MENTIS: What Belief Changes Under Alignment? Measuring Multi-Scale Latent Torsion in Language Models
Preference alignment has substantially improved the observable behavior of large language models, yet it remains unclear what alignment changes internally. Aligned systems still fail under jailbreaks, prompt injection, and retrieval-time corruption, suggesting behavior-level evaluation alone is incomplete. Post-training should leave measurable traces in internal computation. We ask: when an instruction-tuned (IT) model becomes a preference-aligned (PA) model, what geometric structure changes, where do those changes concentrate, and how selectively do they vary across concepts, prompts, and model families? We introduce MENTIS, a geometry-first framework for measuring alignment-induced inte...
| Authors | Partha Pratim Saha, Samarth Raina, Mayur Parvatikar et al. |
|---|---|
| Primary category | cs.CL |
| Published | 31 May 2026 |
| Last updated at source | 7 June 2026 |
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.