Update app.py

app.py

@@ -284,7 +284,7 @@ def captioning(img):
             **inputs,
             do_sample=False,
             num_beams=5,
-            max_length=256,
+            max_length=64,
             #min_length=42,
             top_p=0.9,
             repetition_penalty=1.5,
@@ -294,18 +294,16 @@ def captioning(img):
         generated_text = processor5.batch_decode(generated_ids, skip_special_tokens=True)[0].strip()
         response_text = generated_text.replace(prompt, "").strip() #Or could try .split(prompt, 1)[-1].strip()
         output_prompt.append(response_text)
-        print(f"{response_text}\n") # Print only the response text
+        print(f"{response_text}\n")
     inputf = processor5(
-       # images=img,
+        images=img,
         text=generated_text + 'So therefore',
         return_tensors="pt"
     ).to('cuda')
-    generated_ids = model5.generate(**inputf, max_length=256)
+    generated_ids = model5.generate(**inputf, max_length=128)
     generated_texta = processor5.batch_decode(generated_ids, skip_special_tokens=True)[0].strip()
     response_text = generated_texta.replace(generated_text, "").strip()
-    #print(response_text)
     output_prompt.append(response_text)
-    #print(output_prompt)
     output_prompt = " ".join(output_prompt)
     return output_prompt