Hesitation Has a Geometry: Entropy-Trained Hyperbolic Probes for Sparse Activation Steering
cs.LG, cs.CL
Submitted: 2026-10-01
Updated: 2026-10-01
Code: https://github.com/huggingface/math-verify
Terminology
Sources
- Tethered Reasoning: Decoupling Entropy from Hallucination in Quantized LLMs via Manifold Steering
- Training Verifiers to Solve Math Word Problems
- The Llama 3 Herd of Models
- Measuring Mathematical Problem Solving With the MATH Dataset
- Contextual Linear Activation Steering of Language Models
- From Noise to Diversity: Random Embedding Injection in LLM Reasoning
- Seek in the Dark: Reasoning via Test-Time Instance-Level Policy Gradient in Latent Space
- Fractional Reasoning via Latent Steering Vectors Improves Inference Time Compute
- HyperGuide: Hyperbolic Guidance for Efficient Multi-Step Reasoning in Large Language Models
- Curveball Steering: The Right Direction To Steer Isn't Always Linear
- HyperSteer: Activation Steering at Scale with Hypernetworks
- Steering Language Models With Activation Engineering
- Qwen2.5-Math Technical Report: Toward Mathematical Expert Model via Self-Improvement
- Representation Engineering: A Top-Down Approach to AI Transparency
- Are language models aware of the road not taken? Token-level uncertainty and hidden state dynamics
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks