Update app.py

app.py

@@ -20,15 +20,15 @@ model_vqa = BlipForQuestionAnswering.from_pretrained("Salesforce/blip-vqa-capfil
 
 from transformers import BlipProcessor, BlipForConditionalGeneration
 
-cap_processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
-cap_model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")
+cap_processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-large")
+cap_model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large")
 
 
 
 def caption(input_image):
     inputs = cap_processor(input_image, return_tensors="pt")
-    inputs["num_beams"] = 1
-    inputs['num_return_sequences'] =1
+    # inputs["num_beams"] = 1
+    # inputs['num_return_sequences'] =1
     out = cap_model.generate(**inputs)
     return "\n".join(cap_processor.batch_decode(out, skip_special_tokens=True))
 import openai
@@ -111,6 +111,7 @@ def vle(input_image,input_text):
     return [vqa['answer'] for vqa in vqa_answers],[vqa['score'] for vqa in vqa_answers]
 def inference_chat(input_image,input_text):
     cap=caption(input_image)
+    print(cap)
     # inputs = processor(images=input_image, text=input_text,return_tensors="pt")
     # inputs["max_length"] = 10
     # inputs["num_beams"] = 5