BERT4code
Collection
This collection features BERT and RoBERTa-based models fine-tuned for multi-label code classification, designed to accurately tag and categorize code
•
3 items
•
Updated
codebert-base
This model is a fine-tuned version of microsoft/codebert-base on the xCodeEval dataset, more precisely on the multi-tag classification task.. It achieves the following results on the evaluation set:
- Loss: 0.2756
- F1 Macro: 0.3907
- F1 Micro: 0.6159
- Roc Auc: 0.8944
- Accuracy: 0.4118
- Hamming Loss: 0.1088
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss | F1 Macro | F1 Micro | Roc Auc | Accuracy | Hamming Loss |
|---|---|---|---|---|---|---|---|---|
| No log | 1.0 | 287 | 0.3148 | 0.1532 | 0.4274 | 0.8593 | 0.2510 | 0.1314 |
| 0.3402 | 2.0 | 574 | 0.2830 | 0.3484 | 0.5897 | 0.8873 | 0.3765 | 0.1132 |
| 0.3402 | 3.0 | 861 | 0.2756 | 0.3907 | 0.6159 | 0.8944 | 0.4118 | 0.1088 |
Framework versions
- Transformers 4.44.2
- Pytorch 2.2.2
- Datasets 3.0.0
- Tokenizers 0.19.1
- Downloads last month
- 12
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support
Model tree for Mallard74/codebert-xcode-tags-classification
Base model
microsoft/codebert-base