Skip to content

New load balancing strategy to only switch the model once the previous model is run out #552

Description

@lamualfa

This strategy is useful for managing multiple subscription based models (non pay-as-you-go). Current available strategy such as round-robin will deprive the benefit of token caching because it will switch the model every-time based on the setup weights.

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