130 1.5786 0.4119
0.9632 8.0 4720 1.2903 0.6232
0.9329 9.0 5310 0.8528 0.6352
0.9157 10.0 5900 0.8400 0.6557
0.9187 11.0 6490 0.9022 0.6404
0.8408 12.0 7080 0.8227 0.6679
0.8295 13.0 7670 1.4711 0.5606
0.9554 14.0 8260 0.8134 0.6884
0.7759 15.0 8850 0.7988 0.6774
0.7568 16.0 9440 0.9273 0.6031
0.7197 17.0 10030 0.7468 0.6966
0.739 18.0 10620 0.7418 0.6976
0.725 19.0 11210 0.7303 0.7043
0.7215 20.0 11800 0.7322 0.7024
0.7028 21.0 12390 0.7489 0.7073
0.6929 22.0 12980 0.7376 0.7125
0.6907 23.0 13570 0.7165 0.7122
0.6862 24.0 14160 0.7102 0.7101
0.6583 25.0 14750 0.7060 0.7193
0.6713 26.0 15340 0.7305 0.6905
0.6625 27.0 15930 0.7407 0.6914
0.6516 28.0 16520 0.7057 0.7232
0.6465 29.0 17110 0.7047 0.7135
0.6389 30.0 17700 0.7340 0.7272
0.6333 31.0 18290 0.7067 0.7055
0.6212 32.0 18880 0.7071 0.7235
0.6179 33.0 19470 0.6851 0.7202
0.5935 34.0 20060 0.6888 0.7187
0.5851 35.0 20650 0.7105 0.6985
0.5921 36.0 21240 0.6810 0.7284
0.5838 37.0 21830 0.6814 0.7315
0.5746 38.0 22420 0.6984 0.7086
0.5744 39.0 23010 0.6864 0.7214
0.5628 40.0 23600 0.6842 0.7260
0.5694 41.0 24190 0.7091 0.7083
0.5595 42.0 24780 0.6805 0.7214
0.5552 43.0 25370 0.6899 0.7321
0.5553 44.0 25960 0.7324 0.7021
0.5439 45.0 26550 0.6960 0.7122
0.5328 46.0 27140 0.6965 0.7131
0.5367 47.0 27730 0.6844 0.7257
0.5377 48.0 28320 0.6752 0.7275
0.5364 49.0 28910 0.6861 0.7165
0.5224 50.0 29500 0.6903 0.7153
0.5239 51.0 30090 0.6895 0.7202
0.5259 52.0 30680 0.6885 0.7162
0.5235 53.0 31270 0.6772 0.7281
0.5227 54.0 31860 0.7113 0.7141
0.5176 55.0 32450 0.6802 0.7266
0.5116 56.0 33040 0.6807 0.7284
0.5029 57.0 33630 0.6786 0.7239
0.5068 58.0 34220 0.6862 0.7226
0.498 59.0 34810 0.6838 0.7251
0.5037 60.0 35400 0.6847 0.7226

Framework versions

  • Transformers 4.30.0
  • Pytorch 2.0.1+cu117
  • Datasets 2.14.4
  • Tokenizers 0.13.3
Downloads last month
4
Inference API
This model can be loaded on Inference API (serverless).

Dataset used to train Onutoa/2_5e-3_10_0.1