Skip to content
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

Adding Arena Hard Auto #65

Open
wants to merge 9 commits into
base: main
Choose a base branch
from

Conversation

asuvarna31
Copy link

@asuvarna31 asuvarna31 commented Jan 28, 2025

Sample command :

python -m eval.eval\
    --model hf\
    --tasks arena_hard_auto\
    --model_args "pretrained=Qwen/Qwen1.5-1.8B-Chat" \
    --batch_size 2 \
    --output_path logs \
    --annotator_model gpt-4o-mini-2024-07-18 \ 

Sample Output :

"results": {
        "score": 5.56,
        "avg_tokens": 635.0
      }

@asuvarna31
Copy link
Author

@neginraoof @RyanMarten feel free to review and test this.

Copy link
Collaborator

@neginraoof neginraoof left a comment

Choose a reason for hiding this comment

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

Thanks a lot! Can you also update the reproduced_benchmarks.md with results?


## save a leaderboard in leaderboard dir
subprocess.run(['python', '-m', 'eval.chat_benchmarks.arena_hard_auto.show_result', '--judge-name', f'{self.annotator_model}', '--output'])
df = pd.read_csv(f'eval/chat_benchmarks/arena_hard_auto/leaderboard/arena_hard_leaderboard_{self.annotator_model}.csv')
Copy link
Collaborator

Choose a reason for hiding this comment

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

Thanks a lot @asuvarna31 !
Can we just avoid read/write to file? seems like we can just use the input model_results here?

Copy link
Author

Choose a reason for hiding this comment

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

Addressed your comments @neginraoof

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