TY  - RPRT
TI  - CounterAlign: Counterfactual Supervision for Vision-Language-Action Models
AU  - Haru Kondoh
AU  - Kei Ota
AU  - Asako Kanezaki
AU  - Yueh-Hua Wu
PY  - 2026
UR  - https://arxiv.org/abs/2608.21740
ID  - 2608.21740
ER  -