The Stop-Think-AutoRegress Language Diffusion Model (STAR-LDM) integrates latent diffusion planning with autoregressive generation. Unlike conventional autoregressive language models limited to token-by-token decisions, STAR-LDM incorporates a "thinking" phase that pauses generation to refine a semantic plan through diffusion before continuing. This enables global planning in continuous space prior to committing to discrete tokens. Evaluations show STAR-LDM significantly outperforms similar-sized models on language understanding benchmarks and achieves >70% win rates in LLM-as-judge comparisons for narrative coherence and commonsense reasoning. The architecture also allows straightforward control through lightweight classifiers, enabling fine-grained steering of attributes without model retraining while maintaining better fluency-control trade-offs than specialized approaches.
@article{arxiv.2602.20528,
title = {Stop-Think-AutoRegress: Language Modeling with Latent Diffusion Planning},
author = {Justin Lovelace and Christian Belardi and Sofian Zalouk and Adhitya Polavaram and Srivatsa Kundurthy and Kilian Q. Weinberger},
journal= {arXiv preprint arXiv:2602.20528},
year = {2026}
}