How Controllable Are Large Language Models? A Unified Evaluation across Behavioral Granularities
cs.CL, cs.AI, cs.HC, cs.LG
Submitted: 2026-03-03
Updated: 2026-11-16
Code: https://github.com/zjunlp/EasyEdit
Terminology
Sources
- Foundational Challenges in Assuring Alignment and Safety of Large Language Models
- Persona Vectors: Monitoring and Controlling Character Traits in Language Models
- Representation Engineering for Large-Language Models: Survey and Research Challenges
- Steering Large Language Model Activations in Sparse Spaces
- Belief Dynamics Reveal the Dual Nature of In-Context Learning and Activation Steering
- STEER-BENCH: A Benchmark for Evaluating the Steerability of Large Language Models
- SEAL: Steerable Reasoning Calibration of Large Language Models for Free
- Length-Controlled AlpacaEval: A Simple Way to Debias Automatic Evaluators
- Scaling and evaluating sparse autoencoders
- SafeSwitch: Steering Unsafe LLM Behavior via Internal Activation Signals
- A Unified Understanding and Evaluation of Steering Methods
- Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2
- The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets
- The Information Geometry of Softmax: Probing and Steering
- Towards Reliable Evaluation of Behavior Steering Interventions in LLMs
- The Shape of Beliefs: Geometry, Dynamics, and Interventions along Representation Manifolds of Language Models' Posteriors
- Open Problems in Mechanistic Interpretability
- AlphaSteer: Learning Refusal Steering with Principled Null-Space Constraint
- HyperSteer: Activation Steering at Scale with Hypernetworks
- Towards Understanding Steering Strength
Related papers
- ALARM: Audio-Language Alignment for Reasoning Models
- ThinkSum: Probabilistic reasoning over sets using large language models
- Coherence boosting: When your pretrained language model is not paying enough attention
- From Facts to Insights: A Persona-Driven Dual Memory Framework and Dataset for Role-Playing Agents
- Improving General Role-Playing Agents via Psychology-Grounded Reasoning and Role-Aware Policy Optimization
- The Art of Asking: Multilingual Prompt Optimization for Synthetic Data