Incoherence in Goal-Conditioned Autoregressive Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Karwowski, Jacek, Douglas, Raymond
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910092169838592
author Karwowski, Jacek
Douglas, Raymond
author_facet Karwowski, Jacek
Douglas, Raymond
contents We investigate mathematically the notion of incoherence: a structural issue with reinforcement learning policies derived by naive goal-conditioning of autoregressive models. We focus on the process of re-training models on their own actions, that is, fine-tuning offline-learned policies with online RL. We prove that it decreases incoherence and leads to an improvement in return, and we aim to characterize the resulting trajectory of policies. By re-framing standard notions of control-as-inference and soft Q learning, we establish a three-way correspondence with two other ways of understanding the iterative re-training process: as folding the posterior into the reward and, in the deterministic case, as decreasing the temperature parameter; the correspondence has computational content via the training-inference trade-off. Through soft-conditioning generative models, we discuss the link between incoherence and the effective horizon.
format Preprint
id arxiv_https___arxiv_org_abs_2510_06545
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Incoherence in Goal-Conditioned Autoregressive Models
Karwowski, Jacek
Douglas, Raymond
Machine Learning
Artificial Intelligence
We investigate mathematically the notion of incoherence: a structural issue with reinforcement learning policies derived by naive goal-conditioning of autoregressive models. We focus on the process of re-training models on their own actions, that is, fine-tuning offline-learned policies with online RL. We prove that it decreases incoherence and leads to an improvement in return, and we aim to characterize the resulting trajectory of policies. By re-framing standard notions of control-as-inference and soft Q learning, we establish a three-way correspondence with two other ways of understanding the iterative re-training process: as folding the posterior into the reward and, in the deterministic case, as decreasing the temperature parameter; the correspondence has computational content via the training-inference trade-off. Through soft-conditioning generative models, we discuss the link between incoherence and the effective horizon.
title Incoherence in Goal-Conditioned Autoregressive Models
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2510.06545