Name | Modified | Size | Downloads / Week |
---|---|---|---|
Parent folder | |||
README.md | 2025-02-28 | 3.1 kB | |
v1.0.1_ SOTA on SWE-Bench Full source code.tar.gz | 2025-02-28 | 25.8 MB | |
v1.0.1_ SOTA on SWE-Bench Full source code.zip | 2025-02-28 | 26.0 MB | |
Totals: 3 Items | 51.8 MB | 0 |
SWE-agent 1.0.1
News: After our announcements for SOTA on SWE-Bench Lite and Verified, we now can claim SOTA on the full set of 2k GitHub issues of SWE-Bench full:
Interestingly, the improvement on the issues that are not also in the Lite/Verified subsets is much higher than the gain for the Lite/Verified subsets. Evaluating only on Lite/Verified doesn’t tell the whole story! -
What's Changed
This fixup release brings fixes mostly to the compatibility with local models. We have also significantly expanded the documentation in that aspect (models & keys documentation).
Changed
- Change: Make anthropic_filemap the new default config by @klieret in https://github.com/SWE-agent/SWE-agent/pull/927
Added
- Enh: Set timeout for post_startup_commands by @klieret in https://github.com/SWE-agent/SWE-agent/pull/973
- Enh: Allow to override max_input_tokens for local models by @klieret in https://github.com/SWE-agent/SWE-agent/pull/992
Fixes
- Fix: Handling local models cost lookup issues by @klieret in https://github.com/SWE-agent/SWE-agent/pull/937
- Fix: Requires-python >= 3.11 by @klieret in https://github.com/SWE-agent/SWE-agent/pull/940
- traj inspector viewport reset by @klieret in https://github.com/SWE-agent/SWE-agent/pull/946
- Fix: Reset viewport when next/prev step/traj by @klieret in https://github.com/SWE-agent/SWE-agent/pull/948
- Fix: Disable highlighting of model outputs by @klieret in https://github.com/SWE-agent/SWE-agent/pull/949
- Fix: Create PRs by @klieret in https://github.com/SWE-agent/SWE-agent/pull/954
- Fix: Add init,py to agent/hooks by @RNabel in https://github.com/SWE-agent/SWE-agent/pull/961
- Fix: Pin textual to version 1.0.0 by @RNabel in https://github.com/SWE-agent/SWE-agent/pull/960
- Fix: OpenAI API: Don't pass None tool_calls to the OpenAI API by @RNabel in https://github.com/SWE-agent/SWE-agent/pull/967
- Fix: Forces platform to be linux/amd64 for swe-bench batch runs by @carlosejimenez in https://github.com/SWE-agent/SWE-agent/pull/942
- Fix "TypeError: Cannot read properties of null (reading 'replace')" in Trajectory viewer by @0xba1a in https://github.com/SWE-agent/SWE-agent/pull/989
- Fix: No retries if costs cannot be calculated by @klieret in https://github.com/SWE-agent/SWE-agent/pull/990
- Fix: Race condition/size change during iteration by @klieret in https://github.com/SWE-agent/SWE-agent/pull/993
- Fix: Handle total cost limit exceeded by @klieret in https://github.com/SWE-agent/SWE-agent/pull/994
New Contributors
- @RNabel made their first contribution in https://github.com/SWE-agent/SWE-agent/pull/961
- @dhruvji made their first contribution in https://github.com/SWE-agent/SWE-agent/pull/963
- @0xba1a made their first contribution in https://github.com/SWE-agent/SWE-agent/pull/989
Full Changelog: https://github.com/SWE-agent/SWE-agent/compare/v1.0.0...v1.0.1