6 citations · 6 across the 4 of their papers we have counts for
1 paper · 1 filter
Bonnie Li
World models trained via pixel reconstruction can struggle in visually complex environments, where irrelevant information dominates the objective and distract the model from inform…