Increase accuracy of SWT bench results #340
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- Updated eval_infer.py to improve evaluation accuracy - Modified patch_utils.py to enhance patch handling - Applied code formatting Co-authored-by: openhands <openhands@all-hands.dev>
- Revert from keep_only_test_files_in_patch back to remove_files_from_patch - Keep new evaluation arguments (--patch_types vanilla, --build_mode api) - Remove keep_only_test_files_in_patch function from patch_utils.py Co-authored-by: openhands <openhands@all-hands.dev>
Co-authored-by: openhands <openhands@all-hands.dev>
Co-authored-by: openhands <openhands@all-hands.dev>
neubig
reviewed
Jan 18, 2026
| def run_swtbench_evaluation( | ||
| predictions_file: str, | ||
| dataset: str = "eth-sri/SWT-bench_Verified_bm25_27k_zsp", | ||
| dataset: str = "princeton-nlp/SWE-bench_Verified", |
simonrosenberg
approved these changes
Jan 18, 2026
Co-authored-by: openhands <openhands@all-hands.dev>
…ench Co-authored-by: openhands <openhands@all-hands.dev>
Open
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
We were getting low SWT bench results due to harness errors. Shoutout to @simonrosenberg , we tested many things to get to this PR.
SWT-bench dataset is based on the SWE-bench dataset but contains less instances (433 vs 500). v0 implementation of SWT-bench used the SWE-bench dataset, when migrating we decided to implement SWT-bench dataset.
It seems that their harness contains a bug when running with the SWT bench dataset, but works with the SWE-bench dataset.
This fix increases the results of experiment with time stamp 26-01-16-19-09 from
to