1 paper · 1 filter
Sanjoy Chowdhury, Dinesh Manocha
Language models often generate long chain-of-thought traces, but it remains unclear how much of this reasoning is necessary for preserving the final prediction. We study this throu…