# Video Question Answering on MSRVTT-QA LLaVA-Hound out-of-domain (test) Source: [Video Question Answering on MSRVTT-QA LLaVA-Hound out-of-domain (test)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:29:57.278Z Evaluation of model performance on the out-of-domain test split of the MSRVTT-QA dataset, specifically using LLaVA-Hound generated questions, for Video Question Answering tasks, 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 | Video Question Answering | | Dataset (including any stated split) | MSRVTT-QA LLaVA-Hound out-of-domain (test) | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 11 of 11 | | 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](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | MM1.5-Video | 59.3 | Not recorded | 68fa75051e209131cea65504 | | MM1.5-Video | 57.2 | Not recorded | 68fa75051e209131cea65504 | | MM1.5-Video | 55.3 | Not recorded | 68fa75051e209131cea65504 | | LLAVA-HOUND-SFT | 52.6 | Not recorded | 68fa75051e209131cea65504 | | MM1.5-Video | 48.1 | Not recorded | 68fa75051e209131cea65504 | | MM1.5-Video | 46 | Not recorded | 68fa75051e209131cea65504 | | MM1.5-Video | 43.3 | Not recorded | 68fa75051e209131cea65504 | | Video-LLaVA | 30.8 | Not recorded | 68fa75051e209131cea65504 | | Chat-UniVi | 25.9 | Not recorded | 68fa75051e209131cea65504 | | Video-ChatGPT | 25.7 | Not recorded | 68fa75051e209131cea65504 | | LLaMA-VID | 25 | Not recorded | 68fa75051e209131cea65504 | ### MM1.5-Video Accuracy: 59.3. Result ID: 699370673a6c31484915ef60. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"7B","Protocol":"SFT"}. This record does not assert independent reproduction. ### MM1.5-Video Accuracy: 57.2. Result ID: 699370673a6c31484915ef5f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"3B","Protocol":"SFT"}. This record does not assert independent reproduction. ### MM1.5-Video Accuracy: 55.3. Result ID: 699370673a6c31484915ef5e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"1B","Protocol":"SFT"}. This record does not assert independent reproduction. ### LLAVA-HOUND-SFT Accuracy: 52.6. Result ID: 699370673a6c31484915ef5a. [Reporting paper]() Recorded evaluation configuration: {"Model Scale":"7B"}. This record does not assert independent reproduction. ### MM1.5-Video Accuracy: 48.1. Result ID: 699370673a6c31484915ef5d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"7B","Protocol":"Training-free"}. This record does not assert independent reproduction. ### MM1.5-Video Accuracy: 46. Result ID: 699370673a6c31484915ef5c. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"3B","Protocol":"Training-free"}. This record does not assert independent reproduction. ### MM1.5-Video Accuracy: 43.3. Result ID: 699370673a6c31484915ef5b. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"1B","Protocol":"Training-free"}. This record does not assert independent reproduction. ### Video-LLaVA Accuracy: 30.8. Result ID: 699370673a6c31484915ef59. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"7B"}. This record does not assert independent reproduction. ### Chat-UniVi Accuracy: 25.9. Result ID: 699370673a6c31484915ef58. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"7B"}. This record does not assert independent reproduction. ### Video-ChatGPT Accuracy: 25.7. Result ID: 699370673a6c31484915ef56. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"7B"}. This record does not assert independent reproduction. ### LLaMA-VID Accuracy: 25. Result ID: 699370673a6c31484915ef57. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"7B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.