# Software Engineering Question Answering on SWE-QA Reflex

Source: [Software Engineering Question Answering on SWE-QA Reflex](<https://www.sota2.com/research/sota/software-engineering-question-answering-on-swe-qa-reflex>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-05-18T06:47:38.143Z

Evaluation of model performance on the SWE-QA Reflex dataset for Software Engineering Question Answering.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Software Engineering Question Answering |
| Dataset (including any stated split) | SWE-QA Reflex |
| Metric | Overall Score |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Overall Score](<https://www.sota2.com/research/sota/software-engineering-question-answering-on-swe-qa-reflex.md?metric=Overall%20Score>): Higher is better; unit: SCALAR.

- [Interaction Rounds](<https://www.sota2.com/research/sota/software-engineering-question-answering-on-swe-qa-reflex.md?metric=Interaction%20Rounds>): Lower is better; unit: SCALAR.

- [Tokens (K)](<https://www.sota2.com/research/sota/software-engineering-question-answering-on-swe-qa-reflex.md?metric=Tokens%20(K)>): Lower is better; unit: SIZE.

## Recorded results

| Method | Overall Score | Result date | Report paper |
| --- | --- | --- | --- |
| LaMR | 8.15 | Not recorded | 6a0a9d0adbc836b87d82b37d |
| SWE-Pruner | 8.14 | Not recorded | 6a0a9d0adbc836b87d82b37d |
| SWE-Pruner | 8.07 | Not recorded | 6a0a9d0adbc836b87d82b37d |
| Unpruned | 8.06 | Not recorded | 6a0a9d0adbc836b87d82b37d |
| LaMR | 7.88 | Not recorded | 6a0a9d0adbc836b87d82b37d |
| Unpruned | 7.75 | Not recorded | 6a0a9d0adbc836b87d82b37d |

### LaMR

Overall Score: 8.15. Result ID: 6a0ab4e447f34ce924f03853.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d0adbc836b87d82b37d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0a9d0adbc836b87d82b37d.md>)

Recorded evaluation configuration: {"Backbone":"Claude Opus 4.6"}. This record does not assert independent reproduction.

### SWE-Pruner

Overall Score: 8.14. Result ID: 6a0ab4e447f34ce924f03852.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d0adbc836b87d82b37d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6976f4f1a2b28e761d87c147.md>)

[Implementation](<https://github.com/Ayanami1314/swe-pruner>)

Recorded evaluation configuration: {"Backbone":"Claude Opus 4.6"}. This record does not assert independent reproduction.

### SWE-Pruner

Overall Score: 8.07. Result ID: 6a0ab4e447f34ce924f0384f.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d0adbc836b87d82b37d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6976f4f1a2b28e761d87c147.md>)

[Implementation](<https://github.com/Ayanami1314/swe-pruner>)

Recorded evaluation configuration: {"Backbone":"Claude Sonnet 4.5"}. This record does not assert independent reproduction.

### Unpruned

Overall Score: 8.06. Result ID: 6a0ab4e447f34ce924f03851.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d0adbc836b87d82b37d.md>)

Recorded evaluation configuration: {"Backbone":"Claude Opus 4.6"}. This record does not assert independent reproduction.

### LaMR

Overall Score: 7.88. Result ID: 6a0ab4e447f34ce924f03850.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d0adbc836b87d82b37d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0a9d0adbc836b87d82b37d.md>)

Recorded evaluation configuration: {"Backbone":"Claude Sonnet 4.5"}. This record does not assert independent reproduction.

### Unpruned

Overall Score: 7.75. Result ID: 6a0ab4e447f34ce924f0384e.

[Reporting paper](<https://www.sota2.com/research/paper/6a0a9d0adbc836b87d82b37d.md>)

Recorded evaluation configuration: {"Backbone":"Claude Sonnet 4.5"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.