CeLLaTe-tapt-bioformer16l-tokenizer-adapted-baseline-combinedData
This model is a fine-tuned version of bioformers/bioformer-16l on the Mardiyyah/Combined_TAPT_dataset dataset. It achieves the following results on the evaluation set:
- Loss: 1.3221
- Accuracy: 0.7303
- Perplexity: 3.7513
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: 3e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 3407
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- 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: 100
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Perplexity |
|---|---|---|---|---|---|
| 1.8754 | 1.0 | 30 | 1.6820 | 0.7051 | 5.3764 |
| 1.7991 | 2.0 | 60 | 1.5862 | 0.7118 | 4.8850 |
| 1.7464 | 3.0 | 90 | 1.5605 | 0.7125 | 4.7614 |
| 1.7094 | 4.0 | 120 | 1.4866 | 0.7219 | 4.4222 |
| 1.6502 | 5.0 | 150 | 1.4803 | 0.7198 | 4.3942 |
| 1.6085 | 6.0 | 180 | 1.4975 | 0.7180 | 4.4704 |
| 1.5739 | 7.0 | 210 | 1.4710 | 0.7163 | 4.3535 |
| 1.5157 | 8.0 | 240 | 1.4215 | 0.7250 | 4.1435 |
| 1.4978 | 9.0 | 270 | 1.4305 | 0.7228 | 4.1807 |
| 1.4813 | 10.0 | 300 | 1.4162 | 0.7253 | 4.1213 |
| 1.4611 | 11.0 | 330 | 1.3819 | 0.7264 | 3.9826 |
| 1.453 | 12.0 | 360 | 1.3862 | 0.7285 | 3.9997 |
| 1.4432 | 13.0 | 390 | 1.3699 | 0.7343 | 3.9351 |
| 1.3998 | 14.0 | 420 | 1.3644 | 0.7299 | 3.9134 |
| 1.4128 | 15.0 | 450 | 1.3938 | 0.7282 | 4.0303 |
| 1.3744 | 16.0 | 480 | 1.3660 | 0.7286 | 3.9196 |
| 1.3798 | 17.0 | 510 | 1.3708 | 0.7301 | 3.9385 |
| 1.3572 | 18.0 | 540 | 1.3453 | 0.7355 | 3.8393 |
| 1.3587 | 19.0 | 570 | 1.3519 | 0.7328 | 3.8648 |
| 1.331 | 20.0 | 600 | 1.3408 | 0.7323 | 3.8219 |
| 1.3222 | 21.0 | 630 | 1.3611 | 0.7311 | 3.9003 |
| 1.3242 | 22.0 | 660 | 1.3513 | 0.7302 | 3.8626 |
| 1.3203 | 23.0 | 690 | 1.3528 | 0.7290 | 3.8683 |
| 1.2956 | 24.0 | 720 | 1.3827 | 0.7279 | 3.9858 |
| 1.2845 | 25.0 | 750 | 1.3308 | 0.7304 | 3.7843 |
| 1.2831 | 26.0 | 780 | 1.3218 | 0.7351 | 3.7501 |
| 1.2819 | 27.0 | 810 | 1.3448 | 0.7323 | 3.8376 |
| 1.2562 | 28.0 | 840 | 1.3406 | 0.7293 | 3.8215 |
| 1.2574 | 29.0 | 870 | 1.3337 | 0.7328 | 3.7950 |
| 1.2548 | 30.0 | 900 | 1.3226 | 0.7325 | 3.7531 |
| 1.2539 | 31.0 | 930 | 1.3170 | 0.7364 | 3.7322 |
| 1.2383 | 32.0 | 960 | 1.3265 | 0.7317 | 3.7680 |
| 1.2345 | 33.0 | 990 | 1.3253 | 0.7350 | 3.7632 |
| 1.2243 | 34.0 | 1020 | 1.2938 | 0.7365 | 3.6468 |
| 1.2288 | 35.0 | 1050 | 1.3150 | 0.7346 | 3.7248 |
| 1.2158 | 36.0 | 1080 | 1.3086 | 0.7355 | 3.7011 |
| 1.1943 | 37.0 | 1110 | 1.3148 | 0.7360 | 3.7239 |
| 1.2043 | 38.0 | 1140 | 1.3342 | 0.7329 | 3.7968 |
| 1.1872 | 39.0 | 1170 | 1.3314 | 0.7340 | 3.7864 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0