# Video Question Answering on HD-EPIC 1 (test)

Source: [Video Question Answering on HD-EPIC 1 (test)](<https://www.sota2.com/research/sota/video-question-answering-on-hd-epic-1-test>)

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

Data updated: 2026-04-14T07:23:12.790Z

Evaluation of model performance on the test split of the HD-EPIC 1 dataset for Video Question Answering, covering specialized aspects like Recipe, Ingredient, Nutrition, Action, 3D, and Motion understanding.

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 | Video Question Answering |
| Dataset (including any stated split) | HD-EPIC 1 (test) |
| Metric | Recipe |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 9 of 9 |
| 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

- [Recipe](<https://www.sota2.com/research/sota/video-question-answering-on-hd-epic-1-test.md?metric=Recipe>): Higher is better; unit: PERCENT.

- [Ingredient](<https://www.sota2.com/research/sota/video-question-answering-on-hd-epic-1-test.md?metric=Ingredient>): Higher is better; unit: PERCENT.

- [Nutrition](<https://www.sota2.com/research/sota/video-question-answering-on-hd-epic-1-test.md?metric=Nutrition>): Higher is better; unit: PERCENT.

- [Action](<https://www.sota2.com/research/sota/video-question-answering-on-hd-epic-1-test.md?metric=Action>): Higher is better; unit: PERCENT.

- [3D](<https://www.sota2.com/research/sota/video-question-answering-on-hd-epic-1-test.md?metric=3D>): Higher is better; unit: PERCENT.

- [Motion](<https://www.sota2.com/research/sota/video-question-answering-on-hd-epic-1-test.md?metric=Motion>): Higher is better; unit: PERCENT.

- [Gaze](<https://www.sota2.com/research/sota/video-question-answering-on-hd-epic-1-test.md?metric=Gaze>): Higher is better; unit: PERCENT.

- [Average Performance](<https://www.sota2.com/research/sota/video-question-answering-on-hd-epic-1-test.md?metric=Average%20Performance>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Recipe | Result date | Report paper |
| --- | --- | --- | --- |
| Sample Human Baseline | 96.7 | Not recorded | 69ddca047d0172140ec79930 |
| Gemini Pro | 60.5 | Not recorded | 69ddca047d0172140ec79930 |
| Belief + VLM + RL | 54.9 | Not recorded | 69ddca047d0172140ec79930 |
| VLM + RL | 49.5 | Not recorded | 69ddca047d0172140ec79930 |
| Gemini Pro | 38 | Not recorded | 69ddca047d0172140ec79930 |
| LLaVA-Video | 36.3 | Not recorded | 69ddca047d0172140ec79930 |
| Llama 3.2 | 33.5 | Not recorded | 69ddca047d0172140ec79930 |
| VideoLlama 2 | 30.8 | Not recorded | 69ddca047d0172140ec79930 |
| LongVA | 29.6 | Not recorded | 69ddca047d0172140ec79930 |

### Sample Human Baseline

Recipe: 96.7. Result ID: 69ddeb4647f34ce924c4f4d6.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca047d0172140ec79930.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Gemini Pro

Recipe: 60.5. Result ID: 69ddeb4647f34ce924c4f4d3.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca047d0172140ec79930.md>)

Recorded evaluation configuration: {"Modality":"Video-Language"}. This record does not assert independent reproduction.

### Belief + VLM + RL

Recipe: 54.9. Result ID: 69ddeb4647f34ce924c4f4d5.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca047d0172140ec79930.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddca047d0172140ec79930.md>)

Recorded evaluation configuration: {"Modality":"Video-Language","Backbone":"InternVL-2B"}. This record does not assert independent reproduction.

### VLM + RL

Recipe: 49.5. Result ID: 69ddeb4647f34ce924c4f4d4.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca047d0172140ec79930.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddca047d0172140ec79930.md>)

Recorded evaluation configuration: {"Modality":"Video-Language","Backbone":"InternVL-2B"}. This record does not assert independent reproduction.

### Gemini Pro

Recipe: 38. Result ID: 69ddeb4647f34ce924c4f4cf.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca047d0172140ec79930.md>)

Recorded evaluation configuration: {"Modality":"Blind - Language Only"}. This record does not assert independent reproduction.

### LLaVA-Video

Recipe: 36.3. Result ID: 69ddeb4647f34ce924c4f4d2.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca047d0172140ec79930.md>)

Recorded evaluation configuration: {"Modality":"Video-Language"}. This record does not assert independent reproduction.

### Llama 3.2

Recipe: 33.5. Result ID: 69ddeb4647f34ce924c4f4ce.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca047d0172140ec79930.md>)

Recorded evaluation configuration: {"Modality":"Blind - Language Only"}. This record does not assert independent reproduction.

### VideoLlama 2

Recipe: 30.8. Result ID: 69ddeb4647f34ce924c4f4d0.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca047d0172140ec79930.md>)

Recorded evaluation configuration: {"Modality":"Video-Language"}. This record does not assert independent reproduction.

### LongVA

Recipe: 29.6. Result ID: 69ddeb4647f34ce924c4f4d1.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca047d0172140ec79930.md>)

Recorded evaluation configuration: {"Modality":"Video-Language"}. This record does not assert independent reproduction.

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