Skip to content

Feature/ifbench #113

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 4 commits into from
Jul 25, 2025
Merged

Feature/ifbench #113

merged 4 commits into from
Jul 25, 2025

Conversation

Jianshu1only
Copy link
Collaborator

Checklist Before Starting

  • Search for similar PR(s).

What does this PR do?

Create IFBench dataset verification function and data preprocess code

High-Level Design

Add ifbench in reward_score

Specific Changes

Add ifbench verification function in reward_score, add ood_ifbench in util, add data preprocess function

API

NA

Usage Example

NA

Test

python verl/utils/reward_score/ifbench/test_ifbench.py
sbatch verl/scripts/train/ifbench_test.sh

Additional Info.

NA

Checklist Before Submitting

  • [✓] Read the Contribute Guide.
  • [✓] Apply pre-commit checks.
  • [✓] Add [BREAKING] to the PR title if it breaks any API.
  • [✓] Update the documentation about your changes in the docs.
  • [✓] New CI unit test(s) are added to cover the code path.
  • [✓] Rely on existing unit tests on CI that covers the code path.

"""Split the fixed IFBench data into 90% train and 10% test."""

# Input file (fixed data)
input_file = "/mnt/sharefs/users/jianshu.she/ood__ifbench_95.1k_fixed.parquet"
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There are absolute paths here. Is this file necessary for production?

"""Split the fixed IFBench data into 90% train and 10% test."""

# Input file (fixed data)
input_file = "/mnt/sharefs/users/jianshu.she/ood__ifbench_95.1k_fixed.parquet"
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The same


# Fix the data
print("\n=== Fixing data ===")
fix_ifbench_data(original_file, "/mnt/sharefs/users/jianshu.she/ood__ifbench_95.1k_fixed.parquet")
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The same

@BlankCheng
Copy link
Collaborator

Hi @Jianshu1only , this looks awesome! One issue is some files are not directly used in RL training, which don't need to be included, e.g., check_ifbench_data.py. We can keep the minimal runnable files for RL here. Please correct me if I misunderstand. Also, let's remove the absolute paths involved.
Overall, the IFBench would be great to have. Thanks for the contribution!

@Jianshu1only
Copy link
Collaborator Author

Hi @BlankCheng , I’ve deleted all unnecessary files in the IFbench folder. Please review the PR again when you have a chance. Thanks!

@BlankCheng
Copy link
Collaborator

Thanks @Jianshu1only! I delete the bash files as we only keep the template there. Let's merge.

@BlankCheng BlankCheng self-requested a review July 25, 2025 04:22
@BlankCheng BlankCheng merged commit bf5c3dc into main Jul 25, 2025
@BlankCheng BlankCheng deleted the feature/ifbench branch July 25, 2025 04:28
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants