Spaces:
Runtime error
Runtime error
acecalisto3
commited on
Update app.py
Browse files
app.py
CHANGED
@@ -303,44 +303,34 @@ class MultimodalRAG:
|
|
303 |
except Exception as e:
|
304 |
logger.error(f"Error loading template embeddings: {e}")
|
305 |
|
306 |
-
def
|
307 |
-
"""
|
308 |
try:
|
309 |
-
|
310 |
-
|
311 |
-
|
312 |
-
|
313 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
314 |
except Exception as e:
|
315 |
-
logger.error(f"Error
|
|
|
316 |
|
317 |
-
def encode_image(self, image: Image.Image) -> np.ndarray:
|
318 |
-
"""Encode image using BLIP"""
|
319 |
-
try:
|
320 |
-
model, processor = self.model_manager.load_model("image_processor")
|
321 |
-
|
322 |
-
# Process image
|
323 |
-
inputs = processor(images=image, return_tensors="pt").to(model.device)
|
324 |
-
|
325 |
-
# Get image features using the proper method
|
326 |
-
with torch.no_grad():
|
327 |
-
outputs = model.get_image_features(**inputs)
|
328 |
-
image_features = outputs.last_hidden_state.mean(dim=1) # Average pooling
|
329 |
-
|
330 |
-
return image_features.cpu().numpy()
|
331 |
-
|
332 |
-
except Exception as e:
|
333 |
-
logger.error(f"Error encoding image: {str(e)}")
|
334 |
-
raise ModelError(f"Error encoding image: {str(e)}")
|
335 |
-
|
336 |
def encode_text(self, text: str) -> np.ndarray:
|
337 |
"""Encode text using sentence-transformers"""
|
338 |
-
|
339 |
try:
|
340 |
return self.text_encoder.encode(text)
|
341 |
except Exception as e:
|
342 |
raise ModelError(f"Error encoding text: {str(e)}")
|
343 |
|
|
|
|
|
344 |
def generate_code(self, description: str, template_code: str) -> str:
|
345 |
"""Generate code using StarCoder"""
|
346 |
try:
|
|
|
303 |
except Exception as e:
|
304 |
logger.error(f"Error loading template embeddings: {e}")
|
305 |
|
306 |
+
def encode_image(self, image: Image.Image) -> np.ndarray:
|
307 |
+
"""Encode image using BLIP"""
|
308 |
try:
|
309 |
+
model, processor = self.model_manager.load_model("image_processor")
|
310 |
+
|
311 |
+
# Process image
|
312 |
+
inputs = processor(images=image, return_tensors="pt").to(model.device)
|
313 |
+
|
314 |
+
# Get image features using the proper method
|
315 |
+
with torch.no_grad():
|
316 |
+
outputs = model.get_image_features(**inputs)
|
317 |
+
image_features = outputs.last_hidden_state.mean(dim=1) # Average pooling
|
318 |
+
|
319 |
+
return image_features.cpu().numpy()
|
320 |
+
|
321 |
except Exception as e:
|
322 |
+
logger.error(f"Error encoding image: {str(e)}")
|
323 |
+
raise ModelError(f"Error encoding image: {str(e)}")
|
324 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
325 |
def encode_text(self, text: str) -> np.ndarray:
|
326 |
"""Encode text using sentence-transformers"""
|
|
|
327 |
try:
|
328 |
return self.text_encoder.encode(text)
|
329 |
except Exception as e:
|
330 |
raise ModelError(f"Error encoding text: {str(e)}")
|
331 |
|
332 |
+
# ... rest of the MultimodalRAG class methods ...
|
333 |
+
|
334 |
def generate_code(self, description: str, template_code: str) -> str:
|
335 |
"""Generate code using StarCoder"""
|
336 |
try:
|