About Xent.Tech
We are a team of researchers and engineers interested in using LLMs to go beyond the current chatbot paradigm, in particular to reveal interesting patterns in data.
Our History and Future
Summer 2024
Clément starts thinking more seriously about the idea of generating synthetic training data for LLMs using the knowledge that they already have.
Fall 2024
Clément and Andrew begin initial research and development around synthetic data generation.
Winter 2024
Clément and Andrew develop the XentGame around a computationally intractable but conceptually interesting synthetic data generation possibility with the goal of understanding better what good or optimal solutions for these sorts of problems even look like.
Spring 2025
Clément and Andrew develop the XeGa language, a programming language for xent-based games. The XeGa language is designed to cover the space of all useful games that LLMs can play to benchmark and improve their performance.
June 2025
Our first paper is released!
Summer 2025
Development of the XeGa benchmark and language continues. A collection of XeGa game definitions along with a system for the evaluation of LLMs on these games is created with an planned open source release soon.
Who We Are
Clément Hongler
Clément is a professor of mathematics with a special interest in the theory and practice of machine learning.
Andrew Emil
Andrew is a software engineer with too much startup experience and an interest in LLMs (and Pickleball)