TY  - RPRT
TI  - The Optimal Sample Complexity of Learning Autoregressive Chain-of-Thought
AU  - Zhiyuan Li
PY  - 2026
UR  - https://arxiv.org/abs/2607.07423
ID  - 2607.07423
ER  -