Self-Augmenting Retrieval for Diffusion Language Models

Paul Jünger, Justin Lovelace, Linxi Zhao, Dongyoung Go, Kilian Q Weinberger
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:55110-55124, 2026.

Abstract

Diffusion language models generate text by iteratively denoising an entire response in parallel, holding a noisy draft of the answer at every step. We show that these drafts surface salient entities early in the denoising trajectory, making them a useful lookahead signal for retrieval-augmented generation. We further show that grounding generation in retrieved evidence sharply reduces inter-token dependence, making RAG particularly well suited to parallel decoding. We exploit both through Self-Augmenting Retrieval for Diffusion Language Models (SARDI), a dynamic RAG framework that uses these intermediate drafts to guide retrieval during denoising. SARDI is training-free, retriever-agnostic, and applicable to any reasoning-capable discrete diffusion language model. Across five multi-hop QA benchmarks, SARDI outperforms training-free diffusion and autoregressive retrieval baselines at up to $5\times$ higher throughput.

Cite this Paper


BibTeX
@InProceedings{pmlr-v306-junger26a, title = {Self-Augmenting Retrieval for Diffusion Language Models}, author = {J\"{u}nger, Paul and Lovelace, Justin and Zhao, Linxi and Go, Dongyoung and Weinberger, Kilian Q}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {55110--55124}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/junger26a/junger26a.pdf}, url = {https://proceedings.mlr.press/v306/junger26a.html}, abstract = {Diffusion language models generate text by iteratively denoising an entire response in parallel, holding a noisy draft of the answer at every step. We show that these drafts surface salient entities early in the denoising trajectory, making them a useful lookahead signal for retrieval-augmented generation. We further show that grounding generation in retrieved evidence sharply reduces inter-token dependence, making RAG particularly well suited to parallel decoding. We exploit both through Self-Augmenting Retrieval for Diffusion Language Models (SARDI), a dynamic RAG framework that uses these intermediate drafts to guide retrieval during denoising. SARDI is training-free, retriever-agnostic, and applicable to any reasoning-capable discrete diffusion language model. Across five multi-hop QA benchmarks, SARDI outperforms training-free diffusion and autoregressive retrieval baselines at up to $5\times$ higher throughput.} }
Endnote
%0 Conference Paper %T Self-Augmenting Retrieval for Diffusion Language Models %A Paul Jünger %A Justin Lovelace %A Linxi Zhao %A Dongyoung Go %A Kilian Q Weinberger %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-junger26a %I PMLR %P 55110--55124 %U https://proceedings.mlr.press/v306/junger26a.html %V 306 %X Diffusion language models generate text by iteratively denoising an entire response in parallel, holding a noisy draft of the answer at every step. We show that these drafts surface salient entities early in the denoising trajectory, making them a useful lookahead signal for retrieval-augmented generation. We further show that grounding generation in retrieved evidence sharply reduces inter-token dependence, making RAG particularly well suited to parallel decoding. We exploit both through Self-Augmenting Retrieval for Diffusion Language Models (SARDI), a dynamic RAG framework that uses these intermediate drafts to guide retrieval during denoising. SARDI is training-free, retriever-agnostic, and applicable to any reasoning-capable discrete diffusion language model. Across five multi-hop QA benchmarks, SARDI outperforms training-free diffusion and autoregressive retrieval baselines at up to $5\times$ higher throughput.
APA
Jünger, P., Lovelace, J., Zhao, L., Go, D. & Weinberger, K.Q.. (2026). Self-Augmenting Retrieval for Diffusion Language Models. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:55110-55124 Available from https://proceedings.mlr.press/v306/junger26a.html.

Related Material