deberta-v3-large-wic-with_rationale
This model is a fine-tuned version of microsoft/deberta-v3-large on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.5503
- Accuracy: 0.7618
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: 1e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.6832 | 1.0 | 170 | 0.5426 | 0.7210 |
| 0.4322 | 2.0 | 340 | 0.6440 | 0.7006 |
| 0.2921 | 3.0 | 510 | 0.6459 | 0.7602 |
| 0.1524 | 4.0 | 680 | 0.7986 | 0.7476 |
| 0.0785 | 5.0 | 850 | 1.0614 | 0.7649 |
| 0.0353 | 6.0 | 1020 | 1.3359 | 0.7524 |
| 0.0277 | 7.0 | 1190 | 1.3471 | 0.7665 |
| 0.025 | 8.0 | 1360 | 1.4481 | 0.7555 |
| 0.0183 | 9.0 | 1530 | 1.5351 | 0.7727 |
| 0.0091 | 10.0 | 1700 | 1.5503 | 0.7618 |
Framework versions
- Transformers 4.57.6
- Pytorch 2.8.0+cu128
- Datasets 4.5.0
- Tokenizers 0.22.2