Exploring Next
Exploring Next — Ep 243 w/ Justy & Cody — Reasoning as Compression: Unifying Budget Forcing via the Conditional Information Bottleneck
Chain-of-Thought prompting makes LLMs more accurate but expensive. This research reframes efficient reasoning as a compression problem, introducing a conditional information bottleneck approach that preserves essential reasoning while cutting cognitive bloat. Instead of naive length penalties, they use semantic priors based on token surprisal to compress reasoning traces intelligently.