phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-10-04 04:04:14 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

chore: turn OpenAIMixin into a pydantic.BaseModel

Browse source 
- implement get_api_key instead of relying on LiteLLMOpenAIMixin.get_api_key
 - remove use of LiteLLMOpenAIMixin
 - add default initialize/shutdown methods to OpenAIMixin
 - remove __init__s to allow proper pydantic construction
 - remove dead code from vllm adapter and associated / duplicate unit tests
 - update vllm adapter to use openaimixin for model registration
 - remove ModelRegistryHelper from fireworks & together adapters
 - remove Inference from nvidia adapter
 - complete type hints on embedding_model_metadata
 - allow extra fields on OpenAIMixin, for model_store, __provider_id__, etc
 - new recordings for ollama
This commit is contained in:
Matthew Farrellee 2025-10-02 20:47:54 -04:00
parent ce77c27ff8
commit 60f0056cbc
57 changed files with 12520 additions and 1254 deletions
Download patch file Download diff file Expand all files Collapse all files

20
llama_stack/providers/remote/inference/together/together.py
View file

@ -11,7 +11,6 @@ from together.constants import BASE_URL
from llama_stack.apis.inference import (
ChatCompletionRequest,
Inference,
LogProbConfig,
OpenAIEmbeddingsResponse,
ResponseFormat,
@ -22,7 +21,6 @@ from llama_stack.apis.inference.inference import OpenAIEmbeddingUsage
from llama_stack.apis.models import Model, ModelType
from llama_stack.core.request_headers import NeedsRequestProviderData
from llama_stack.log import get_logger
from llama_stack.providers.utils.inference.model_registry import ModelRegistryHelper
from llama_stack.providers.utils.inference.openai_compat import (
convert_message_to_openai_dict,
get_sampling_options,
@ -38,8 +36,10 @@ from .config import TogetherImplConfig
logger = get_logger(name=__name__, category="inference::together")
class TogetherInferenceAdapter(OpenAIMixin, Inference, NeedsRequestProviderData):
embedding_model_metadata = {
class TogetherInferenceAdapter(OpenAIMixin, NeedsRequestProviderData):
config: TogetherImplConfig
embedding_model_metadata: dict[str, dict[str, int]] = {
"togethercomputer/m2-bert-80M-32k-retrieval": {"embedding_dimension": 768, "context_length": 32768},
"BAAI/bge-large-en-v1.5": {"embedding_dimension": 1024, "context_length": 512},
"BAAI/bge-base-en-v1.5": {"embedding_dimension": 768, "context_length": 512},
@ -47,11 +47,7 @@ class TogetherInferenceAdapter(OpenAIMixin, Inference, NeedsRequestProviderData)
"intfloat/multilingual-e5-large-instruct": {"embedding_dimension": 1024, "context_length": 512},
}
def __init__(self, config: TogetherImplConfig) -> None:
ModelRegistryHelper.__init__(self)
self.config = config
self.allowed_models = config.allowed_models
self._model_cache: dict[str, Model] = {}
_model_cache: dict[str, Model] = {}
def get_api_key(self):
return self.config.api_key.get_secret_value()
@ -59,12 +55,6 @@ class TogetherInferenceAdapter(OpenAIMixin, Inference, NeedsRequestProviderData)
def get_base_url(self):
return BASE_URL
async def initialize(self) -> None:
pass
async def shutdown(self) -> None:
pass
def _get_client(self) -> AsyncTogether:
together_api_key = None
config_api_key = self.config.api_key.get_secret_value() if self.config.api_key else None