Techniques for Layer Fusion Group Selection in Deep Learning Compilers
Summary
USPTO published patent application US20260099779A1 on April 9, 2026, covering techniques for layer fusion group selection in deep learning compilers. The invention by Nien-En Lee and Tsung-Han Lin relates to optimizing compiler operations by determining operation fusion groups based on data sizes and target memory storage capacity. The patent application was filed on October 9, 2024, under application number 18910130.
What changed
USPTO published patent application US20260099779A1 titled 'Techniques for Layer Fusion Group Selection in Deep Learning Compilers' on April 9, 2026. The application covers methods for compilers to optimize operation fusion groups based on data sizes of adjacent operations and target memory storage capacity, ensuring intermediate data remains within corresponding target memories.
Technology companies developing deep learning frameworks, AI compilers, or optimization tools may benefit from reviewing this patent application to understand the technical approaches covered. Patent applications do not create compliance obligations but provide intellectual property landscape awareness for technology developers and researchers working in compiler optimization and deep learning acceleration.
What to do next
- Monitor for updates
Archived snapshot
Apr 14, 2026GovPing captured this document from the original source. If the source has since changed or been removed, this is the text as it existed at that time.
TECHNIQUES FOR LAYER FUSION GROUP SELECTION IN DEEP LEARNING COMPILERS
Application US20260099779A1 Kind: A1 Apr 09, 2026
Inventors
Nien-En Lee, Tsung-Han Lin
Abstract
In an aspect of the disclosure, a method, a computer-readable medium, and an apparatus are provided. The apparatus may be a compiler. The compiler obtains data size of operations in a model graph corresponding to the workflow. The model graph includes multiple operations and an execution order of the operations. The compiler determines one or more operation fusion groups based on data sizes of adjacent operations and a storage capacity of one or more target memories, the intermediate data generated during processing by each operation fusion group is not transferred outside a corresponding target memory. The compiler updates the model graph based on the determined one or more operation fusion groups.
CPC Classifications
G06Q 10/0631 G06Q 10/0633
Filing Date
2024-10-09
Application No.
18910130
Related changes
Get daily alerts for USPTO Patent Applications - Business Methods (G06Q)
Daily digest delivered to your inbox.
Free. Unsubscribe anytime.
Source
About this page
Every important government, regulator, and court update from around the world. One place. Real-time. Free. Our mission
Source document text, dates, docket IDs, and authority are extracted directly from USPTO.
The summary, classification, recommended actions, deadlines, and penalty information are AI-generated from the original text and may contain errors. Always verify against the source document.
Classification
Who this affects
Taxonomy
Browse Categories
Get alerts for this source
We'll email you when USPTO Patent Applications - Business Methods (G06Q) publishes new changes.
Subscribed!
Optional. Filters your digest to exactly the updates that matter to you.