Update handler.py

handler.py

@@ -90,7 +90,16 @@ class EndpointHandler:
             return [{"error": "Missing 'clip' or 'prompt' in input data"}]
 
         # Prepare the inputs for the model
-        inputs_video = self.processor(text=prompt, videos=clip, padding=True, return_tensors="pt").to(self.model.device)
+        inputs_video = self.processor(
+            text=prompt, 
+            videos=clip, 
+            padding=True, 
+            return_tensors="pt"
+        ).to(self.model.device)
+        
+        # Check if 'pixel_values_videos' needs to be renamed to 'pixel_values'
+        if 'pixel_values_videos' in inputs_video:
+            inputs_video['pixel_values'] = inputs_video.pop('pixel_values_videos')
 
         # Generate output from the model
         generate_kwargs = {"max_new_tokens": 512, "do_sample": True, "top_p": 0.9}