Supplementary Data for: Conformal Scaling in Trained Transformer Attention

Read full text Summarise with VIRAT
Authors: Ariel Umphrey, Eldon Umphrey
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Supplementary data and reproduction scripts for 'Conformal Scaling in Trained Transformer Attention: Evidence for an SYK Fixed Point' (Umphrey 2026). Contains: full per-head conformal dimension tables for GPT-2 (144 heads), Pythia training checkpoint data (70m, 160m, 410m), fine checkpoint sampling for prethermal plateau analysis, and all Python scripts for reproduction.

Fields of Study & Subjects

Topic Modeling Quantum many-body systems Adversarial Robustness in Machine Learning

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
25 March 2026
Language
EN