Locally tuned Large Language Model (LLM) to empower digitalization of borehole logs for 3D stratigraphic modelling

Jun-Cheng Yao

doi:10.1016/j.tust.2026.108013

Locally tuned Large Language Model (LLM) to empower digitalization of borehole logs for 3D stratigraphic modelling

2026-08-11

0

PRE

AI

J

Jun-Cheng Yao

B

Borui Lyu

王宇 (Yu Wang) *

DOI：10.1016/j.tust.2026.108013

Original

Original request for help

Save

Abstract

En 中文

• A novel LLM-empowered method is proposed to automatically digitalize borehole logs for 3D stratigraphic modelling. • A locally deployed LLM, called StratumGPT, is developed from DeepSeek and fine-tuned using a regional dataset. • Performance of StratumGPT on borehole log digitalization and stratigraphic classification is compared with a general model. • The proposed method is illustrated using real-world database and locally deployed StratumGPT, with data confidentiality.

Keywords:

Geological text classification

Borehole logs

Artificial intelligence

Digitalization

Data confidentiality

Fine-tuning optimization

Locally tuned Large Language Model (LLM) to empower digitalization of borehole logs for 3D stratigraphic modelling

Abstract

Journal

Organization

Cited Papers

Citing Papers