TY  - RPRT
TI  - Measuring in-context algorithmic reasoning in language models against an exact Bayes-optimal reference
AU  - Luan Ozelim
AU  - Hector Zenil
PY  - 2026
UR  - https://arxiv.org/abs/2608.01575
ID  - 2608.01575
ER  -