CeLLaTe-tapt-pubmedbert-tokenizer-original-spanmask-combinedData
This model is a fine-tuned version of microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext on the Mardiyyah/Combined_TAPT_dataset dataset. It achieves the following results on the evaluation set:
- Loss: 2.1023
- Accuracy: 0.5977
- Perplexity: 8.1850
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 |
|---|---|---|---|---|---|
| 2.6735 | 1.0 | 28 | 2.2438 | 0.5810 | 9.4294 |
| 2.4083 | 2.0 | 56 | 2.1526 | 0.5907 | 8.6071 |
| 2.375 | 3.0 | 84 | 2.1275 | 0.5909 | 8.3943 |
| 2.3415 | 4.0 | 112 | 2.1319 | 0.5954 | 8.4312 |
| 2.3423 | 5.0 | 140 | 2.1031 | 0.5948 | 8.1917 |
| 2.3198 | 6.0 | 168 | 2.0806 | 0.5985 | 8.0092 |
| 2.31 | 7.0 | 196 | 2.0744 | 0.6028 | 7.9594 |
| 2.3199 | 8.0 | 224 | 2.1230 | 0.5918 | 8.3558 |
| 2.2729 | 9.0 | 252 | 2.0403 | 0.6020 | 7.6928 |
| 2.2842 | 10.0 | 280 | 2.1096 | 0.5930 | 8.2446 |
| 2.2639 | 11.0 | 308 | 2.0864 | 0.5989 | 8.0560 |
| 2.2578 | 12.0 | 336 | 2.0610 | 0.6004 | 7.8541 |
| 2.2211 | 13.0 | 364 | 2.1497 | 0.5861 | 8.5819 |
| 2.2328 | 14.0 | 392 | 2.0952 | 0.5916 | 8.1274 |
Framework versions
- Transformers 4.48.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.21.0