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
14.6k stars 1.71k forks source link

[Feature]: Support native vllm server #1114

Open krrishdholakia opened 11 months ago

krrishdholakia commented 11 months ago

The Feature

The normal vllm server, supports these inputs. By supporting this, we can handle prompt formatting on the proxy instead of the individual vllm server

Screenshot 2023-12-13 at 8 06 38 AM

Motivation, pitch

User had to do more setup work once their vllm - litellm setup was done, to support the individual prompt template

Twitter / LinkedIn details

No response

krrishdholakia commented 11 months ago

User needs it to support sending top_k, top_p, etc. parameters to their vllm model