CeLLaTe-tapt-bioformer16l-tokenizer-original-wwmask
This model is a fine-tuned version of bioformers/bioformer-16l on the Mardiyyah/TAPT_CeLLaTe2.0 dataset. It achieves the following results on the evaluation set:
- Loss: 1.3129
- Accuracy: 0.7322
- Perplexity: 3.7168
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 |
|---|---|---|---|---|---|
| 1.5638 | 1.0 | 15 | 1.3829 | 0.7211 | 3.9863 |
| 1.5566 | 2.0 | 30 | 1.3512 | 0.7298 | 3.8622 |
| 1.5428 | 3.0 | 45 | 1.3470 | 0.7249 | 3.8457 |
| 1.5242 | 4.0 | 60 | 1.3387 | 0.7330 | 3.8143 |
| 1.509 | 5.0 | 75 | 1.3191 | 0.7296 | 3.7400 |
| 1.4827 | 6.0 | 90 | 1.3234 | 0.7304 | 3.7563 |
| 1.4929 | 7.0 | 105 | 1.3199 | 0.7295 | 3.7431 |
| 1.4475 | 8.0 | 120 | 1.3436 | 0.7242 | 3.8329 |
| 1.4111 | 9.0 | 135 | 1.2958 | 0.7323 | 3.6540 |
| 1.426 | 10.0 | 150 | 1.3413 | 0.7235 | 3.8240 |
| 1.4173 | 11.0 | 165 | 1.3045 | 0.7290 | 3.6859 |
| 1.4146 | 12.0 | 180 | 1.3439 | 0.7259 | 3.8338 |
| 1.3824 | 13.0 | 195 | 1.3025 | 0.7308 | 3.6785 |
| 1.3899 | 14.0 | 210 | 1.3377 | 0.7239 | 3.8102 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0