Skip to content

How does scaling up search turns using ASearcher perform for 3B models? #8

Open

Description

@cjz007

Hi!

Nice work!

May I ask if you have tried scaling up search turns using ASearcher for 3B models? I think Search-R1 has the experiment for 3B models, but the result is not shown in ASearcher's paper. May I ask does scaling up search turns also help 3B models?

Thank you!

Activity

  1. samjia2000 commented on Aug 24, 2025

    @samjia2000
    Collaborator

    Hi, we have not run training for 3B models. I have just launched a run for 3B model. The results will be available soon. If you are interested, you can also run single-node 3B training on your own, which is expected to finish within 6~8 hours.

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