x
Reasoning Models Sometimes Output Illegible Chains of Thought — AI Alignment Forum