microsoft / microsoft/LLMLingua
[Question]: How was edit sim for code tasks calculated?
Open
@iofu728 is already working on this.
Since Jul 3, 2025.
question
- Dominant language
- Python
- Stars
- 6.7k
- Forks
- 428
- Avg merge
- 2d 4h
- Merged PRs (30d)
- 1
Description
@pzs19
In the longbench evaluation fuzzywuzzy.ratio was used to calculate edit similarity (Levenshtein distance) for the predicted code lines (see here).
However I noticed that this function returns different results depending on whether python-Levenshtein, an optional dependency of fuzzywuzzy, is installed or not. Therefore I wanted to know if you used python-Levenshtein or not.
On another note, in both cases this library does not actually return the correct Levenshtein distance: https://github.com/THUDM/LongBench/issues/96
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.
Assessment
This issue has not been assessed yet.