Sampling Reveals Style: Unsupervised, Training-Free Discovery of Prompt-Conditional Stylistic Axes in LLM Activations

arXiv:2609.19150 · cs.CL, cs.LG · Submitted 2026-07-20 · Read on arXiv

cs.CL, cs.LG

Submitted: 2026-07-20

Updated: 2026-09-18

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Steering — 24× in this paper · explained in 2 episodes
• Alignment — 17× in this paper · explained in 1 episode
• Entanglement — 11× in this paper · explained in 4 episodes
• Interpretability — 6× in this paper · explained in 15 episodes
• Activation Steering — 5× in this paper · explained in 4 episodes
• Annotator Agreement — 5× in this paper · explained in 1 episode
• Inter-Annotator Agreement — 5× in this paper
• Accuracy — 4× in this paper · explained in 1 episode
• Large Language Models (LLMs) — 4× in this paper · explained in 6 episodes
• Principal Component Analysis — 4× in this paper
• Quantization — 4× in this paper · explained in 12 episodes
• Representation Engineering — 4× in this paper · explained in 2 episodes
• Grounding — 3× in this paper · explained in 3 episodes
• LLM judge — 3× in this paper · explained in 2 episodes
• Principal Component Analysis (PCA) — 3× in this paper · explained in 4 episodes
• Core — 2× in this paper · explained in 2 episodes
• Robustness — 2× in this paper · explained in 5 episodes
• Semantic Alignment — 2× in this paper · explained in 3 episodes
• Activation Addition — 1× in this paper · explained in 1 episode
• Adam — 1× in this paper · explained in 2 episodes
• Cosine Similarity — 1× in this paper · explained in 3 episodes
• Lexical Diversity — 1× in this paper · explained in 2 episodes
• Resource Constraints — 1× in this paper · explained in 3 episodes
• Trace — 1× in this paper · explained in 2 episodes
• Transformer — 1× in this paper · explained in 2 episodes

Sources

• Sparse Autoencoders Find Highly Interpretable Features in Language Models
• I Have Covered All the Bases Here: Interpreting Reasoning Features in Large Language Models via Sparse Autoencoders
• The Llama 3 Herd of Models
• Qwen2.5-Coder Technical Report
• Steering LLMs? Actually, Sparse Autoencoders can outperform simple baselines
• Are Sparse Autoencoders Useful? A Case Study in Sparse Probing
• Steerable Chatbots: Exploring Personalization Control Interfaces via LLM Activation Steering
• Quantifying Feature Space Universality Across Large Language Models via Sparse Autoencoders
• DeepSeek-V3 Technical Report
• Qwen3.5-Omni Technical Report
• Mechanistically Eliciting Latent Behaviors in Language Models
• The Linear Representation Hypothesis and the Geometry of Large Language Models
• Automatically Interpreting Millions of Features in Large Language Models
• Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet
• Steering Language Models With Activation Engineering
• Taxonomy, Opportunities, and Challenges of Representation Engineering for Large Language Models
• AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoders
• Representation Engineering: A Top-Down Approach to AI Transparency
• MPNet: Masked and Permuted Pre-training for Language Understanding

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering