EleutherAI/lm-evaluation-harness

Hendrycks Math extraction rule seems too strict

Open

#2,552 opened on Dec 8, 2024

View on GitHub
 (4 comments) (1 reaction) (0 assignees)Python (3,306 forks)auto 404
good first issuevalidation

Repository metrics

Stars
 (12,755 stars)
PR merge metrics
 (Avg merge 15d 7h) (11 merged PRs in 30d)

Description

Hi thanks for the library! It seems that the way how math answer is extracted, i.e. https://github.com/EleutherAI/lm-evaluation-harness/blob/bcb4cbf464668450f1969a4f1bef28f8638d1648/lm_eval/tasks/hendrycks_math/utils.py#L20-L24, may be too strict.

For example, the following answer: ... some reasoning logic ... Thus the answer is \[ \boxed{42} \] is not extracted, because it is not a $.

Contributor guide

Research direction
Inspect the regex pattern in lm eval/tasks/hendrycks math/utils.py lines 20-24, and extend it to match LaTeX \boxed{} expressions that may be enclosed in \[ \] brackets instead of $.
Tech stack
python
Domain
machine learning
Issue type
Bug
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
2
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-3 hours
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
Active
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
Clear
Prerequisites
Git
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
75