Is this your dataset? Claim it with the Hugging Face account that owns it.

TIGER-Lab/
MMEB-V2

Name: MMEB-V2
Creator: TIGER-Lab
License: https://spdx.org/licenses/Apache-2.0.html

Views

No views yet

MMEB-V2 (Massive Multimodal Embedding Benchmark)

Introduction

Building upon on our original MMEB, MMEB-V2 expands the evaluation scope to include five new tasks: four video-based tasks — Video Retrieval, Moment Retrieval, Video Classification, and Video Question Answering — and one task focused on visual documents, Visual Document Retrieval. This comprehensive suite enables… See the full description on the dataset page: https://huggingface.co/datasets/TIGER-Lab/MMEB-V2.