BerriAI / litellm

Python SDK, Proxy Server (LLM Gateway) to call 100+ LLM APIs in OpenAI format - [Bedrock, Azure, OpenAI, VertexAI, Cohere, Anthropic, Sagemaker, HuggingFace, Replicate, Groq]
https://docs.litellm.ai/docs/
Other
12.5k stars 1.45k forks source link

[Feature]: Voyage API handle batching embedding calls #4241

Open krrishdholakia opened 3 months ago

krrishdholakia commented 3 months ago

The Feature

Support automatically batching embeddings according to provider limit (e.g. Vonage supports 128 batch size

Motivation, pitch

Vonage supports 128 batch size

If i send batch size 129 it returns an error

If user on proxy sends 129, would make sense as an optional setting to batch the request automatically into two requests and return as one cohesive (as a feature that can be enabled)

Twitter / LinkedIn details

cc: @PSU3D0