try chonk

app.py

@@ -18,10 +18,10 @@ blip_processor_large = AutoProcessor.from_pretrained("Salesforce/blip-image-capt
 blip_model_large = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large").to(device)
 
 blip2_processor = AutoProcessor.from_pretrained("Salesforce/blip2-opt-6.7b")
-blip2_model_4_bit = Blip2ForConditionalGeneration.from_pretrained("Salesforce/blip2-opt-6.7b", device_map="auto", load_in_4bit=True, torch_dtype=torch.float16)
+blip2_model = Blip2ForConditionalGeneration.from_pretrained("Salesforce/blip2-opt-6.7b", device_map="auto", torch_dtype=torch.float16)
 
 instructblip_processor = AutoProcessor.from_pretrained("Salesforce/instructblip-vicuna-7b")
-instructblip_model_4_bit = InstructBlipForConditionalGeneration.from_pretrained("Salesforce/instructblip-vicuna-7b", device_map="auto", load_in_4bit=True, torch_dtype=torch.float16)
+instructblip_model = InstructBlipForConditionalGeneration.from_pretrained("Salesforce/instructblip-vicuna-7b", device_map="auto", torch_dtype=torch.float16)
 
 def generate_caption(processor, model, image, tokenizer=None, use_float_16=False):
     inputs = processor(images=image, return_tensors="pt").to(device)
@@ -58,11 +58,11 @@ def generate_captions(image):
 
     caption_blip_large = generate_caption(blip_processor_large, blip_model_large, image)
 
-    caption_blip2_4_bit = generate_caption_blip2(blip2_processor, blip2_model_4_bit, image).strip()
+    caption_blip2 = generate_caption_blip2(blip2_processor, blip2_model, image).strip()
 
-    caption_instructblip_4_bit = generate_caption_blip2(instructblip_processor, instructblip_model_4_bit, image)
+    caption_instructblip = generate_caption_blip2(instructblip_processor, instructblip_model, image)
 
-    return caption_git_large_coco, caption_blip_large, caption_blip2_4_bit, caption_instructblip_4_bit
+    return caption_git_large_coco, caption_blip_large, caption_blip2, caption_instructblip
 
    
 examples = [["cats.jpg"], ["stop_sign.png"], ["astronaut.jpg"]]