Edit model card

deberta-v3-large__sst2__train-16-9

This model is a fine-tuned version of microsoft/deberta-v3-large on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 1.2598
  • Accuracy: 0.7809

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: 2e-05
  • train_batch_size: 4
  • eval_batch_size: 4
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 50
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Accuracy
0.6887 1.0 7 0.7452 0.2857
0.6889 2.0 14 0.7988 0.2857
0.6501 3.0 21 0.8987 0.2857
0.4286 4.0 28 0.9186 0.4286
0.3591 5.0 35 0.5566 0.7143
0.0339 6.0 42 1.1130 0.5714
0.013 7.0 49 1.8296 0.7143
0.0041 8.0 56 1.7069 0.7143
0.0023 9.0 63 1.1942 0.7143
0.0022 10.0 70 0.6054 0.7143
0.0011 11.0 77 0.3872 0.7143
0.0006 12.0 84 0.3217 0.7143
0.0005 13.0 91 0.2879 0.8571
0.0005 14.0 98 0.2640 0.8571
0.0004 15.0 105 0.2531 0.8571
0.0003 16.0 112 0.2384 0.8571
0.0004 17.0 119 0.2338 0.8571
0.0003 18.0 126 0.2314 0.8571
0.0003 19.0 133 0.2276 0.8571
0.0003 20.0 140 0.2172 0.8571
0.0003 21.0 147 0.2069 0.8571
0.0002 22.0 154 0.2018 0.8571
0.0002 23.0 161 0.2005 0.8571
0.0002 24.0 168 0.1985 0.8571
0.0002 25.0 175 0.1985 1.0
0.0002 26.0 182 0.1955 1.0
0.0002 27.0 189 0.1967 1.0
0.0002 28.0 196 0.1918 1.0
0.0002 29.0 203 0.1888 1.0
0.0002 30.0 210 0.1864 1.0
0.0002 31.0 217 0.1870 1.0
0.0002 32.0 224 0.1892 1.0
0.0002 33.0 231 0.1917 1.0
0.0002 34.0 238 0.1869 1.0
0.0002 35.0 245 0.1812 1.0
0.0001 36.0 252 0.1777 1.0
0.0002 37.0 259 0.1798 1.0
0.0002 38.0 266 0.1824 0.8571
0.0002 39.0 273 0.1846 0.8571
0.0002 40.0 280 0.1839 0.8571
0.0001 41.0 287 0.1826 0.8571
0.0001 42.0 294 0.1779 0.8571
0.0002 43.0 301 0.1762 0.8571
0.0001 44.0 308 0.1742 1.0
0.0002 45.0 315 0.1708 1.0
0.0001 46.0 322 0.1702 1.0
0.0001 47.0 329 0.1699 1.0
0.0001 48.0 336 0.1695 1.0
0.0001 49.0 343 0.1683 1.0
0.0001 50.0 350 0.1681 1.0

Framework versions

  • Transformers 4.15.0
  • Pytorch 1.10.2+cu102
  • Datasets 1.18.2
  • Tokenizers 0.10.3
Downloads last month
16

Finetuned from