Computing US 12,505,307 B2 Not in force

Natural language training and/or augmentation with large language models

This invention describes a method for teaching a smaller AI language model to generate text by having a larger, more powerful AI model act as its teacher. The larger AI analyzes a dataset for a specific task and generates example text outputs. The smaller AI then learns to imitate these examples through a process called imitation learning, allowing it to develop its own ability to produce similar high-quality text for that task.

Why it matters: When this was filed in 2023, the techniques for efficiently transferring knowledge from large language models to smaller ones were still rapidly evolving. By 2026, methods like imitation learning and model distillation have matured significantly, making it more practical and effective to train smaller, specialized AI models using the guidance of larger ones.

Status: Not in forceListed as no longer active. The specific reason is not in the record we hold.
How hard to build: SpecializedSoftware logic, LLMs, SLMs, training data

3 AI reinvention directions for this patent

See all three

AI gives you a few directions you could take this. Pick one, and we check whether your version is different enough to patent, then write the filing.

Reinvent this with AI

Where this one came from

Every morning we go through millions of abandoned patents and pick out 15 worth your time. This was one of today’s.

+10