[Frontend] New allowed_token_ids decoding request parameter (#6753)
This commit is contained in:
@@ -134,7 +134,7 @@ class OpenAIServingChat(OpenAIServing):
|
||||
|
||||
request_id = f"chat-{random_uuid()}"
|
||||
try:
|
||||
sampling_params = request.to_sampling_params()
|
||||
sampling_params = request.to_sampling_params(tokenizer)
|
||||
decoding_config = await self.engine.get_decoding_config()
|
||||
guided_decoding_backend = request.guided_decoding_backend \
|
||||
or decoding_config.guided_decoding_backend
|
||||
|
||||
Reference in New Issue
Block a user