The Unequal Influence of Bad Advice: Using Training Data Attribution to Modulate Emergent Misalignment
cs.CL, cs.AI
Submitted: 2026-09-29
Updated: 2026-09-29
Code: https://github.com/huggingface/peft
Terminology
Sources
- Emergent and Subliminal Misalignment Through the Lens of Data-Mediated Transfer
- Weird Generalization and Inductive Backdoors: New Ways to Corrupt LLMs
- Steering Out-of-Distribution Generalization with Concept Ablation Fine-Tuning
- In-Training Defenses against Emergent Misalignment in Language Models
- A Rank Stabilization Scaling Factor for Fine-Tuning with LoRA
- Concept Influence: Leveraging Interpretability to Improve Performance and Efficiency in Training Data Attribution
- Natural Emergent Misalignment from Reward Hacking in Production RL
- Bergson: An Open Source Library for Data Attribution
- Qwen2.5 Technical Report
- Qwen3 Technical Report
- Shaping capabilities with token-level data filtering
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- Convergent Linear Representations of Emergent Misalignment
- Olmo 3
- Model Organisms for Emergent Misalignment
- Inoculation Prompting: Instructing LLMs to misbehave at train-time improves test-time alignment
- Probe-Based Data Attribution: Discovering and Mitigating Undesirable Behaviors in LLM Post-Training
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