@randall_balestr
SIGReg for pretraining Video Foundation Models! Our LeVJEPA opens many doors... - stable recipe with a simple loss (sigreg + prediction) - no tubelet, frame aggregation, EMA, stop-gradient, .... - 20X more FLOP efficient than VJEPA1/2 pretraining - open source + reproducible https://t.co/rCwJRx2Dim