TY  - RPRT
TI  - Learning Evolving Latent Strategies for Multi-Agent Language Systems without Model Fine-Tuning
AU  - Wenlong Tang
PY  - 2026
UR  - https://arxiv.org/abs/2512.20629
ID  - 2512.20629
ER  -