Close Menu
    Facebook LinkedIn
    Geo365
    Facebook LinkedIn
    BESTILL Login ABONNÉR PÅ NYHETSBREV
    • Hjem
    • Anlegg og infrastruktur
    • Aktuelt
    • Bergindustri
    • Dyphavsmineraler
    • Miljø
    • Olje og gass
    • Geofunn
    • Download Media Guide
    Geo365
    You are at:Home » Fine-tuning LLMs for geoscience
    Olje og gass

    Fine-tuning LLMs for geoscience

    Large language models promise new ways of working with data, but they lack domain expertise. At Dig X Subsurface 2025, Viridien will show how annotating data from 10 million wells unlocks real AI value for subsurface geoscience.
    By Redaksjonennovember 27, 2025
    Del denne artikkelen Facebook Twitter LinkedIn Email

    Annotation effort spanning millions of wells and over a million curated data points. Illustration: Viridien / Conference abstract

    Facebook Twitter LinkedIn Email

    ChatGPT, Claude, Grok, and other generic large language models (LLMs) excel at summarizing, drafting, and coding support – but their limits quickly show when applied to highly specialized technical domains like subsurface geoscience. Vast datasets, inconsistent formats, and complex terminology demand a tailored approach.

    To address this, Viridien launched one of the largest geoscience annotation initiatives to date.

    Starting with a collection of publicly available well data – over 65 million documents from 10 million worldwide – subject matter experts have carefully curated and annotated more than one million data points.

    Each annotation is guided by a detailed taxonomy of 30 document classes, 90 table classes, and 50 figure classes, ensuring broad coverage across document types and geoscience disciplines.

    The annotation process blends human expertise with AI feedback loops. By monitoring model errors during curation, the team continuously refines both the taxonomy and the dataset, capturing edge cases and improving class balance. This iterative approach not only ensures quality but also drives measurable gains in model performance.

    The payoff is clear: fine-tuned classifiers and embedding models trained on this annotated dataset consistently outperform generic LLMs in classification accuracy and document retrieval relevance. In other words, they move beyond hype and into practical, explainable AI tailored to subsurface workflows.

    At the Dig X Subsurface 2025 conference in December, Jamie Richardson (Viridien) will share how this evolving dataset is laying the groundwork for operational AI in geoscience — helping industry teams turn decades of underutilized data into discovery.

    Join us at Scandic Fornebu, December 03-04, 2025, to learn more. The program can be found on the conference website.

    Related Posts

    92 000 brønndokumenter klare for KI

    november 18, 2026

    Trenger vi flere selskaper på norsk sokkel?

    november 16, 2026

    Langt over forventning

    november 10, 2026
    Add A Comment

    Comments are closed.

    NYHETSBREV
    Abonner på vårt nyhetsbrev
    geo365.no: ledende leverandør av nyheter og kunnskap som vedrører geofaget og geofaglige problemstillinger relatert til norsk samfunnsliv og næringsliv.
    KONFERANSER

    Lekker hundrevis av liter olje hver dag
    Nov 20, 2026

    Lekker hundrevis av liter olje hver dag

    From workflow to agent
    Nov 20, 2026

    From workflow to agent

    Steinsprang i Longyearbyen registrert med fiberoptisk teknologi
    Nov 17, 2026

    Steinsprang i Longyearbyen registrert med fiberoptisk teknologi

    Skred i Kvam registrert i Sverige
    Nov 13, 2026

    Skred i Kvam registrert i Sverige

    Norske fjorder, sølv og marmor får internasjonal anerkjennelse
    Nov 10, 2026

    Norske fjorder, sølv og marmor får internasjonal anerkjennelse

    Is the oil industry cannibalising itself in geoscience recruitment?
    Nov 20, 2026

    Is the oil industry cannibalising itself in geoscience recruitment?

    A tale of a cave or how much in-place volume you might be missing
    Nov 19, 2026

    A tale of a cave or how much in-place volume you might be missing

    Polymetallic nodule discovery onshore Peru
    Nov 18, 2026

    Polymetallic nodule discovery onshore Peru

    A jump in the deep
    Nov 17, 2026

    A jump in the deep

    It boils down to connectivity
    Nov 16, 2026

    It boils down to connectivity

    OLJEPRIS
    BCOUSD quotes by TradingView
    GULLPRIS
    GOLD quotes by TradingView
    KOBBERPRIS
    Track all markets on TradingView
    GeoPublishing AS

    GeoPublishing AS
    Trollkleiva 23
    N-1389 Heggedal

    Publisher & General Manager

    Ingvild Ryggen Carstens
    ingvild@geopublishing.no
    cell: +47 974 69 090

    Editor in Chief

    Ronny Setså
    ronny@geopublishing.no
    +47 901 08 659

    Media Guide

    Download Media Guide

    ABONNEMENT
    NYHETSBREV
    Abonner på vårt nyhetsbrev
    © 2026 GeoPublishing AS - All rights reserved.

    Type above and press Enter to search. Press Esc to cancel.