<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
bert-finetuned-mutation-recognition-1
This model is a fine-tuned version of bert-base-cased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0380
- Proteinmutation F1: 0.8631
- Dnamutation F1: 0.7522
- Snp F1: 1.0
- Precision: 0.8061
- Recall: 0.8386
- F1: 0.8221
- Accuracy: 0.9942
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 | Proteinmutation F1 | Dnamutation F1 | Snp F1 | Precision | Recall | F1 | Accuracy |
---|---|---|---|---|---|---|---|---|---|---|
No log | 1.0 | 259 | 0.0273 | 0.8072 | 0.5762 | 0.975 | 0.6685 | 0.7580 | 0.7104 | 0.9924 |
0.0597 | 2.0 | 518 | 0.0260 | 0.8148 | 0.6864 | 0.9873 | 0.7363 | 0.8004 | 0.7670 | 0.9936 |
0.0597 | 3.0 | 777 | 0.0338 | 0.8252 | 0.7221 | 1.0 | 0.7857 | 0.7941 | 0.7899 | 0.9935 |
0.0046 | 4.0 | 1036 | 0.0299 | 0.8707 | 0.7214 | 0.9873 | 0.7773 | 0.8450 | 0.8098 | 0.9941 |
0.0046 | 5.0 | 1295 | 0.0353 | 0.9035 | 0.7364 | 0.9873 | 0.8130 | 0.8493 | 0.8307 | 0.9941 |
0.0014 | 6.0 | 1554 | 0.0361 | 0.8941 | 0.7391 | 0.9873 | 0.8093 | 0.8471 | 0.8278 | 0.9941 |
0.0014 | 7.0 | 1813 | 0.0367 | 0.8957 | 0.7249 | 1.0 | 0.8090 | 0.8365 | 0.8225 | 0.9940 |
0.0004 | 8.0 | 2072 | 0.0381 | 0.8714 | 0.7578 | 1.0 | 0.8266 | 0.8301 | 0.8284 | 0.9940 |
0.0004 | 9.0 | 2331 | 0.0380 | 0.8732 | 0.7550 | 1.0 | 0.8148 | 0.8408 | 0.8276 | 0.9942 |
0.0002 | 10.0 | 2590 | 0.0380 | 0.8631 | 0.7522 | 1.0 | 0.8061 | 0.8386 | 0.8221 | 0.9942 |
Framework versions
- Transformers 4.17.0
- Pytorch 1.10.2
- Datasets 2.0.0
- Tokenizers 0.12.1