Introspecting Alignment Shifts Beyond Behaviors Implanted Through Fine-Tuning
cs.LG
Submitted: 2026-08-05
Updated: 2026-09-26
Terminology
Sources
- Layer Normalization
- The Llama 3 Herd of Models
- DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
- How to use and interpret activation patching
- Delta-Crosscoder: Robust Crosscoder Model Diffing in Narrow Fine-Tuning Regimes
- Emergent Introspective Awareness in Large Language Models
- Can LLMs Introspect? A Reality Check
- Gemma 3 Technical Report
- HuggingFace's Transformers: State-of-the-art Natural Language Processing
- Qwen3 Technical Report
- Universal and Transferable Adversarial Attacks on Aligned Language Models
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: Local-Consistency Optimisation for Bayesian Neural Networks
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks