<!-- 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. -->
protBERTbfd_AAV2_classification
This model is a fine-tuned version of Rostlab/prot_bert_bfd on AAV2 dataset with ~230k sequences (Bryant et al 2020).
The WT sequence (aa561-588): D E E E I R T T N P V A T E Q Y G S V S T N L Q R G N R Maximum length: 50
It achieves the following results on the evaluation set. Note:this is result of the last epoch, I think the pushed model is loaded with best checkpoint - best val_loss, I'm not so sure though :/
- Loss: 0.1341
- Accuracy: 0.9615
- F1: 0.9627
- Precision: 0.9637
- Recall: 0.9618
- Auroc: 0.9615
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: 5e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 64
- total_train_batch_size: 2048
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 200
- num_epochs: 8
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Precision | Recall | Auroc |
|---|---|---|---|---|---|---|---|---|
| No log | 1.0 | 116 | 0.2582 | 0.9064 | 0.9157 | 0.8564 | 0.9839 | 0.9038 |
| No log | 2.0 | 232 | 0.1447 | 0.9424 | 0.9432 | 0.9618 | 0.9252 | 0.9430 |
| No log | 3.0 | 348 | 0.1182 | 0.9542 | 0.9556 | 0.9573 | 0.9539 | 0.9542 |
| No log | 4.0 | 464 | 0.1129 | 0.9585 | 0.9602 | 0.9520 | 0.9685 | 0.9581 |
| 0.2162 | 5.0 | 580 | 0.1278 | 0.9553 | 0.9558 | 0.9776 | 0.9351 | 0.9561 |
| 0.2162 | 6.0 | 696 | 0.1139 | 0.9587 | 0.9607 | 0.9465 | 0.9752 | 0.9581 |
| 0.2162 | 7.0 | 812 | 0.1127 | 0.9620 | 0.9633 | 0.9614 | 0.9652 | 0.9619 |
| 0.2162 | 8.0 | 928 | 0.1341 | 0.9615 | 0.9627 | 0.9637 | 0.9618 | 0.9615 |
Framework versions
- Transformers 4.18.0
- Pytorch 1.11.0+cu113
- Datasets 2.1.0
- Tokenizers 0.12.1
