Causal and Interpretable Structures in LLM Compositional Tasks
cs.CL, cs.AI, cs.LG
Submitted: 2026-09-28
Updated: 2026-09-28
Comments: 46 pages, 28 figures
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Arithmetic in the Wild: Llama uses Base-10 Addition to Reason About Cyclic Concepts
- Towards Empirical Interpretation of Internal Circuits and Properties in Grokked Transformers on Modular Polynomials
- Finding Alignments Between Interpretable Causal Variables and Distributed Neural Representations
- Gemma 2: Improving Open Language Models at a Practical Size
- Language Models Use Trigonometry to Do Addition
- The Llama 3 Herd of Models
- The Origins of Representation Manifolds in Large Language Models
- Qwen3 Technical Report
- The Shape of Beliefs: Geometry, Dynamics, and Interventions along Representation Manifolds of Language Models' Posteriors
Related papers
- Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
- Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
- Subliminal Steering: Stronger Encoding of Hidden Signals
- MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
- The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
- Untangling the Mechanisms of Misleading Context in Medical Question Answering