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.
More from Research
- Mathematician praises Lean community's soundness work, admits skepticism is easier than building — mboehme_ · 2026-11-23
- New proposal: mandatory compute disclosures plus 15% capabilities cap could delay intelligence explosion a year — sjgadler · 2026-11-23
- Prune, train, upcycle: one model that scales with your RAM — minhnguyen · 2026-11-23
- Claude formalizes GRH proof approach in 17M lines of Lean, code checks out — elonmusk · 2026-11-23
- Riemann proof built with Claude spans 17M lines of Lean — no human will ever read it — r0ck3t23 · 2026-11-23
- CLM author on extending editable context to video: RL may teach models where to look, prefill cut 30% — RulinShao · 2026-11-23