# Video Question Answering on CinePile Source: [Video Question Answering on CinePile]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:30:23.479Z Evaluation of model performance on the CinePile dataset for Video Question Answering, 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) | CinePile | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 13 of 13 | | 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 | | --- | --- | --- | --- | | VITED (Temporal Evidence Distillation) | 58.98 | Not recorded | 68fa77751e209131cea733ca | | VITED (Dense Caption Distillation) | 57.85 | Not recorded | 68fa77751e209131cea733ca | | VITED (Temporal Evidence Distillation) | 56.39 | Not recorded | 68fa77751e209131cea733ca | | TimeChat (Video Instruction Tuning) | 55.86 | Not recorded | 68fa77751e209131cea733ca | | LLaVA-Video (Video Instruction Tuning) | 54.97 | Not recorded | 68fa77751e209131cea733ca | | LLaVA-Video (Chain-of-Thought) | 54.82 | Not recorded | 68fa77751e209131cea733ca | | VITED (Dense Caption Distillation) | 54.79 | Not recorded | 68fa77751e209131cea733ca | | LLaVA-Video | 53.77 | Not recorded | 68fa77751e209131cea733ca | | LLaVA-OneVision | 46.42 | Not recorded | 68fa77751e209131cea733ca | | LLaMA-3.2V | 39.55 | Not recorded | 68fa77751e209131cea733ca | | VideoLLaVA | 22.51 | Not recorded | 68fa77751e209131cea733ca | | TimeChat (Chain-of-Thought) | 14.27 | Not recorded | 68fa77751e209131cea733ca | | TimeChat | 13.67 | Not recorded | 68fa77751e209131cea733ca | ### VITED (Temporal Evidence Distillation) Accuracy: 58.98. Result ID: 69952ae03badfaffe610f5fa. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Params":"7B","Base Model":"TimeChat Base"}. This record does not assert independent reproduction. ### VITED (Dense Caption Distillation) Accuracy: 57.85. Result ID: 69952ae03badfaffe610f5f9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Params":"7B","Base Model":"TimeChat Base"}. This record does not assert independent reproduction. ### VITED (Temporal Evidence Distillation) Accuracy: 56.39. Result ID: 69952ae03badfaffe610f5ff. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Params":"7B","Base Model":"LLaVA-Video Base"}. This record does not assert independent reproduction. ### TimeChat (Video Instruction Tuning) Accuracy: 55.86. Result ID: 69952ae03badfaffe610f5f8. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Params":"7B","Base Model":"TimeChat Base"}. This record does not assert independent reproduction. ### LLaVA-Video (Video Instruction Tuning) Accuracy: 54.97. Result ID: 69952ae03badfaffe610f5fd. [Reporting paper]() Recorded evaluation configuration: {"Params":"7B","Base Model":"LLaVA-Video Base"}. This record does not assert independent reproduction. ### LLaVA-Video (Chain-of-Thought) Accuracy: 54.82. Result ID: 69952ae03badfaffe610f5fc. [Reporting paper]() Recorded evaluation configuration: {"Params":"7B","Base Model":"LLaVA-Video Base"}. This record does not assert independent reproduction. ### VITED (Dense Caption Distillation) Accuracy: 54.79. Result ID: 69952ae03badfaffe610f5fe. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Params":"7B","Base Model":"LLaVA-Video Base"}. This record does not assert independent reproduction. ### LLaVA-Video Accuracy: 53.77. Result ID: 69952ae03badfaffe610f5fb. [Reporting paper]() Recorded evaluation configuration: {"Params":"7B","Base Model":"LLaVA-Video Base"}. This record does not assert independent reproduction. ### LLaVA-OneVision Accuracy: 46.42. Result ID: 69952ae03badfaffe610f5f5. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Params":"7B"}. This record does not assert independent reproduction. ### LLaMA-3.2V Accuracy: 39.55. Result ID: 69952ae03badfaffe610f5f4. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Params":"11B"}. This record does not assert independent reproduction. ### VideoLLaVA Accuracy: 22.51. Result ID: 69952ae03badfaffe610f5f3. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Params":"7B"}. This record does not assert independent reproduction. ### TimeChat (Chain-of-Thought) Accuracy: 14.27. Result ID: 69952ae03badfaffe610f5f7. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Params":"7B","Base Model":"TimeChat Base"}. This record does not assert independent reproduction. ### TimeChat Accuracy: 13.67. Result ID: 69952ae03badfaffe610f5f6. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Params":"7B","Base Model":"TimeChat Base"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.