Measuring, Localizing, and Ablating Alignment Signatures in LLMs

arXiv:2605.30526 · cs.LG, cs.CL · Submitted 2026-05-28 · Read on arXiv

cs.LG, cs.CL

Submitted: 2026-05-28

Updated: 2026-10-20

Terminology

• Ablation — 2× in this paper · explained in 1 episode
• Post-training Alignment Signature Targeted Ablation — 1× in this paper

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: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks