LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristics
arxiv.org
2 threads
A breath of fresh air from the over-concentration on LLM research. Something different for once.
Seems like this signals Yann Lecun's direction now that he's leaving Meta
The EMA teacher model still seems like black magic to me (present in DINO and JEPA series but gone now)