# Video Question Answering on LLaVA-Video

Source: [Video Question Answering on LLaVA-Video](<https://www.sota2.com/research/sota/video-question-answering-on-llava-video>)

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

Data updated: 2026-04-21T10:17:33.481Z

Evaluation of model performance on the LLaVA-Video dataset for Video Question Answering, utilizing a specialized set of metrics including VYR, TIHR, and Overall score.

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) | LLaVA-Video |
| Metric | VYR |
| Unit | SCALAR |
| 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

- [VYR](<https://www.sota2.com/research/sota/video-question-answering-on-llava-video.md?metric=VYR>): Higher is better; unit: SCALAR.

- [TIHR](<https://www.sota2.com/research/sota/video-question-answering-on-llava-video.md?metric=TIHR>): Higher is better; unit: SCALAR.

- [WHR](<https://www.sota2.com/research/sota/video-question-answering-on-llava-video.md?metric=WHR>): Higher is better; unit: SCALAR.

- [SCSI](<https://www.sota2.com/research/sota/video-question-answering-on-llava-video.md?metric=SCSI>): Higher is better; unit: SCALAR.

- [HRR](<https://www.sota2.com/research/sota/video-question-answering-on-llava-video.md?metric=HRR>): Higher is better; unit: SCALAR.

- [HSR](<https://www.sota2.com/research/sota/video-question-answering-on-llava-video.md?metric=HSR>): Higher is better; unit: SCALAR.

- [TIB](<https://www.sota2.com/research/sota/video-question-answering-on-llava-video.md?metric=TIB>): Higher is better; unit: SCALAR.

- [ICR](<https://www.sota2.com/research/sota/video-question-answering-on-llava-video.md?metric=ICR>): Higher is better; unit: SCALAR.

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

- [TLSR](<https://www.sota2.com/research/sota/video-question-answering-on-llava-video.md?metric=TLSR>): Higher is better; unit: SCALAR.

- [ASLSR](<https://www.sota2.com/research/sota/video-question-answering-on-llava-video.md?metric=ASLSR>): Higher is better; unit: SCALAR.

- [AALSR](<https://www.sota2.com/research/sota/video-question-answering-on-llava-video.md?metric=AALSR>): Higher is better; unit: SCALAR.

- [SRLSR](<https://www.sota2.com/research/sota/video-question-answering-on-llava-video.md?metric=SRLSR>): Higher is better; unit: SCALAR.

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

## Recorded results

| Method | VYR | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen3-VL-30B | 63.2 | Not recorded | 69e7049f7d0172140ec7ed93 |
| Baseline (Qwen3-VL-8B-Instruct) | 59.5 | Not recorded | 69e7049f7d0172140ec7ed93 |
| Baseline+SFT | 57 | Not recorded | 69e7049f7d0172140ec7ed93 |
| Baseline+CoT | 55 | Not recorded | 69e7049f7d0172140ec7ed93 |
| Internvl3.5-VL-241B | 52.4 | Not recorded | 69e7049f7d0172140ec7ed93 |
| Qwen3-VL-235B | 49.6 | Not recorded | 69e7049f7d0172140ec7ed93 |
| Baseline+CoT+SFT | 30.9 | Not recorded | 69e7049f7d0172140ec7ed93 |
| VTHM-MoE | 4.6 | Not recorded | 69e7049f7d0172140ec7ed93 |
| Gemini-3.1-Pro | 1.1 | Not recorded | 69e7049f7d0172140ec7ed93 |

### Qwen3-VL-30B

VYR: 63.2. Result ID: 69e74eac47f34ce924ce2806.

[Reporting paper](<https://www.sota2.com/research/paper/69e7049f7d0172140ec7ed93.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Parameters":"30B"}. This record does not assert independent reproduction.

### Baseline (Qwen3-VL-8B-Instruct)

VYR: 59.5. Result ID: 69e74eac47f34ce924ce2809.

[Reporting paper](<https://www.sota2.com/research/paper/69e7049f7d0172140ec7ed93.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B-Instruct"}. This record does not assert independent reproduction.

### Baseline+SFT

VYR: 57. Result ID: 69e74eac47f34ce924ce280b.

[Reporting paper](<https://www.sota2.com/research/paper/69e7049f7d0172140ec7ed93.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B-Instruct","Supervised Fine-Tuning":"true"}. This record does not assert independent reproduction.

### Baseline+CoT

VYR: 55. Result ID: 69e74eac47f34ce924ce280a.

[Reporting paper](<https://www.sota2.com/research/paper/69e7049f7d0172140ec7ed93.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B-Instruct","Chain-of-Thought":"true"}. This record does not assert independent reproduction.

### Internvl3.5-VL-241B

VYR: 52.4. Result ID: 69e74eac47f34ce924ce2808.

[Reporting paper](<https://www.sota2.com/research/paper/69e7049f7d0172140ec7ed93.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7af01e209131cea85dc9.md>)

Recorded evaluation configuration: {"Parameters":"241B"}. This record does not assert independent reproduction.

### Qwen3-VL-235B

VYR: 49.6. Result ID: 69e74eac47f34ce924ce2807.

[Reporting paper](<https://www.sota2.com/research/paper/69e7049f7d0172140ec7ed93.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Parameters":"235B"}. This record does not assert independent reproduction.

### Baseline+CoT+SFT

VYR: 30.9. Result ID: 69e74eac47f34ce924ce280c.

[Reporting paper](<https://www.sota2.com/research/paper/69e7049f7d0172140ec7ed93.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B-Instruct","Chain-of-Thought":"true","Supervised Fine-Tuning":"true"}. This record does not assert independent reproduction.

### VTHM-MoE

VYR: 4.6. Result ID: 69e74eac47f34ce924ce280d.

[Reporting paper](<https://www.sota2.com/research/paper/69e7049f7d0172140ec7ed93.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e7049f7d0172140ec7ed93.md>)

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

### Gemini-3.1-Pro

VYR: 1.1. Result ID: 69e74eac47f34ce924ce2805.

[Reporting paper](<https://www.sota2.com/research/paper/69e7049f7d0172140ec7ed93.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

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

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