Skip to content

Create a local implementation of the auto tuner using vllm #1

@hgsmn

Description

@hgsmn

As a user, I want to tune my vLLM configuration using only vLLM-native settings, so that my setup remains lightweight and free from external framework dependencies.

Therefore, we need to implement :

  • a local backend
  • add a config params to switch between Ray / Local
  • adapt differents controllers to support local backend

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions