mirror of
https://github.com/BerriAI/litellm.git
synced 2025-04-27 03:34:10 +00:00
0519c0c507
* test: fix import for test
* fix: fix bad error string
* docs: cleanup files docs
* fix(files/main.py): cleanup error string
* style: initial commit with a provider/config pattern for files api
google ai studio files api onboarding
* fix: test
* feat(gemini/files/transformation.py): support gemini files api response transformation
* fix(gemini/files/transformation.py): return file id as gemini uri
allows id to be passed in to chat completion request, just like openai
* feat(llm_http_handler.py): support async route for files api on llm_http_handler
* fix: fix linting errors
* fix: fix model info check
* fix: fix ruff errors
* fix: fix linting errors
* Revert "fix: fix linting errors"
This reverts commit 926a5a527f.
* fix: fix linting errors
* test: fix test
* test: fix tests
81 lines
2.5 KiB
Python
81 lines
2.5 KiB
Python
"""
|
|
*New config* for using aiohttp to make the request to the custom OpenAI-like provider
|
|
|
|
This leads to 10x higher RPS than httpx
|
|
https://github.com/BerriAI/litellm/issues/6592
|
|
|
|
New config to ensure we introduce this without causing breaking changes for users
|
|
"""
|
|
|
|
from typing import TYPE_CHECKING, Any, List, Optional
|
|
|
|
from aiohttp import ClientResponse
|
|
|
|
from litellm.llms.openai_like.chat.transformation import OpenAILikeChatConfig
|
|
from litellm.types.llms.openai import AllMessageValues
|
|
from litellm.types.utils import Choices, ModelResponse
|
|
|
|
if TYPE_CHECKING:
|
|
from litellm.litellm_core_utils.litellm_logging import Logging as _LiteLLMLoggingObj
|
|
|
|
LiteLLMLoggingObj = _LiteLLMLoggingObj
|
|
else:
|
|
LiteLLMLoggingObj = Any
|
|
|
|
|
|
class AiohttpOpenAIChatConfig(OpenAILikeChatConfig):
|
|
def get_complete_url(
|
|
self,
|
|
api_base: Optional[str],
|
|
api_key: Optional[str],
|
|
model: str,
|
|
optional_params: dict,
|
|
litellm_params: dict,
|
|
stream: Optional[bool] = None,
|
|
) -> str:
|
|
"""
|
|
Ensure - /v1/chat/completions is at the end of the url
|
|
|
|
"""
|
|
if api_base is None:
|
|
api_base = "https://api.openai.com"
|
|
|
|
if not api_base.endswith("/chat/completions"):
|
|
api_base += "/chat/completions"
|
|
return api_base
|
|
|
|
def validate_environment(
|
|
self,
|
|
headers: dict,
|
|
model: str,
|
|
messages: List[AllMessageValues],
|
|
optional_params: dict,
|
|
api_key: Optional[str] = None,
|
|
api_base: Optional[str] = None,
|
|
) -> dict:
|
|
return {"Authorization": f"Bearer {api_key}"}
|
|
|
|
async def transform_response( # type: ignore
|
|
self,
|
|
model: str,
|
|
raw_response: ClientResponse,
|
|
model_response: ModelResponse,
|
|
logging_obj: LiteLLMLoggingObj,
|
|
request_data: dict,
|
|
messages: List[AllMessageValues],
|
|
optional_params: dict,
|
|
litellm_params: dict,
|
|
encoding: Any,
|
|
api_key: Optional[str] = None,
|
|
json_mode: Optional[bool] = None,
|
|
) -> ModelResponse:
|
|
_json_response = await raw_response.json()
|
|
model_response.id = _json_response.get("id")
|
|
model_response.choices = [
|
|
Choices(**choice) for choice in _json_response.get("choices")
|
|
]
|
|
model_response.created = _json_response.get("created")
|
|
model_response.model = _json_response.get("model")
|
|
model_response.object = _json_response.get("object")
|
|
model_response.system_fingerprint = _json_response.get("system_fingerprint")
|
|
return model_response
|