Skip to content

could not reproduce the results #2

Description

@lihuiliullh

Dear Authors,

I followed the instructions in your README, including the fine-tuning step for the LLM, followed by inference and self-improvement using the 1B model. However, I obtained a performance of around 56% to 60%, rather than the reported 73.5%.

I also followed the hyperparameters and settings provided in the paper as closely as possible. Could you please let me know if there are any additional settings, checkpoints, or implementation details that are necessary to reproduce the reported 73.5% result?

I would greatly appreciate any guidance on how to reproduce the reported results.

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