Instructions to use codesage/codesage-small with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use codesage/codesage-small with Transformers:
# Load model directly from transformers import CodeSage model = CodeSage.from_pretrained("codesage/codesage-small", trust_remote_code=True, dtype="auto") - Notebooks
- Google Colab
- Kaggle
Update modeling_codesage.py
Browse files- modeling_codesage.py +3 -1
modeling_codesage.py
CHANGED
|
@@ -351,7 +351,9 @@ class CodeSageForSequenceClassification(CodeSagePreTrainedModel):
|
|
| 351 |
|
| 352 |
self.transformer = CodeSageModel(config)
|
| 353 |
classifier_dropout = (
|
| 354 |
-
config.classifier_dropout
|
|
|
|
|
|
|
| 355 |
)
|
| 356 |
self.dropout = nn.Dropout(classifier_dropout)
|
| 357 |
self.classifier = nn.Linear(config.hidden_size, config.num_labels)
|
|
|
|
| 351 |
|
| 352 |
self.transformer = CodeSageModel(config)
|
| 353 |
classifier_dropout = (
|
| 354 |
+
config.classifier_dropout
|
| 355 |
+
if hasattr(config, 'classifier_dropout') and config.classifier_dropout is not None
|
| 356 |
+
else config.residual_dropout_prob
|
| 357 |
)
|
| 358 |
self.dropout = nn.Dropout(classifier_dropout)
|
| 359 |
self.classifier = nn.Linear(config.hidden_size, config.num_labels)
|