phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-12-17 08:52:39 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

[#391] Add support for json structured output for vLLM

Browse source
This commit is contained in:
Aidan Do 2024-11-26 09:40:17 +00:00
parent 4e6c984c26
commit 1801aa145d
2 changed files with 13 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

11
llama_stack/providers/remote/inference/vllm/vllm.py
View file

@ -100,6 +100,7 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
tool_prompt_format=tool_prompt_format,
stream=stream,
logprobs=logprobs,
response_format=response_format,
)
if stream:
return self._stream_chat_completion(request, self.client)
@ -180,6 +181,16 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
self.formatter,
)
if fmt := request.response_format:
if fmt.type == ResponseFormatType.json_schema.value:
input_dict["extra_body"] = {
"guided_json": request.response_format.json_schema
}
elif fmt.type == ResponseFormatType.grammar.value:
raise NotImplementedError("Grammar response format not supported yet")
else:
raise ValueError(f"Unknown response format {fmt.type}")
return {
"model": request.model,
**input_dict,