Understanding Unreliability of Steering Vectors in Language Models: Geometric Predictors and the Limits of Linear Approximations
Understanding Unreliability of Steering Vectors in Language Models: Geometric Predictors and the Limits of Linear Approximations explores Develop a tool to diagnose and improve the reliability of steering vectors in language models, addressing non-linear behavior representations.. Commercial viability score: 4/10 in AI System Diagnostics.
BUILDER'S SANDBOX
Build This Paper
Use an AI coding agent to implement this research.
OpenAI CodexAI Agent
Lightweight coding agent in your terminal.
Claude CodeAI Agent
Agentic coding tool for terminal workflows.
AntiGravity IDEScaffolding
AI agent mindset installer and workflow scaffolder.
CursorIDE
AI-first code editor built on VS Code.
VS CodeIDE
Free, open-source editor by Microsoft.
Recommended Stack
Startup Essentials
Estimated $9K - $13K over 6-10 weeks.
See exactly what it costs to build this -- with 3 comparable funded startups.
7-day free trial. Cancel anytime.
Discover the researchers behind this paper and find similar experts.
7-day free trial. Cancel anytime.
Related Papers
- Builds On ThisBehavioral Steering in a 35B MoE Language Model via SAE-Decoded Probe Vectors: One Agency Axis, Not Five TraitsViability: 2.0
- Builds On ThisCurveball Steering: The Right Direction To Steer Isn't Always LinearViability: 2.0
- Prior WorkSteering Code LLMs with Activation Directions for Language and Library ControlViability: 4.0
- Higher ViabilityReliable Control-Point Selection for Steering Reasoning in Large Language ModelsViability: 7.0
- Higher ViabilityAnalysing the Safety Pitfalls of Steering Vectors