NEC Corporation Predicts Language Model Performance Across Multiple Languages
Summary
NEC Corporation has obtained USPTO Patent Application US20260099714A1 for a prediction device that forecasts language model performance when trained across multiple languages. The device acquires model size, training data amount, and target language ratio to predict model loss using a function combining model size, training data, and a power of the target language ratio. The patent covers 5 CPC classifications including G06N 3/084 and G06F 18/24147.
What changed
NEC Corporation has been granted USPTO Patent Application US20260099714A1 for a prediction device and method to estimate language model performance in multilingual training scenarios. The device includes an acquisition unit for model size, training data amount, and target language ratio, along with a prediction unit applying a mathematical function combining these parameters.
Technology companies developing AI language models should monitor this patent for potential licensing or design-around considerations. The patent's mathematical approach to predicting model loss based on language distribution may influence how multilingual AI systems are developed and optimized.
What to do next
- Monitor patent landscape for competing AI/ML prediction technologies
- Review patent claims for potential licensing implications
Archived snapshot
Apr 11, 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.
PREDICTION DEVICE, PREDICTION METHOD, AND NON-TRANSITORY COMPUTER READABLE MEDIUM STORING PREDICTION PROGRAM FOR SUPPORTING DECISION MAKING
Application US20260099714A1 Kind: A1 Apr 09, 2026
Assignee
NEC Corporation
Inventors
Kosuke AKIMOTO, Masafumi Oyamada
Abstract
A prediction device capable of predicting performance of a language model in a case where the language model is trained using a plurality of languages is implemented. The prediction device includes an acquisition unit for acquiring a model size of the language model for a target language to be trained using the plurality of languages, a training data amount used for learning processing, and a target language ratio indicating a ratio of a data amount of the target language in the training data amount, and a prediction unit for predicting a loss of the language model using a product of a function depending on the model size and the training data amount and a constant power of the target language ratio.
CPC Classifications
G06N 3/084 G06F 18/24147 G06F 18/27 G06N 3/0475
Filing Date
2025-09-24
Application No.
19338324
Named provisions
Related changes
Get daily alerts for USPTO Patent Applications - AI & Computing (G06N)
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 - AI & Computing (G06N) publishes new changes.
Subscribed!
Optional. Filters your digest to exactly the updates that matter to you.