Datasetpublicgated

MLVU/MVLU

MLVU: Multi-task Long Video Understanding Benchmark This repo contains the annotation data and evaluation code for the paper "MLVU: A Comprehensive Benchmark for Multi-Task Long Video Understanding". 🔔 News: 🆕 7/28/2024: The data for the MLVU-Test set has been released (🤗 Link)! The test set includes 11 different tasks, featuring our newly added Sports Question Answering (SQA, single-detail LVU) and Tutorial… See the full description on the dataset page: https://huggingface.co/datasets/MLVU/MVLU.

sourceHugging Facecc-by-nc-sa-4.0updated 2y agoView on Hugging Face

40likes16kdownloads

rootMLVUjson1_plotQA.json

1_plotQA.jsonDownload Raw Back to json

This repository is gated, so its file contents are only served once you have accepted the publisher's terms at Hugging Face. Open it at the source above.