Localization Does Not Imply Steerability: A Mechanistic Study of Sign Errors in LLM Mathematics
cs.LG, cs.AI, cs.CL
Submitted: 2026-11-05
Updated: 2026-11-05
Code: https://github.com/signcircuits-experiments/sign-circuit-scaling
Terminology
Sources
- LinAlg-Bench: A Forensic Benchmark Revealing Structural Failure Modes in LLM Mathematical Reasoning
- Eliciting Latent Predictions from Transformers with the Tuned Lens
- The Hydra Effect: Emergent Self-repair in Language Model Computations
- Steering Language Models With Activation Engineering
- Representation Engineering: A Top-Down Approach to AI Transparency
- Explorations of Self-Repair in Language Models
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis