Entity · paper

Distill to Detect: Exposing Stealth Biases in LLMs through Cartridge Distillation

paperactiveprovisionaldistill-to-detect-exposing-stealth-biases-in-llms-through-cartridge-distillation-144e7c86·1 events·first seen Jul 2, 2026

Aliases: Distill to Detect: Exposing Stealth Biases in LLMs through Cartridge Distillation

Co-occurring entities

Distill to Detect

More like this (12)

Distill to Detect Weak-to-Strong Distillation Generalized Distillation ensemble distillation Model Distillation Learning from the Self-future: On-policy Self-distillation for dLLMs Self-Distillation DeepSeek-R1-Distill-Llama-8B Rank-to-Distill Be My Tutor: On-Policy Co-Distillation for Mutual LLM Improvement via Peer Feedback distilabel Backdoor Unlearning Generalization: A Path Toward the Removal of Unknown Triggers in LLMs

Recent events (1)

6arXiv · cs.LG·Jul 2, 2026·source ↗

Distill to Detect (D2D): surfacing stealth preferential biases in LLMs via cartridge distillation

Researchers introduce Distill to Detect (D2D), a method for exposing hidden preferential biases in language models by distilling the distributional shift between a suspected model and its base into a KV-cache prefix adapter (cartridge), amplifying the bias signal into generated text. The work addresses a fundamental asymmetry in bias detection: stealth biases introduced anywhere in a model's supply chain can remain invisible to text-based inspection, weight analysis, or representation probing. D2D is grounded in a theoretical framework based on Fisher-weighted projection of logit distribution shifts and is validated across multiple bias types. The technique offers a practical auditing tool for detecting hidden behaviors in deployed LLMs.

Evaluation and Benchmarking AI Safety Research Distill to Detect: Exposing Stealth Biases in LLMs through Cartridge Distillation Distill to Detect