Derives an interaction measure between crosscoder features from reconstruction error in compact proofs and applies it to produce computationally sparse crosscoders retaining 60% MLP performance with single-feature selection versus 10% for standard crosscoders.
Seshia, Dorsa Sadigh, and S
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
years
2026 3representative citing papers
A 12,000-line Rocq development proves that effect-level governance on AI workflows is semantically transparent, preserves expressivity, and separates decidable governance predicates from undecidable semantic properties.
Skilldex provides a TypeScript-based package manager with line-level format validation and skillset bundling to ensure coherent LLM agent skills.
citing papers explorer
-
Interactions Between Crosscoder Features: A Compact Proofs Perspective
Derives an interaction measure between crosscoder features from reconstruction error in compact proofs and applies it to produce computationally sparse crosscoders retaining 60% MLP performance with single-feature selection versus 10% for standard crosscoders.
-
Effect-Transparent Governance for AI Workflow Architectures: Semantic Preservation, Expressive Minimality, and Decidability Boundaries
A 12,000-line Rocq development proves that effect-level governance on AI workflows is semantically transparent, preserves expressivity, and separates decidable governance predicates from undecidable semantic properties.
-
Skilldex: A Package Manager and Registry for Agent Skill Packages with Hierarchical Scope-Based Distribution
Skilldex provides a TypeScript-based package manager with line-level format validation and skillset bundling to ensure coherent LLM agent skills.