TY  - RPRT
TI  - On the Capability Separation Between World-Model Policy Learning and Imitated World-Action Models
AU  - Yang Yu
PY  - 2026
UR  - https://arxiv.org/abs/2608.22197
ID  - 2608.22197
ER  -