Drilling US 20,260,049,549 A1 Under examination

Benchmarking and evaluation of llms for geoscience domain

This invention describes a method for creating specialized test questions and answers to evaluate how well large language models (LLMs) perform in the oil and gas industry. It involves feeding an LLM existing data and a specific task, then having that LLM generate new, realistic questions and their correct responses. These generated pairs then form a dataset to benchmark other LLMs tailored for geoscience.

Why it matters: Filed just a year ago, the rapid advancements in LLM capabilities for generating high-quality synthetic data make this approach more viable and efficient today. The increasing demand for specialized LLMs in niche domains like geoscience also highlights the growing need for robust benchmarking datasets.

Status: Under examinationA pending application. It has not been granted, and it has not been abandoned.
How hard to build: SpecializedHeavy software system, LLM integration, domain data handling

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