# Multi-choice Spatial Question Answering on MSQA

Source: [Multi-choice Spatial Question Answering on MSQA](<https://www.sota2.com/research/sota/multi-choice-spatial-question-answering-on-msqa>)

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

Data updated: 2026-02-26T06:03:57.302Z

Evaluation of model performance on the MSQA dataset for Multi-choice Spatial Question Answering tasks, typically measured using a general Score metric.

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 | Multi-choice Spatial Question Answering |
| Dataset (including any stated split) | MSQA |
| Metric | Score |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 3 of 3 |
| 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

- [Score](<https://www.sota2.com/research/sota/multi-choice-spatial-question-answering-on-msqa.md?metric=Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Score | Result date | Report paper |
| --- | --- | --- | --- |
| LSceneLLM | 11.7 | Not recorded | 6940e7cf456146936284fb0a |
| Lemon | 10.68 | Not recorded | 6940e7cf456146936284fb0a |
| LEO | 7.84 | Not recorded | 6940e7cf456146936284fb0a |

### LSceneLLM

Score: 11.7. Result ID: 69984c6f5698665ec4a9c5f1.

[Reporting paper](<https://www.sota2.com/research/paper/6940e7cf456146936284fb0a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76c71e209131cea6ef9e.md>)

[Implementation](<https://github.com/Hoyyyaard/LSceneLLM>)

Recorded evaluation configuration: {"zero-shot":true}. This record does not assert independent reproduction.

### Lemon

Score: 10.68. Result ID: 69984c6f5698665ec4a9c5f3.

[Reporting paper](<https://www.sota2.com/research/paper/6940e7cf456146936284fb0a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6940e7cf456146936284fb0a.md>)

Recorded evaluation configuration: {"zero-shot":true}. This record does not assert independent reproduction.

### LEO

Score: 7.84. Result ID: 69984c6f5698665ec4a9c5f2.

[Reporting paper](<https://www.sota2.com/research/paper/6940e7cf456146936284fb0a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa732c1e209131cea5aefe.md>)

[Implementation](<https://embodied-generalist.github.io/>)

Recorded evaluation configuration: {"zero-shot":true}. This record does not assert independent reproduction.

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