Update app.py

app.py

@@ -1,8 +1,24 @@
 import string
 import gradio as gr
 import requests
+import torch
+
+
+from transformers import BlipForQuestionAnswering, BlipProcessor
+
+device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+
+processor = BlipProcessor.from_pretrained("Salesforce/blip-vqa-base")
+model_vqa = BlipForQuestionAnswering.from_pretrained("Salesforce/blip-vqa-base").to(device)
 def inference_chat(input_image,input_text):
-    return input_text
+    inputs = processor(images=input_image, text=input_text,return_tensors="pt")
+
+
+    inputs["max_length"] = 20
+    inputs["num_beams"] = 5
+
+    out = model_vqa.generate(**inputs)
+    return processor.batch_decode(out, skip_special_tokens=True)[0]
     
 with gr.Blocks(
     css="""
@@ -17,8 +33,6 @@ with gr.Blocks(
     #gr.Markdown(article)
 
     with gr.Row():
-        with gr.Column():
-            caption_output = gr.Textbox(lines=1, label="VQA Output")
         with gr.Column(scale=1):
             image_input = gr.Image(type="pil")
 
@@ -34,7 +48,7 @@ with gr.Blocks(
                         ],
                         [ caption_output],
                     )
-    
+                    caption_output = gr.Textbox(lines=1, label="VQA Output")
                     with gr.Row():
                         clear_button = gr.Button(value="Clear", interactive=True)
                         clear_button.click(