CeLLaTe-tapt-bioformer16l-tokenizer-adapted-wwmask-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: 1.4792
- Accuracy: 0.7139
- Perplexity: 4.3895
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 |
|---|---|---|---|---|---|
| 1.9266 | 1.0 | 30 | 1.7206 | 0.6978 | 5.5881 |
| 1.8986 | 2.0 | 60 | 1.6633 | 0.7021 | 5.2767 |
| 1.8293 | 3.0 | 90 | 1.6614 | 0.7005 | 5.2666 |
| 1.8045 | 4.0 | 120 | 1.6091 | 0.7056 | 4.9982 |
| 1.7507 | 5.0 | 150 | 1.5792 | 0.7076 | 4.8511 |
| 1.7157 | 6.0 | 180 | 1.5732 | 0.7019 | 4.8221 |
| 1.7036 | 7.0 | 210 | 1.5474 | 0.7079 | 4.6994 |
| 1.647 | 8.0 | 240 | 1.5618 | 0.7075 | 4.7673 |
| 1.6048 | 9.0 | 270 | 1.5185 | 0.7092 | 4.5652 |
| 1.6109 | 10.0 | 300 | 1.5061 | 0.7127 | 4.5090 |
| 1.592 | 11.0 | 330 | 1.5194 | 0.7112 | 4.5696 |
| 1.5659 | 12.0 | 360 | 1.4839 | 0.7152 | 4.4103 |
| 1.552 | 13.0 | 390 | 1.4644 | 0.7169 | 4.3250 |
| 1.537 | 14.0 | 420 | 1.4882 | 0.7140 | 4.4291 |
| 1.5244 | 15.0 | 450 | 1.4913 | 0.7147 | 4.4430 |
| 1.4891 | 16.0 | 480 | 1.4797 | 0.7124 | 4.3915 |
| 1.4937 | 17.0 | 510 | 1.4396 | 0.7209 | 4.2189 |
| 1.4898 | 18.0 | 540 | 1.4614 | 0.7203 | 4.3120 |
| 1.4762 | 19.0 | 570 | 1.4471 | 0.7183 | 4.2507 |
| 1.4581 | 20.0 | 600 | 1.4833 | 0.7111 | 4.4076 |
| 1.4349 | 21.0 | 630 | 1.4684 | 0.7156 | 4.3423 |
| 1.4304 | 22.0 | 660 | 1.4600 | 0.7167 | 4.3062 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0