scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-29 01:42:19 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 100

fix(embedding): normalize list encoding_format from model config to first element

Browse source 
When `encoding_format` is configured as a list in model litellm_params (e.g.
`encoding_format: [float, base64]` to declare supported formats), the list was
forwarded verbatim to the upstream embedding API, causing vLLM to reject the
request with validation errors.

Normalize `encoding_format` to its first element when it arrives as a list,
matching the function's `Optional[str]` contract. If the list is empty, treat
it as `None` (unset). This affects all providers (hosted_vllm, openai, azure,
etc.) uniformly.

Fixes #28239
This commit is contained in:
PRABHU KIRAN VANDRANKI 2026-05-19 20:19:16 -04:00
parent a72414a061
commit eb28ed1dc6
1 changed files with 2 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
litellm/main.py
View file

@ -4809,6 +4809,8 @@ def embedding( # noqa: PLR0915
mock_response: Optional[List[float]] = kwargs.get("mock_response", None) # type: ignore
azure_ad_token_provider = kwargs.get("azure_ad_token_provider", None)
aembedding: Optional[bool] = kwargs.get("aembedding", None)
if isinstance(encoding_format, list):
encoding_format = encoding_format[0] if encoding_format else None
extra_headers = kwargs.get("extra_headers", None)
headers = kwargs.get("headers", None) or extra_headers
if headers is None: