A balloon effect: ban one channel and a hidden trait finds another
cs.AI
Submitted: 2026-11-02
Updated: 2026-11-02
Code: https://github.com/kkt94/balloon-effect
Terminology
Sources
- Distilling the Knowledge in a Neural Network
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Towards Understanding Subliminal Learning: When and How Hidden Biases Transfer
- Subliminal Learning Is Steering Vector Distillation
- Learning Through Noise: Why Subliminal Learning Works and When It Fails
- Channel Location Constrains the Auditability of Subliminal Learning
- Covert Influence Between Language Models
- Subliminal Signals in Preference Labels
- Qwen3 Technical Report
- The Llama 3 Herd of Models
- Representation Engineering: A Top-Down Approach to AI Transparency
- Towards Best Practices of Activation Patching in Language Models: Metrics and Methods
- Locating and Editing Factual Associations in GPT
- Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small
- Retrieval Head Mechanistically Explains Long-Context Factuality
- Copy Suppression: Comprehensively Understanding an Attention Head
- Tokenization counts: the impact of tokenization on arithmetic in frontier LLMs
- Narrow Finetuning Leaves Clearly Readable Traces in Activation Differences
- Massive Activations in Large Language Models
- Persona Vectors: Monitoring and Controlling Character Traits in Language Models
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