llm-probing

Tag

Cards List
#llm-probing

Hesitation Has a Geometry: Entropy-Trained Hyperbolic Probes for Sparse Activation Steering

arXiv cs.LG ↗ · 2026-10-05 Cached

The paper proposes HEST, a method that embeds LLM hidden states into the Poincaré ball with an entropy-trained probe and steers only high-entropy (hesitant) tokens along hyperbolic geodesics, improving math reasoning accuracy on Qwen2.5-Math and Llama-3.1 models where fixed-vector Euclidean steering fails.

0 favorites 0 likes
#llm-probing

Does the Truthfulness Signal Survive Code-Mixing? Probing Hidden States for Hallucination Detection in Hinglish

arXiv cs.CL ↗ · 2026-09-22 Cached

The study investigates whether hallucination detection probes trained on monolingual text transfer to Hinglish, finding robust performance and higher hallucination rates in Hindi and Hinglish compared to English.

0 favorites 0 likes
#llm-probing

Safety Beyond the Interface: Detecting Harm via Latent States in Large Language Models

arXiv cs.AI ↗ · 2026-09-18 Cached

This paper proposes using lightweight MLP probes on activations from LLaMA-3.1-8B to detect harmful prompts with high F1 scores, offering a cost-effective alternative to large external guardrail models.

0 favorites 0 likes
← Back to home

Submit Feedback