enabled blip large

app.py

@@ -11,8 +11,8 @@ torch.hub.download_url_to_file('http://images.cocodataset.org/val2017/0000000397
 torch.hub.download_url_to_file('https://huggingface.co/datasets/nielsr/textcaps-sample/resolve/main/stop_sign.png', 'stop_sign.png')
 torch.hub.download_url_to_file('https://cdn.openai.com/dall-e-2/demos/text2im/astronaut/horse/photo/0.jpg', 'astronaut.jpg')
 
-git_processor_base = AutoProcessor.from_pretrained("microsoft/git-base-coco")
-git_model_base = AutoModelForCausalLM.from_pretrained("microsoft/git-base-coco")
+# git_processor_base = AutoProcessor.from_pretrained("microsoft/git-base-coco")
+# git_model_base = AutoModelForCausalLM.from_pretrained("microsoft/git-base-coco")
 
 # git_processor_large_coco = AutoProcessor.from_pretrained("microsoft/git-large-coco")
 # git_model_large_coco = AutoModelForCausalLM.from_pretrained("microsoft/git-large-coco")
@@ -20,11 +20,11 @@ git_model_base = AutoModelForCausalLM.from_pretrained("microsoft/git-base-coco")
 # git_processor_large_textcaps = AutoProcessor.from_pretrained("microsoft/git-large-r-textcaps")
 # git_model_large_textcaps = AutoModelForCausalLM.from_pretrained("microsoft/git-large-r-textcaps")
 
-blip_processor_base = AutoProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
-blip_model_base = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")
+# blip_processor_base = AutoProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
+# blip_model_base = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")
 
-# blip_processor_large = AutoProcessor.from_pretrained("Salesforce/blip-image-captioning-large")
-# blip_model_large = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large")
+blip_processor_large = AutoProcessor.from_pretrained("Salesforce/blip-image-captioning-large")
+blip_model_large = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large")
 
 # blip2_processor = AutoProcessor.from_pretrained("Salesforce/blip2-opt-2.7b")
 # blip2_model = Blip2ForConditionalGeneration.from_pretrained("Salesforce/blip2-opt-2.7b", torch_dtype=torch.float16)
@@ -43,11 +43,11 @@ blip_model_base = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-
 
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
-git_model_base.to(device)
-blip_model_base.to(device)
+# git_model_base.to(device)
+# blip_model_base.to(device)
 # git_model_large_coco.to(device)
 # git_model_large_textcaps.to(device)
-# blip_model_large.to(device)
+blip_model_large.to(device)
 # vitgpt_model.to(device)
 # coca_model.to(device)
 # blip2_model.to(device)
@@ -76,15 +76,15 @@ def generate_caption_coca(model, transform, image):
 
 
 def generate_captions(image):
-    caption_git_base = generate_caption(git_processor_base, git_model_base, image)
+    # caption_git_base = generate_caption(git_processor_base, git_model_base, image)
 
     # caption_git_large_coco = generate_caption(git_processor_large_coco, git_model_large_coco, image)
 
     # caption_git_large_textcaps = generate_caption(git_processor_large_textcaps, git_model_large_textcaps, image)
 
-    caption_blip_base = generate_caption(blip_processor_base, blip_model_base, image)
+    # caption_blip_base = generate_caption(blip_processor_base, blip_model_base, image)
 
-    # caption_blip_large = generate_caption(blip_processor_large, blip_model_large, image)
+    caption_blip_large = generate_caption(blip_processor_large, blip_model_large, image)
 
     # caption_vitgpt = generate_caption(vitgpt_processor, vitgpt_model, image, vitgpt_tokenizer)
 
@@ -95,22 +95,22 @@ def generate_captions(image):
     # caption_blip2_8_bit = generate_caption(blip2_processor_8_bit, blip2_model_8_bit, image, use_float_16=True).strip()
 
     # return caption_git_large_coco, caption_git_large_textcaps, caption_blip_large, caption_coca, caption_blip2_8_bit
-    return caption_git_base, caption_blip_base
+    return caption_blip_large
 
 
    
 examples = [["cats.jpg"], ["stop_sign.png"], ["astronaut.jpg"]]
 # outputs = [gr.outputs.Textbox(label="Caption generated by GIT-large fine-tuned on COCO"), gr.outputs.Textbox(label="Caption generated by GIT-large fine-tuned on TextCaps"), gr.outputs.Textbox(label="Caption generated by BLIP-large"), gr.outputs.Textbox(label="Caption generated by CoCa"), gr.outputs.Textbox(label="Caption generated by BLIP-2 OPT 6.7b")] 
 outputs = [
-    gr.outputs.Textbox(label="Caption generated by GIT-base fine-tuned on COCO"), 
+    # gr.outputs.Textbox(label="Caption generated by GIT-base fine-tuned on COCO"), 
            # gr.outputs.Textbox(label="Caption generated by GIT-large fine-tuned on COCO"),
            # gr.outputs.Textbox(label="Caption generated by GIT-large fine-tuned on TextCaps"),
-           gr.outputs.Textbox(label="Caption generated by BLIP-base"),
-           # gr.outputs.Textbox(label="Caption generated by BLIP-large"),
+           # gr.outputs.Textbox(label="Caption generated by BLIP-base"),
+           gr.outputs.Textbox(label="Caption generated by BLIP-large"),
            # gr.outputs.Textbox(label="Caption generated by vitgpt")
           ] 
 
-title = "Interactive demo: comparing image captioning models"
+title = "Interactive demo: blip-large"
 description = "Gradio Demo to compare GIT, BLIP, CoCa, and BLIP-2, 4 state-of-the-art vision+language models. To use it, simply upload your image and click 'submit', or click one of the examples to load them. Read more at the links below."
 article = "<p style='text-align: center'><a href='https://huggingface.co/docs/transformers/main/model_doc/blip' target='_blank'>BLIP docs</a> | <a href='https://huggingface.co/docs/transformers/main/model_doc/git' target='_blank'>GIT docs</a></p>"