LV-MAE: Learning Long Video Representations through Masked-Embedding Autoencoders Paper • 2504.03501 • Published Apr 4, 2025 • 2