phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-07-11 23:56:11 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

OpenAI compat embeddings API

Browse source
This commit is contained in:
Hardik Shah 2025-05-29 15:27:59 -07:00
parent 2603f10f95
commit f2c2a05f58
20 changed files with 706 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

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

@ -22,6 +22,7 @@ from llama_stack.apis.inference import (
Inference,
LogProbConfig,
Message,
OpenAIEmbeddingsResponse,
ResponseFormat,
SamplingParams,
TextTruncation,
@ -197,3 +198,13 @@ class BedrockInferenceAdapter(
response_body = json.loads(response.get("body").read())
embeddings.append(response_body.get("embedding"))
return EmbeddingsResponse(embeddings=embeddings)
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
raise NotImplementedError()

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

@ -21,6 +21,7 @@ from llama_stack.apis.inference import (
Inference,
LogProbConfig,
Message,
OpenAIEmbeddingsResponse,
ResponseFormat,
SamplingParams,
TextTruncation,
@ -194,3 +195,13 @@ class CerebrasInferenceAdapter(
task_type: EmbeddingTaskType | None = None,
) -> EmbeddingsResponse:
raise NotImplementedError()
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
raise NotImplementedError()

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

@ -20,6 +20,7 @@ from llama_stack.apis.inference import (
Inference,
LogProbConfig,
Message,
OpenAIEmbeddingsResponse,
ResponseFormat,
SamplingParams,
TextTruncation,
@ -152,3 +153,13 @@ class DatabricksInferenceAdapter(
task_type: EmbeddingTaskType | None = None,
) -> EmbeddingsResponse:
raise NotImplementedError()
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
raise NotImplementedError()

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

@ -37,6 +37,7 @@ from llama_stack.apis.inference.inference import (
OpenAIChatCompletion,
OpenAIChatCompletionChunk,
OpenAICompletion,
OpenAIEmbeddingsResponse,
OpenAIMessageParam,
OpenAIResponseFormatParam,
)
@ -286,6 +287,16 @@ class FireworksInferenceAdapter(ModelRegistryHelper, Inference, NeedsRequestProv
embeddings = [data.embedding for data in response.data]
return EmbeddingsResponse(embeddings=embeddings)
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
raise NotImplementedError()
async def openai_completion(
self,
model: str,

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

@ -29,6 +29,7 @@ from llama_stack.apis.inference import (
Inference,
LogProbConfig,
Message,
OpenAIEmbeddingsResponse,
ResponseFormat,
SamplingParams,
TextTruncation,
@ -238,6 +239,16 @@ class NVIDIAInferenceAdapter(Inference, ModelRegistryHelper):
#
return EmbeddingsResponse(embeddings=[embedding.embedding for embedding in response.data])
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
raise NotImplementedError()
async def chat_completion(
self,
model_id: str,

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

@ -32,6 +32,7 @@ from llama_stack.apis.inference import (
JsonSchemaResponseFormat,
LogProbConfig,
Message,
OpenAIEmbeddingsResponse,
ResponseFormat,
SamplingParams,
TextTruncation,
@ -370,6 +371,16 @@ class OllamaInferenceAdapter(
return model
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
raise NotImplementedError()
async def openai_completion(
self,
model: str,

52
llama_stack/providers/remote/inference/openai/openai.py
View file

@ -14,6 +14,9 @@ from llama_stack.apis.inference.inference import (
OpenAIChatCompletion,
OpenAIChatCompletionChunk,
OpenAICompletion,
OpenAIEmbeddingData,
OpenAIEmbeddingsResponse,
OpenAIEmbeddingUsage,
OpenAIMessageParam,
OpenAIResponseFormatParam,
)
@ -38,6 +41,7 @@ logger = logging.getLogger(__name__)
# | batch_chat_completion | LiteLLMOpenAIMixin |
# | openai_completion | AsyncOpenAI |
# | openai_chat_completion | AsyncOpenAI |
# | openai_embeddings | AsyncOpenAI |
#
class OpenAIInferenceAdapter(LiteLLMOpenAIMixin):
def __init__(self, config: OpenAIConfig) -> None:
@ -171,3 +175,51 @@ class OpenAIInferenceAdapter(LiteLLMOpenAIMixin):
user=user,
)
return await self._openai_client.chat.completions.create(**params)
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
model_id = (await self.model_store.get_model(model)).provider_resource_id
if model_id.startswith("openai/"):
model_id = model_id[len("openai/") :]
# Prepare parameters for OpenAI embeddings API
params = {
"model": model_id,
"input": input,
}
if encoding_format is not None:
params["encoding_format"] = encoding_format
if dimensions is not None:
params["dimensions"] = dimensions
if user is not None:
params["user"] = user
# Call OpenAI embeddings API
response = await self._openai_client.embeddings.create(**params)
data = []
for i, embedding_data in enumerate(response.data):
data.append(
OpenAIEmbeddingData(
embedding=embedding_data.embedding,
index=i,
)
)
usage = OpenAIEmbeddingUsage(
prompt_tokens=response.usage.prompt_tokens,
total_tokens=response.usage.total_tokens,
)
return OpenAIEmbeddingsResponse(
data=data,
model=response.model,
usage=usage,
)

