open-compass / open-compass/opencompass
[Feature] Add Math Dataset as in "Let’s Verify Step by Step"
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 7.5k
- Forks
- 869
- Avg merge
- 17h 52m
- Merged PRs (30d)
- 13
Description
Describe the feature
Some community models have adopted the methods from the "Let’s Verify Step by Step" paper for training, thus there is a risk of Math data leakage.
https://arxiv.org/pdf/2305.20050.pdf
We may consider using a testing method with only 500 samples for reference.
On the other hand, our current evaluation approach cannot measure the extent of data leakage at all. Whether there is a need to treat Math separately is also worth discussing.
Will you implement it?
- I would like to implement this feature and create a PR!
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reading the linked “Let’s Verify Step by Step” paper and reviewing the repository’s current evaluation approach. The issue proposes a separate math evaluation using about 500 samples, but names no files, tests, entry points, or agreed completion criteria; clarify the design and leakage-measurement requirements before implementation.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- data, machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100