CeLLaTe-tapt-bioformer16l-tokenizer-original-spanmask
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: 2.6539
- Accuracy: 0.5310
- Perplexity: 14.2095
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 |
|---|---|---|---|---|---|
| 3.2148 | 1.0 | 15 | 2.8562 | 0.5216 | 17.3957 |
| 2.9835 | 2.0 | 30 | 2.6301 | 0.5422 | 13.8757 |
| 2.856 | 3.0 | 45 | 2.6085 | 0.5389 | 13.5783 |
| 2.8112 | 4.0 | 60 | 2.5362 | 0.5460 | 12.6320 |
| 2.8094 | 5.0 | 75 | 2.6204 | 0.5335 | 13.7416 |
| 2.7926 | 6.0 | 90 | 2.6063 | 0.5361 | 13.5493 |
| 2.7496 | 7.0 | 105 | 2.5235 | 0.5495 | 12.4725 |
| 2.6944 | 8.0 | 120 | 2.5665 | 0.5433 | 13.0207 |
| 2.7186 | 9.0 | 135 | 2.5562 | 0.5440 | 12.8863 |
| 2.685 | 10.0 | 150 | 2.5532 | 0.5415 | 12.8486 |
| 2.6691 | 11.0 | 165 | 2.5285 | 0.5454 | 12.5349 |
| 2.6679 | 12.0 | 180 | 2.6046 | 0.5385 | 13.5254 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0