phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-08-02 00:34:44 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix: model to model_id

Browse source
This commit is contained in:
seyeong-han 2024-11-24 01:31:47 -06:00
parent 09490e4ee3
commit 73b51308d3
1 changed files with 4 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

8
llama_stack/providers/remote/inference/ssambanova/ssambanova.py
View file

@ -74,7 +74,7 @@ class SsambanovaInferenceAdapter(ModelRegistryHelper, Inference):
async def completion( async def completion(
self, self,
model: str, model_id: str,
content: InterleavedTextMedia, content: InterleavedTextMedia,
sampling_params: Optional[SamplingParams] = SamplingParams(), sampling_params: Optional[SamplingParams] = SamplingParams(),
response_format: Optional[ResponseFormat] = None, response_format: Optional[ResponseFormat] = None,
@ -85,7 +85,7 @@ class SsambanovaInferenceAdapter(ModelRegistryHelper, Inference):
async def chat_completion( async def chat_completion(
self, self,
model: str, model_id: str,
messages: List[Message], messages: List[Message],
sampling_params: Optional[SamplingParams] = SamplingParams(), sampling_params: Optional[SamplingParams] = SamplingParams(),
response_format: Optional[ResponseFormat] = None, response_format: Optional[ResponseFormat] = None,
@ -96,7 +96,7 @@ class SsambanovaInferenceAdapter(ModelRegistryHelper, Inference):
logprobs: Optional[LogProbConfig] = None, logprobs: Optional[LogProbConfig] = None,
) -> AsyncGenerator: ) -> AsyncGenerator:
request = ChatCompletionRequest( request = ChatCompletionRequest(
model=model, model=model_id,
messages=messages, messages=messages,
sampling_params=sampling_params, sampling_params=sampling_params,
tools=tools or [], tools=tools or [],
@ -147,7 +147,7 @@ class SsambanovaInferenceAdapter(ModelRegistryHelper, Inference):
async def embeddings( async def embeddings(
self, self,
model: str, model_id: str,
contents: List[InterleavedTextMedia], contents: List[InterleavedTextMedia],
) -> EmbeddingsResponse: ) -> EmbeddingsResponse:
raise NotImplementedError() raise NotImplementedError()