# Video Question Answering on Video-MME short and medium videos

Source: [Video Question Answering on Video-MME short and medium videos](<https://www.sota2.com/research/sota/video-question-answering-on-video-mme-short-and-medium-videos>)

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

Data updated: 2026-06-23T10:20:59.249Z

Evaluation of Video Question Answering performance on the Video-MME short and medium videos dataset, reporting detailed accuracy metrics across temporal, object, spatial, attribute, counting, OCR, and synopsis reasoning tasks.

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) | Video-MME short and medium videos |
| Metric | Temporal Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

- [Temporal Accuracy](<https://www.sota2.com/research/sota/video-question-answering-on-video-mme-short-and-medium-videos.md?metric=Temporal%20Accuracy>): Higher is better; unit: PERCENT.

- [Action Reasoning Accuracy](<https://www.sota2.com/research/sota/video-question-answering-on-video-mme-short-and-medium-videos.md?metric=Action%20Reasoning%20Accuracy>): Higher is better; unit: PERCENT.

- [Object Reasoning Accuracy](<https://www.sota2.com/research/sota/video-question-answering-on-video-mme-short-and-medium-videos.md?metric=Object%20Reasoning%20Accuracy>): Higher is better; unit: PERCENT.

- [Spatial Reasoning Accuracy](<https://www.sota2.com/research/sota/video-question-answering-on-video-mme-short-and-medium-videos.md?metric=Spatial%20Reasoning%20Accuracy>): Higher is better; unit: PERCENT.

- [Attribute Perception Accuracy](<https://www.sota2.com/research/sota/video-question-answering-on-video-mme-short-and-medium-videos.md?metric=Attribute%20Perception%20Accuracy>): Higher is better; unit: PERCENT.

- [Counting Accuracy](<https://www.sota2.com/research/sota/video-question-answering-on-video-mme-short-and-medium-videos.md?metric=Counting%20Accuracy>): Higher is better; unit: PERCENT.

- [OCR Accuracy](<https://www.sota2.com/research/sota/video-question-answering-on-video-mme-short-and-medium-videos.md?metric=OCR%20Accuracy>): Higher is better; unit: PERCENT.

- [Synopsis Accuracy](<https://www.sota2.com/research/sota/video-question-answering-on-video-mme-short-and-medium-videos.md?metric=Synopsis%20Accuracy>): Higher is better; unit: PERCENT.

- [Overall Accuracy](<https://www.sota2.com/research/sota/video-question-answering-on-video-mme-short-and-medium-videos.md?metric=Overall%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Temporal Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Oracle-RLAIF | 50.1 | Not recorded | 68fa7be51e209131cea8b20e |
| Oracle-DPO | 49.5 | Not recorded | 68fa7be51e209131cea8b20e |
| VLM SFT | 48.2 | Not recorded | 68fa7be51e209131cea8b20e |
| VLM-RLAIF | 38.8 | Not recorded | 68fa7be51e209131cea8b20e |

### Oracle-RLAIF

Temporal Accuracy: 50.1. Result ID: 6a3a5df8fbc373592c36af8d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7be51e209131cea8b20e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7be51e209131cea8b20e.md>)

Recorded evaluation configuration: {"Optimization":"Rank-based RLAIF"}. This record does not assert independent reproduction.

### Oracle-DPO

Temporal Accuracy: 49.5. Result ID: 6a3a5df8fbc373592c36af8b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7be51e209131cea8b20e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7be51e209131cea8b20e.md>)

Recorded evaluation configuration: {"Optimization":"DPO"}. This record does not assert independent reproduction.

### VLM SFT

Temporal Accuracy: 48.2. Result ID: 6a3a5df8fbc373592c36af8e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7be51e209131cea8b20e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73b11e209131cea5df25.md>)

Recorded evaluation configuration: {"Optimization":"Supervised Fine-Tuning"}. This record does not assert independent reproduction.

### VLM-RLAIF

Temporal Accuracy: 38.8. Result ID: 6a3a5df8fbc373592c36af8c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7be51e209131cea8b20e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73b11e209131cea5df25.md>)

Recorded evaluation configuration: {"Optimization":"RLAIF (scalar reward + PPO)"}. This record does not assert independent reproduction.

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