feat: support stop_token_ids parameter. (#1097)

vllm/entrypoints/openai/api_server.py

@@ -217,6 +217,7 @@ async def create_chat_completion(request: ChatCompletionRequest,
             temperature=request.temperature,
             top_p=request.top_p,
             stop=request.stop,
+            stop_token_ids=request.stop_token_ids,
             max_tokens=request.max_tokens,
             best_of=request.best_of,
             top_k=request.top_k,
@@ -418,6 +419,7 @@ async def create_completion(request: CompletionRequest, raw_request: Request):
             top_p=request.top_p,
             top_k=request.top_k,
             stop=request.stop,
+            stop_token_ids=request.stop_token_ids,
             ignore_eos=request.ignore_eos,
             max_tokens=request.max_tokens,
             logprobs=request.logprobs,