A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 07:15 UTC
Support allowed_token_ids in ChatCompletionRequest (#19143)
Signed-off-by: Xu Song <xusong.vip@gmail.com>
Xu Song committed
Jun 6, 2025 at 13:06 UTC
3da2313d781f73c4b3b6bd57a130f85b7c0f0ca4
1 file changed
+2
vllm/entrypoints/openai/protocol.py
+2
@@ -271,6 +271,7 @@ class ChatCompletionRequest(OpenAIBaseModel):
271
spaces_between_special_tokens: bool = True
272
truncate_prompt_tokens: Optional[Annotated[int, Field(ge=1)]] = None
273
prompt_logprobs: Optional[int] = None
274
+ allowed_token_ids: Optional[list[int]] = None
275
# --8<-- [end:chat-completion-sampling-params]
276
277
# --8<-- [start:chat-completion-extra-params]
@@ -549,6 +550,7 @@ class ChatCompletionRequest(OpenAIBaseModel):
550
else RequestOutputKind.FINAL_ONLY,
551
guided_decoding=guided_decoding,
552
logit_bias=self.logit_bias,
553
+ allowed_token_ids=self.allowed_token_ids,
554
extra_args=({"kv_transfer_params": self.kv_transfer_params}
555
if self.kv_transfer_params else None))
556