Routing Subspaces: Auditing Evaluation-to-Deployment Mismatch in Fine-Tuned Language Models
cs.CL, cs.AI, cs.SE
Submitted: 2026-05-11
Updated: 2026-11-01
Terminology
Sources
- Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training
- Stress-Testing Capability Elicitation With Password-Locked Models
- Alignment faking in large language models
- Inference-Time Intervention: Eliciting Truthful Answers from a Language Model
- Representation Engineering: A Top-Down Approach to AI Transparency
- Steering Language Models With Activation Engineering
- Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small
- Towards Automated Circuit Discovery for Mechanistic Interpretability
- Localizing Model Behavior with Path Patching
- Taken out of context: On measuring situational awareness in LLMs
- Me, Myself, and AI: The Situational Awareness Dataset (SAD) for LLMs
- Large Language Models Often Know When They Are Being Evaluated
- Probing and Steering Evaluation Awareness of Language Models
- Steering Evaluation-Aware Language Models to Act Like They Are Deployed
- In-Context Environments Induce Evaluation-Awareness in Language Models
- Measuring Massive Multitask Language Understanding
- Gemma 2: Improving Open Language Models at a Practical Size
- Qwen2.5 Technical Report
- The Llama 3 Herd of Models
- Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone
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