roberta-base-academic
This model is a fine-tuned version of roberta-base on a combination of Elsevier OA CC-by dataset and other corpora of university essays such as BAWE and MICUSP. It achieves the following results on the evaluation set:
- Loss: 1.4229
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: 0.0001
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 16
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.99) and epsilon=1e-08
- 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 |
|---|---|---|---|
| 1.671 | 1.0 | 338 | 1.5581 |
| 1.6395 | 1.99 | 676 | 1.5276 |
| 1.5991 | 2.99 | 1014 | 1.5108 |
| 1.5659 | 3.99 | 1352 | 1.4903 |
| 1.5393 | 4.99 | 1690 | 1.4668 |
| 1.5178 | 5.98 | 2028 | 1.4621 |
| 1.4962 | 6.98 | 2366 | 1.4388 |
| 1.4783 | 7.98 | 2704 | 1.4320 |
| 1.4652 | 8.97 | 3042 | 1.4216 |
| 1.4542 | 9.97 | 3380 | 1.4180 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu116
- Datasets 2.8.0
- Tokenizers 0.13.2
- Downloads last month
- 12