This inventor holds 1 USPTO granted patent. Top assignee: Google Inc.. Active years: 2026.
Data source: USPTO (United States Patent and Trademark Office) public patent records. Weekly synchronization. How IDiyas builds this profile
Company Filing History:
Years Active: 2026
Title: James Patrick Lee-Thorp: Inventor in Transformer Systems
Introduction
James Patrick Lee-Thorp is an inventor based in San Francisco, California, whose patent work concerns transformer systems for deep learning. His listed patent is associated with Google Inc.
Latest Patents
Lee-Thorp is credited with 1 patent, titled “Mixing Tokens with Spectral Transform.” The patent describes transformer systems and methods for performing deep learning tasks on input sequences to generate encoded outputs. One described encoder block uses a spectral transform mixing layer to process input-token embeddings along the sequence dimension, followed by a feed-forward layer that combines information based on the embeddings and spectral transform output for subsequent processing.
Career Highlights
The available information identifies Lee-Thorp’s work with Google Inc. and highlights his contribution to transformer architecture through a patent focused on spectral-transform token mixing.
Collaborations
His listed coworkers include Joshua Timothy Ainslie and Ilya Eckstein.
Conclusion
James Patrick Lee-Thorp’s patent reflects work on transformer systems that use spectral transforms to process token sequences for deep learning.
Data source: USPTO (United States Patent and Trademark Office) public patent records. Weekly synchronization. How IDiyas builds this profile