Valley: A Video Assistant with Large Language Model Enhanced Ability
A large multi-modal instruction-following dataset for video understanding, comprising 37k conversation pairs, 26k complex reasoning QA pairs and 10k detail description instruction pairs.
BibTex:
