MaxInfo: A Training-Free Key-Frame Selection Method Using Maximum Volume for Enhanced Video Understanding Paper • 2502.03183 • Published Feb 5, 2025 • 5
SA-DVAE: Improving Zero-Shot Skeleton-Based Action Recognition by Disentangled Variational Autoencoders Paper • 2407.13460 • Published Jul 18, 2024
Running Featured 1.35k FineWeb: decanting the web for the finest text data at scale 🍷 1.35k Explore and download the FineWeb web‑scale text dataset