bert-base-uncased-finetuned-academic
This model is a fine-tuned version of bert-base-uncased on the elsevier-oa-cc-by dataset. It achieves the following results on the evaluation set:
- Loss: 2.5893
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: 40
- eval_batch_size: 40
- seed: 42
- optimizer: Adam with betas=(0.9,0.97) and epsilon=0.0001
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 2.9591 | 0.25 | 820 | 2.6567 |
| 2.7993 | 0.5 | 1640 | 2.6006 |
| 2.7519 | 0.75 | 2460 | 2.5707 |
| 2.7319 | 1.0 | 3280 | 2.5763 |
| 2.7359 | 1.25 | 4100 | 2.5866 |
| 2.7451 | 1.5 | 4920 | 2.5855 |
| 2.7421 | 1.75 | 5740 | 2.5770 |
| 2.7319 | 2.0 | 6560 | 2.5762 |
| 2.7356 | 2.25 | 7380 | 2.5807 |
| 2.7376 | 2.5 | 8200 | 2.5813 |
| 2.7386 | 2.75 | 9020 | 2.5841 |
| 2.7378 | 3.0 | 9840 | 2.5737 |
Framework versions
- Transformers 4.19.0.dev0
- Pytorch 1.11.0+cu113
- Datasets 2.3.2
- Tokenizers 0.12.1