Skip to content

Token Classification: Dispatches1

This project develops a text/string detection & classification model by fine-tuning large language model architectures. The model detects and classifies text/strings vis-à-vis a set of classes. The model depends on Reuters International News Bulletins of a specific time period, therefore the data's timeliness and context will impact its performance in various settings.

The modelling package, and the supplementary packages, are adaptable components.


In brief

 CommentExpires / Decommissioning
Intelligence Hub Hosts model & data details, and [automatically updated] dynamic model card components. End of November 2025; further developments, updates, cease a brief time hereafter. 1
basic model interface A simple interface for interacting with the developed model; an endless number of interface designs are possible. The classes in focus are organisation, person, time, geographic entity, geo-political entity [excluding geographic items]
Repositories Hub Hosts the project's git repositories.
  1. Further developments by external/interested parties will appear elsewhere.












  1. Herein, token classification is the detection and classification of text or strings via fine-tuned large language model architectures. Each model detects and classifies text or strings of vis-à-vis a set of classes.