Skip to content

How do you handle timeout for broadcasting? #48

Description

@zfj1998

我注意到您在这个地方用了broadcast来同步group rank的数据。但我有一些疑问:

  1. 这里直接用src=0,如何保证group 1的rank 0不会错误地broadcast到其它group的ranks?
  2. 如何避免tool call花费时间太长,导致broadcast超时?
  3. 为什么不在ray_trainer实现tool call的逻辑,这样就不需要对vllm的model parallel做同步了

broadcast_data = vllm_ps._TP.broadcast_object(broadcast_data, src=0)

No activity

Activity on this issue will appear here.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

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