SAGE Uses Topological Guidance to Fix Long-Horizon LLM Reasoning Biases
Xinyue Zeng · hf · 2026-09-28
A new HF paper, "SAGE: Mitigating Long-Horizon Reasoning Biases via Topological Guidance," attributes brittle long-horizon reasoning under sparse rewards to two biases: exploration bias (models drawn to locally plausible but structurally unstable branches) and compounding bias (small local deviations accumulate across depth).
It introduces Symbolic Closure Analysis (SCA) as both a theoretical lens and design principle, and builds SAGE combining algebraic sparsification (projecting candidates onto operator-indexed algebraic subspaces) with hyperbolic structural guidance (embedding reasoning states in negatively curved space for dense depth-wise signals).
Across 12 benchmarks and 7 model families SAGE beats competitive baselines, with up to an 8-fold improvement on the open Andrews-Curtis problem. Code is open-sourced.
More from Research
- New blog surveys world models: definitions, SOTA, and eval axes — mervenoyann · 2026-09-29
- Researcher Flags Risk That Pedagogical RL's Student-Likelihood Optimization Filters Rare Reasoning — novasarc01 · 2026-09-29
- Goodfire grants geometric_intel lab funding for AI interpretability research — ninamiolane · 2026-09-29
- Bespoke Labs Launches AutoResearchExam Benchmark, Again, With a Demo Video — gregd_nlp · 2026-09-28
- Agentick benchmark accepted at NeurIPS: LLM vs RL agents on same tasks, no single winner — pcastr · 2026-09-28
- IROS 2026 has 1,933 papers — researcher curates 130-paper reading list on VLA and robot learning — GlenBerseth · 2026-09-28