# Audio-Visual Question Answering on Music-AVQA 2000 samples

Source: [Audio-Visual Question Answering on Music-AVQA 2000 samples](<https://www.sota2.com/research/sota/audio-visual-question-answering-on-music-avqa-2000-samples>)

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

Data updated: 2026-02-26T05:24:04.926Z

Evaluation of model performance on the 2000-sample subset of the Music-AVQA dataset for the Audio-Visual Question Answering task, reporting ASR percentage.

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 | Audio-Visual Question Answering |
| Dataset (including any stated split) | Music-AVQA 2000 samples |
| Metric | ASR Rate |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 7 of 7 |
| 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

- [ASR Rate](<https://www.sota2.com/research/sota/audio-visual-question-answering-on-music-avqa-2000-samples.md?metric=ASR%20Rate>): Higher is better; unit: PERCENT.

## Recorded results

| Method | ASR Rate | Result date | Report paper |
| --- | --- | --- | --- |
| Combined Loss | 13.8 | Not recorded | 6976f519a2b28e761d87c22c |
| Encoder-Based Cosine Similarity Loss | 12.92 | Not recorded | 6976f519a2b28e761d87c22c |
| Negative Language Modeling Loss | 10.74 | Not recorded | 6976f519a2b28e761d87c22c |
| Hidden-State Similarity Loss | 10.28 | Not recorded | 6976f519a2b28e761d87c22c |
| Audio Attention Amplification Loss | 9.81 | Not recorded | 6976f519a2b28e761d87c22c |
| Attention Randomization Loss | 9.81 | Not recorded | 6976f519a2b28e761d87c22c |
| Vision Attention Suppression Loss | 7.94 | Not recorded | 6976f519a2b28e761d87c22c |

### Combined Loss

ASR Rate: 13.8. Result ID: 699a052b5698665ec4878b90.

[Reporting paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

Recorded evaluation configuration: {"Attack Objective":"L(combined)","Source Model":"VideoLLAMA2","Training Dataset":"AVQA","Training Epochs":"150"}. This record does not assert independent reproduction.

### Encoder-Based Cosine Similarity Loss

ASR Rate: 12.92. Result ID: 699a052b5698665ec4878b8b.

[Reporting paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

Recorded evaluation configuration: {"Attack Objective":"L(cos)","Source Model":"VideoLLAMA2","Training Dataset":"AVQA","Training Epochs":"150"}. This record does not assert independent reproduction.

### Negative Language Modeling Loss

ASR Rate: 10.74. Result ID: 699a052b5698665ec4878b8a.

[Reporting paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

Recorded evaluation configuration: {"Attack Objective":"L\_negLM","Source Model":"VideoLLAMA2","Training Dataset":"AVQA","Training Epochs":"150"}. This record does not assert independent reproduction.

### Hidden-State Similarity Loss

ASR Rate: 10.28. Result ID: 699a052b5698665ec4878b8f.

[Reporting paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

Recorded evaluation configuration: {"Attack Objective":"L(hidden-cos)","Source Model":"VideoLLAMA2","Training Dataset":"AVQA","Training Epochs":"150"}. This record does not assert independent reproduction.

### Audio Attention Amplification Loss

ASR Rate: 9.81. Result ID: 699a052b5698665ec4878b8d.

[Reporting paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

Recorded evaluation configuration: {"Attack Objective":"L(audioatt)","Source Model":"VideoLLAMA2","Training Dataset":"AVQA","Training Epochs":"150"}. This record does not assert independent reproduction.

### Attention Randomization Loss

ASR Rate: 9.81. Result ID: 699a052b5698665ec4878b8e.

[Reporting paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

Recorded evaluation configuration: {"Attack Objective":"L(randatt)","Source Model":"VideoLLAMA2","Training Dataset":"AVQA","Training Epochs":"150"}. This record does not assert independent reproduction.

### Vision Attention Suppression Loss

ASR Rate: 7.94. Result ID: 699a052b5698665ec4878b8c.

[Reporting paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6976f519a2b28e761d87c22c.md>)

Recorded evaluation configuration: {"Attack Objective":"L(visionatt)","Source Model":"VideoLLAMA2","Training Dataset":"AVQA","Training Epochs":"150"}. This record does not assert independent reproduction.

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