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

@ -19,6 +19,7 @@ from llama_stack.apis.inference import (
Inference,
LogProbConfig,
Message,
OpenAIEmbeddingsResponse,
ResponseFormat,
SamplingParams,
TextTruncation,
@ -210,6 +211,16 @@ class PassthroughInferenceAdapter(Inference):
task_type=task_type,
)
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
raise NotImplementedError()
async def openai_completion(
self,
model: str,

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

@ -8,6 +8,7 @@ from collections.abc import AsyncGenerator
from openai import OpenAI
from llama_stack.apis.inference import * # noqa: F403
from llama_stack.apis.inference.inference import OpenAIEmbeddingsResponse
# from llama_stack.providers.datatypes import ModelsProtocolPrivate
from llama_stack.providers.utils.inference.model_registry import ModelRegistryHelper
@ -134,3 +135,13 @@ class RunpodInferenceAdapter(
task_type: Optional[EmbeddingTaskType] = None,
) -> EmbeddingsResponse:
raise NotImplementedError()
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
raise NotImplementedError()

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

@ -23,6 +23,7 @@ from llama_stack.apis.inference import (
Inference,
LogProbConfig,
Message,
OpenAIEmbeddingsResponse,
ResponseFormat,
ResponseFormatType,
SamplingParams,
@ -291,6 +292,16 @@ class _HfAdapter(
) -> EmbeddingsResponse:
raise NotImplementedError()
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
raise NotImplementedError()
class TGIAdapter(_HfAdapter):
async def initialize(self, config: TGIImplConfig) -> None:

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

@ -23,6 +23,7 @@ from llama_stack.apis.inference import (
Inference,
LogProbConfig,
Message,
OpenAIEmbeddingsResponse,
ResponseFormat,
ResponseFormatType,
SamplingParams,
@ -267,6 +268,16 @@ class TogetherInferenceAdapter(ModelRegistryHelper, Inference, NeedsRequestProvi
embeddings = [item.embedding for item in r.data]
return EmbeddingsResponse(embeddings=embeddings)
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
raise NotImplementedError()
async def openai_completion(
self,
model: str,

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

@ -38,6 +38,7 @@ from llama_stack.apis.inference import (
JsonSchemaResponseFormat,
LogProbConfig,
Message,
OpenAIEmbeddingsResponse,
ResponseFormat,
SamplingParams,
TextTruncation,
@ -507,6 +508,16 @@ class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
embeddings = [data.embedding for data in response.data]
return EmbeddingsResponse(embeddings=embeddings)
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
raise NotImplementedError()
async def openai_completion(
self,
model: str,

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

@ -21,6 +21,7 @@ from llama_stack.apis.inference import (
Inference,
LogProbConfig,
Message,
OpenAIEmbeddingsResponse,
ResponseFormat,
SamplingParams,
TextTruncation,
@ -260,6 +261,16 @@ class WatsonXInferenceAdapter(Inference, ModelRegistryHelper):
) -> EmbeddingsResponse:
raise NotImplementedError("embedding is not supported for watsonx")
async def openai_embeddings(
self,
model: str,
input: str | list[str],
encoding_format: str | None = "float",
dimensions: int | None = None,
user: str | None = None,
) -> OpenAIEmbeddingsResponse:
raise NotImplementedError()
async def openai_completion(
self,
model: str,