#Jepa

New paper from Robbyant, with inference code and checkpoints already public. LingBot-Touch is a self-supervised tactile encoder trained JEPA-style: it predicts latent representations of masked contact instead of reconstructing raw sensor signals.…

Harold Hotelling’s 1936 Canonical Correlation Analysis (CCA) forms the theoretical and intuitive foundation for modern embedding prediction techniques, including JEPA models.

From Dev.to - machinelearning: Yann LeCun thinks the whole industry is building the wrong thing, and now he has $1B to prove it

In April 2026, Yann LeCun responded on LinkedIn to a charge that he had failed to credit Juergen...