Is Chain-of-Thought Reasoning of LLMs a Mirage? ... Our results reveal that CoT reasoning is a brittle mirage that vanishes when it is pushed beyond training distributions. This work offers a deeper understanding of why and when CoT reasoning fails, emphasizing the ongoing challenge of achieving genuine and generalizable reasoning. ... Our findings reveal that CoT reasoning works effectively when applied to in-distribution or near in-distribution data but becomes fragile and prone to failure even under moderate distribution shifts. In some cases, LLMs generate fluent yet logically inconsistent reasoning steps. The results suggest that what appears to be structured reasoning can be a mirage, emerging from memorized or interpolated patterns in the training data rather than logical inference. ... Together, these findings suggest that LLMs are not principled reasoners but rather sophisticated simulators of reasoning-like text.

Stellan Skarsgård was scared to do his monologue in #Andor S1 because it was so good "I always have problems remembering lines ... especially now since I've had a stroke ... so I have [the lines] in my ear by a prompter" "I didn't want to fuck it up"

A new study from Denmark finds that experiencing prenatal stress may lead children to learn language faster. | Latest Mind Matters column by Susan Pinker. wsj.com/articles/prenatal-st… via @WSJ

In a new paper, our team uses unsupervised program synthesis to make sense of sensory sequences. This system is able to solve intelligence test problems zero-shot, without prior training on similar tasks: dpmd.ai/sensory

This was one of the most extraordinary moments of the entire war. A Polish resistance commander got himself sent to Auschwitz so as to be able to tell the world what was going on. The world didn’t want to believe it.