Inherited or Learned? Backdoor Selectivity in Distilled Language Models
cs.AI
Submitted: 2026-10-31
Updated: 2026-10-31
Terminology
Sources
- Transferring Backdoors between Large Language Models by Knowledge Distillation
- Subliminal Learning: Language models transmit behavioral traits via hidden signals in data
- Pay Attention to the Triggers: Constructing Backdoors That Survive Distillation
- Gemma 3 Technical Report
- BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chain
- Distilling the Knowledge in a Neural Network
- LoRA: Low-Rank Adaptation of Large Language Models
- Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training
- Qwen2.5 Technical Report
- A Study of Backdoors in Instruction Fine-tuned Language Models
- Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents