Update app.py

app.py

@@ -102,10 +102,12 @@ def run_inference(image, prompt, max_new_tokens):
     )
 
     inputs = processor(
-        text=[text_prompt],
-        images=[image],
-        return_tensors="pt"
-    ).to(model.device)
+                            text=[text_prompt],
+                            images=[image],
+                            return_tensors="pt",
+                            truncation=False,        # 🔴 BẮT BUỘC
+                            padding="longest"        # 🔴 BẮT BUỘC
+                        ).to(model.device)
 
     with torch.inference_mode(), torch.cuda.amp.autocast(dtype=torch.float16):
         outputs = model.generate(