deberta_mcq
This model is a fine-tuned version of microsoft/deberta-v3-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.0868
- Accuracy: 0.9962
- F1 Macro: 0.9960
- Map At 3: 0.9981
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: 16
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 6
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 Macro | Map At 3 |
|---|---|---|---|---|---|---|
| 1.6085 | 0.9840 | 46 | 1.5929 | 0.6061 | 0.5985 | 0.7506 |
| 1.3478 | 1.9893 | 93 | 0.7228 | 0.9242 | 0.9211 | 0.9558 |
| 0.4559 | 2.9947 | 140 | 0.2588 | 0.9848 | 0.9839 | 0.9924 |
| 0.3132 | 4.0 | 187 | 0.1472 | 0.9924 | 0.9920 | 0.9962 |
| 0.2326 | 4.9840 | 233 | 0.0957 | 0.9962 | 0.9960 | 0.9981 |
| 0.201 | 5.9037 | 276 | 0.0868 | 0.9962 | 0.9960 | 0.9981 |
Framework versions
- Transformers 4.44.2
- Pytorch 2.13.0+cu130
- Datasets 4.8.5
- Tokenizers 0.19.1