[Bugfix] remove duplicate tokens streamed in required tool choice streaming (vllm-project#23312)

Signed-off-by: Jason Cheng <[email protected]>
Co-authored-by: Chauncey <[email protected]>

Author: 2 people

vllm/entrypoints/openai/serving_chat.py

@@ -828,9 +828,6 @@ async def chat_completion_stream_generator(
                             history_tool_call_cnt += 1
                             tools_streamed[i] = True
 
-                        # update the previous values for the next iteration
-                        previous_texts[i] = current_text
-
                     # handle streaming deltas for tools with "auto" tool choice
                     # and reasoning parser
                     elif tool_choice_auto and self.reasoning_parser: