Instructions to use jinaai/jina-clip-implementation with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use jinaai/jina-clip-implementation with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("jinaai/jina-clip-implementation", dtype="auto") - Notebooks
- Google Colab
- Kaggle
refactor: optimize training and tokenizer
Browse files- modeling_clip.py +0 -1
modeling_clip.py
CHANGED
|
@@ -247,7 +247,6 @@ class JinaCLIPModel(JinaCLIPPreTrainedModel):
|
|
| 247 |
)
|
| 248 |
return self.visual_projection(self.vision_model(x=x))
|
| 249 |
|
| 250 |
-
@property
|
| 251 |
def get_tokenizer(self):
|
| 252 |
if not self.tokenizer:
|
| 253 |
self.tokenizer = AutoTokenizer.from_pretrained(config._name_or_path)
|
|
|
|
| 247 |
)
|
| 248 |
return self.visual_projection(self.vision_model(x=x))
|
| 249 |
|
|
|
|
| 250 |
def get_tokenizer(self):
|
| 251 |
if not self.tokenizer:
|
| 252 |
self.tokenizer = AutoTokenizer.from_pretrained(config._name_or_path)
|