phoenix-oss/llama-stack
1
0
Fork 0
forked from phoenix-oss/llama-stack-mirror
Code Pull requests Releases Packages 2 Activity Actions

Fix conversion to RawMessage everywhere

Browse source
This commit is contained in:
Ashwin Bharambe 2024-12-17 13:38:01 -08:00
parent fbca51d6da
commit b7a7caa9a8
11 changed files with 87 additions and 78 deletions
Download patch file Download diff file Expand all files Collapse all files

6
llama_stack/providers/remote/inference/vllm/vllm.py
View file

@ -77,7 +77,7 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
stream: Optional[bool] = False,
logprobs: Optional[LogProbConfig] = None,
) -> Union[CompletionResponse, CompletionResponseStreamChunk]:
raise NotImplementedError()
raise NotImplementedError("Completion not implemented for vLLM")
async def chat_completion(
self,
@ -167,7 +167,7 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
for m in request.messages
]
else:
input_dict["prompt"] = chat_completion_request_to_prompt(
input_dict["prompt"] = await chat_completion_request_to_prompt(
request,
self.register_helper.get_llama_model(request.model),
self.formatter,
@ -176,7 +176,7 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
assert (
not media_present
), "Together does not support media for Completion requests"
input_dict["prompt"] = completion_request_to_prompt(
input_dict["prompt"] = await completion_request_to_prompt(
request,
self.register_helper.get_llama_model(request.model),
self.formatter,