NRGPT: An Energy-based Alternative for GPT

ICLR 2026 Conference Submission13858 Authors

Keywords: energy-based model, GPT, LLM, small language models

TL;DR: we introduce an energy-based GPT model, where inference is gradient descent on the energy; we evaluate it on three datasets Listops, Shakespeare and OWT

Abstract:

Generative Pre-trained Transformer (GPT) architectures are the most popular design for language modeling. Energy-based modeling is a different paradigm that views inference as a dynamical process operating on an energy landscape. We propose a minimal modification of the GPT setting to unify it with the EBM framework. The inference step of our model, which we call eNeRgy-GPT (NRGPT), is conceptualized as an exploration of the tokens on the energy landscape. We prove, and verify empirically, that under certain circumstances this exploration becomes gradient descent, although they don’t necessarily lead to the best performing models. We demonstrate that our model performs well for simple language (Shakespeare dataset), algebraic ListOPS tasks, and richer settings such as OpenWebText
language modeling. We also observe that our models may be more resistant to overfitting, doing so only during very long training.

Primary Area: generative models

Submission Number: 13858