arXiv AI August 5, 2026 By Abdallah Khemais

A Theory of Conditional Collapse under Low-Rank Weight-Space Ablations: I. The Single-Block Theory and Synthetic Validation

arXiv:2608. 03620v1 Announce Type: cross Abstract: Activation patching and weight-space ablation both claim a component is causally responsible for a behavior, yet they act on different objects: one forward pass versus the parameters behind every forward pass.

Summary generated by The Flow from the publisher's feed. The full article lives at arXiv AI.

arXiv AI

Aug 5

Cross-Layer Interaction under Weight-Space Ablation: A Closed-Form Attention Jacobian Bound and a Test on a Real Pretrained Model

arXiv:2608. 03629v1 Announce Type: new Abstract: A companion paper studies when activation patching and weight-space ablation agree, inside an idealized model where a conditional computation is carried additively through a residual stream.

By Abdallah Khemais

llms benchmarks

arXiv AI

Jul 1

Protein contacts are already in the attention: a single-forward-pass alternative to the Categorical Jacobian

arXiv:2606. 21876v2 Announce Type: replace-cross Abstract: The Categorical Jacobian of Zhang et al.

By Rome Thorstenson

llms

arXiv AI

Jul 21

Every Component is a Lookup: Token Attribution and Composition from a Single Decomposition

arXiv:2605. 23393v2 Announce Type: replace-cross Abstract: Mechanistic interpretability of transformers requires identifying not just which components matter but how they compose into the computational route that produced a prediction.

By Po-Kai Chen, Aske Plaat, Niki van Stein

llms safety

arXiv Machine Learning

Aug 12

Off-Axis, On Purpose: Where a Transformer Computes Concepts and Why it Does So

arXiv:2608. 10251v1 Announce Type: cross Abstract: A transformer's answer lives on one axis: the direction its unembedding reads.

By Mark Oskin

llms benchmarks

arXiv AI

Jul 22

The Information Shadow: Measuring Structural Limits on What Language Models Can Learn

arXiv:2607. 18305v1 Announce Type: cross Abstract: Some limits on what language models know are not gaps in data coverage but structural properties of learning from text.

By Priyansh Srivastava, Romit Chatterjee

llms benchmarks safety

arXiv AI

1d ago

Exact Attention Sensitivity and the Geometry of Transformer Stability

arXiv:2602. 18849v2 Announce Type: replace-cross Abstract: We develop a sensitivity analysis for transformer attention in a geometry aligned with tokenwise computation.

By Seyed Morteza Emadi

llms

Related stories

Cross-Layer Interaction under Weight-Space Ablation: A Closed-Form Attention Jacobian Bound and a Test on a Real Pretrained Model

Protein contacts are already in the attention: a single-forward-pass alternative to the Categorical Jacobian

Every Component is a Lookup: Token Attribution and Composition from a Single Decomposition

Off-Axis, On Purpose: Where a Transformer Computes Concepts and Why it Does So

The Information Shadow: Measuring Structural Limits on What Language Models Can Learn

Exact Attention Sensitivity and the Geometry of Transformer Stability