electramed-small-BC5CDR-ner
This model is a fine-tuned version of giacomomiolo/electramed_small_scivocab on the bc5_cdr dataset. It achieves the following results on the evaluation set:
- Loss: 0.1227
- Precision: 0.8092
- Recall: 0.8829
- F1: 0.8444
- Accuracy: 0.9686
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: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| 0.7177 | 1.0 | 286 | 0.6902 | 0.0 | 0.0 | 0.0 | 0.8864 |
| 0.1561 | 2.0 | 572 | 0.3210 | 0.7334 | 0.8104 | 0.7700 | 0.9636 |
| 0.2511 | 3.0 | 858 | 0.2064 | 0.7809 | 0.8711 | 0.8236 | 0.9666 |
| 0.0512 | 4.0 | 1144 | 0.1599 | 0.7937 | 0.8751 | 0.8324 | 0.9689 |
| 0.083 | 5.0 | 1430 | 0.1449 | 0.7983 | 0.8804 | 0.8373 | 0.9679 |
| 0.0412 | 6.0 | 1716 | 0.1315 | 0.8141 | 0.8825 | 0.8469 | 0.9701 |
| 0.1437 | 7.0 | 2002 | 0.1258 | 0.8227 | 0.8758 | 0.8485 | 0.9699 |
| 0.1894 | 8.0 | 2288 | 0.1226 | 0.8141 | 0.8833 | 0.8473 | 0.9696 |
| 0.0236 | 9.0 | 2574 | 0.1220 | 0.8160 | 0.8824 | 0.8479 | 0.9694 |
| 0.0602 | 10.0 | 2860 | 0.1227 | 0.8092 | 0.8829 | 0.8444 | 0.9686 |
Framework versions
- Transformers 4.21.1
- Pytorch 1.12.1+cu113
- Datasets 2.4.0
- Tokenizers 0.12.1