# Egocentric Question Answering on Egolife (test)

Source: [Egocentric Question Answering on Egolife (test)](<https://www.sota2.com/research/sota/egocentric-question-answering-on-egolife-test>)

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

Data updated: 2026-03-04T13:14:36.167Z

Evaluation of model performance on the test split of the Egolife dataset for Egocentric Question Answering, reporting Accuracy.

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 | Egocentric QA |
| Dataset (including any stated split) | Egolife (test) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/egocentric-question-answering-on-egolife-test.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Reasoning Core (Ours) | 40.9 | Not recorded | 69a66c142d603e6de3e3cdad |
| Qwen2.5 VL | 38.1 | Not recorded | 69a66c142d603e6de3e3cdad |
| Gemini-1.5-Pro | 36.9 | Not recorded | 69a66c142d603e6de3e3cdad |
| GPT-4o | 36.2 | Not recorded | 69a66c142d603e6de3e3cdad |
| LLaVA-OV | 30.8 | Not recorded | 69a66c142d603e6de3e3cdad |

### Reasoning Core (Ours)

Accuracy: 40.9. Result ID: 69a830246dbb7075b94e9775.

[Reporting paper](<https://www.sota2.com/research/paper/69a66c142d603e6de3e3cdad.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a66c142d603e6de3e3cdad.md>)

Recorded evaluation configuration: {"Source":"Reproduction from official code"}. This record does not assert independent reproduction.

### Qwen2.5 VL

Accuracy: 38.1. Result ID: 69a830246dbb7075b94e9774.

[Reporting paper](<https://www.sota2.com/research/paper/69a66c142d603e6de3e3cdad.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773a1e209131cea71d3c.md>)

Recorded evaluation configuration: {"Source":"Reproduction from official code"}. This record does not assert independent reproduction.

### Gemini-1.5-Pro

Accuracy: 36.9. Result ID: 69a830246dbb7075b94e9773.

[Reporting paper](<https://www.sota2.com/research/paper/69a66c142d603e6de3e3cdad.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76361e209131cea6be11.md>)

Recorded evaluation configuration: {"Source":"Original paper"}. This record does not assert independent reproduction.

### GPT-4o

Accuracy: 36.2. Result ID: 69a830246dbb7075b94e9772.

[Reporting paper](<https://www.sota2.com/research/paper/69a66c142d603e6de3e3cdad.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75681e209131cea67a98.md>)

Recorded evaluation configuration: {"Source":"Original paper"}. This record does not assert independent reproduction.

### LLaVA-OV

Accuracy: 30.8. Result ID: 69a830246dbb7075b94e9771.

[Reporting paper](<https://www.sota2.com/research/paper/69a66c142d603e6de3e3cdad.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa756f1e209131cea67e58.md>)

[Implementation](<https://llava-vl.github.io/blog/llava-onevision>)

Recorded evaluation configuration: {"Source":"Original paper"}. This record does not assert independent reproduction.

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