Skip to content

[Bug] Dynamic LoRA load does not handle modules with greater rank correctly #7808

@lifuhuang

Description

@lifuhuang

Checklist

  • 1. I have searched related issues but cannot get the expected help.
  • 2. The bug has not been fixed in the latest version.
  • 3. Please note that if the bug-related issue you submitted lacks corresponding environment info and a minimal reproducible demo, it will be challenging for us to reproduce and resolve the issue, reducing the likelihood of receiving feedback.
  • 4. If the issue you raised is not a bug but a question, please raise a discussion at https://github.com/sgl-project/sglang/discussions/new/choose Otherwise, it will be closed.
  • 5. Please use English, otherwise it will be closed.

Describe the bug

Adding an issue to track a bug for the new dynamic lora support:

Currently the new dynamic lora support creates gpu buffer based on existing adapter max lora ranks, when a new adapter is loaded that has larger lora rank than the initial set, it might not correctly handle it (to be verified).

We need to add logic to reset existing buffers when max-lora-rank changes.

Reproduction

TODO

Environment

N/A

Metadata

Metadata

Assignees

Labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions