variant-tapt_base-LR_2e-05

This model is a fine-tuned version of microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext on the Mardiyyah/TAPT_Variant_FT dataset. It achieves the following results on the evaluation set:

  • Loss: 1.4987
  • Accuracy: 0.7109

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: 16
  • eval_batch_size: 16
  • seed: 3407
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 32
  • optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-06 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.06
  • num_epochs: 50
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Accuracy
1.7258 1.0 19 1.7873 0.6920
1.708 2.0 38 1.7398 0.6962
1.6506 3.0 57 1.6802 0.6973
1.5883 4.0 76 1.6648 0.7041
1.567 5.0 95 1.6342 0.7030
1.5485 6.0 114 1.5430 0.7146
1.5105 7.0 133 1.5296 0.7113
1.4635 8.0 152 1.6214 0.7078
1.4841 9.0 171 1.5212 0.7120
1.4663 10.0 190 1.5628 0.7035
1.4282 11.0 209 1.5351 0.7165
1.4511 12.0 228 1.5300 0.7095
1.4318 13.0 247 1.5256 0.7148
1.4241 14.0 266 1.4872 0.7146
1.4235 15.0 285 1.5431 0.7088
1.3905 16.0 304 1.5831 0.7096
1.3526 17.0 323 1.4920 0.7175
1.3733 18.0 342 1.5018 0.7104
1.3673 19.0 361 1.4766 0.7180
1.3631 20.0 380 1.4878 0.7142
1.3709 21.0 399 1.5422 0.7039
1.3408 22.0 418 1.4855 0.7206
1.3311 23.0 437 1.5095 0.7157
1.3144 24.0 456 1.5173 0.7157
1.297 25.0 475 1.4743 0.7215
1.3343 26.0 494 1.5012 0.7113
1.2949 27.0 513 1.4988 0.7146
1.3182 28.0 532 1.4198 0.7242
1.3005 29.0 551 1.4724 0.7161
1.2821 30.0 570 1.4705 0.7205
1.278 31.0 589 1.4780 0.7201
1.274 32.0 608 1.5008 0.7129
1.2849 33.0 627 1.4571 0.7200
1.2607 34.0 646 1.4253 0.7247
1.2673 35.0 665 1.5112 0.7101
1.259 36.0 684 1.5094 0.7149
1.2348 37.0 703 1.4844 0.7216
1.2561 38.0 722 1.4628 0.7171
1.2464 39.0 741 1.4711 0.7183
1.2483 40.0 760 1.4617 0.7228
1.2392 41.0 779 1.4650 0.7165
1.2306 42.0 798 1.4046 0.7259
1.2328 43.0 817 1.4773 0.7141
1.2493 44.0 836 1.4506 0.7229
1.2349 45.0 855 1.5113 0.7073
1.2352 46.0 874 1.4787 0.7155
1.2469 47.0 893 1.4405 0.7175
1.2215 48.0 912 1.4719 0.7176
1.2238 49.0 931 1.4799 0.7195
1.2371 50.0 950 1.4882 0.7123

Framework versions

  • Transformers 4.48.2
  • Pytorch 2.4.1+cu121
  • Datasets 3.0.2
  • Tokenizers 0.21.0
Downloads last month
-
Safetensors
Model size
0.1B params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for Mardiyyah/variant-tapt_base-LR_2e-05