ActivityNet-QA
The ActivityNet-QA dataset contains 58,000 human-annotated QA pairs on 5,800 videos derived from the popular ActivityNet dataset.
How to run
ltb · activitynet-qa
$ npm i -g lettokenburn-cli$ ltb bench pull activitynet-qa --n 50$ ltb bench run ./activitynet-qa -a claude-code -m anthropic/claude-opus-4-8$ ltb bench push
Runs happen on your machine with your own subscription; only the score summary is pushed. CLI docs
Scores
1 sourced entry| Harness | Source | Confidence | Link | |||
|---|---|---|---|---|---|---|
| LLaVA-Mini | 53.5 | N/A | Independent | medium | Jan 7, 2025 | arxiv.org |
Every number links to where it was published. Low-confidence rows are muted and need per-cell verification.
Reproduced by the community
All runsLoading community runs…
About
The ActivityNet-QA dataset contains 58,000 human-annotated QA pairs on 5,800 videos derived from the popular ActivityNet dataset. The dataset provides a benchmark for testing the performance of VideoQA models on long-term spatio-temporal reasoning.
Source: [ActivityNet-QA: A Dataset for Understanding Complex Web Videos via Question Answering](/paper/activitynet-qa-a-dataset-for-understanding)