bugfix to padding-related issues

app.py

@@ -179,7 +179,8 @@ def chat_response(user_message, chat_history, system_prompt, enable_search,
                 top_k=top_k,
                 top_p=top_p,
                 repetition_penalty=repeat_penalty,
-                do_sample=True
+                do_sample=True,
+                pad_token_id=tokenizer.pad_token_id,
             )
         # Decode the generated tokens.
         generated_text = tokenizer.decode(output_ids[0], skip_special_tokens=True)