fixed Kwargs passed to processor.__call__ have to be in processor_kwargs dict

model_engine.py

@@ -132,13 +132,16 @@ def _generate_from_messages(
         add_generation_prompt=True,
         return_dict=True,
         return_tensors="pt",
-        downsample_mode=downsample_mode,
-        max_slice_nums=36,
+        processor_kwargs={
+            "downsample_mode": downsample_mode,
+            "max_slice_nums": 36,
+        },
     ).to(model.device)
     generated_ids = model.generate(
         **inputs,
         downsample_mode=downsample_mode,
         max_new_tokens=max_new_tokens,
+        pad_token_id=processor.tokenizer.eos_token_id,
     )
     generated_ids_trimmed = [
         output_ids[len(input_ids):]