A fill-in-the-middle scheme to distill hidden reasoning from proprietary models

OwariDa · x · 2026-11-23

OwariDa proposes distilling reasoning from proprietary models that show truncated chains of thought: train a model to reconstruct the omitted reasoning from the prefix alone, rewarded by a continuous likelihood signal — how much probability it adds to P(known answer | prefix + reasoning). Since the model never sees the suffix, it can't leak the answer directly.

Related event: OwariDa Proposes FIM-style Distillation to Reveal Hidden Reasoning of Closed Models(3 posts)→

Original post →

More from Research

Research channel →