SPIN Unprocessed October 23, 2026 ai_technology research
Probing and Steering for Safety-Relevant Behaviours in Large Language Models
View original on arxiv.orgOverview
arXiv:2610.30363v1 Announce Type: new Abstract: Activation probes are computed directions in a language model's residual stream, used both to detect safety-relevant behaviours and, by adding their direction back into the stream, to steer them. We explore how the main design choices behind such probes interact, within one pipeline, on one model (Llama-3.1-8B-Instruct), and across three behaviours: deception, high-stakes awareness and sycophancy. The pipeline crosses two estimators (difference-of-
SpinGraph analysis pending — check back after processing.
Ask AI about this story
Opens with the SpinGraph .md URL and structured context — one click, prompt included.
More from arXiv Machine Learning
View all →- AcoustiPrint: Physics-Guided Neural Design of Manufacturable Acoustic Structures
- Escaping the Phylogenetic Trap: Persistent Homology Enables Cross-Species Ion Channel Classification via Multimodal Graph Neural Networks
- Bayesian Fixed-Confidence Best-Arm Identification with Abstention
- CTRL-FM: Inference-Time Structural Control for Flow-Based Virtual Sensing under Long-Horizon Node Outages
- Spike-Aware Gradient-Based Mixed-Precision Quantization of Spiking Neural Networks for Vision Applications
- GAV: A scientific experimentation environment for PDE agents
Markdown (.md) · JSON-LD schema (.json) · Machine-readable for AI & GEO