Preprint
Jul 2026
SIEVE: Structure-Aware Data Selection for Imitation Learning with VLA Models
SIEVE, a structure-aware data selection method for VLA imitation learning that can surpass full-data training while using only 50% of demonstrations and 50% of training steps, suggests that reusable structure, captured through primitives and transitions, is an important signal for efficient VLA imitation learning.
Changti Wu, Bin Yu, Zhaolong Shen et al.
· 1 citation