This is your work, valued
The first general-purpose action model of its kind made by @seonghyeonye and @joeljang
LAPA. [ICLR 2025] LAPA: Latent Action Pretraining from Videos