TY  - RPRT
TI  - Calibrated Partial Resets: Preventing Policy Collapse in Continual Reinforcement Learning
AU  - Luc McCutcheon
AU  - Evangelos Chatzaroulas
AU  - Saber Fallah
PY  - 2026
UR  - https://arxiv.org/abs/2607.24996
ID  - 2607.24996
ER  -