Track diffusion language model research: what changed, why it matters, and what to try.

Semantic search is unavailable in the public read model; results use lexical matching.

Window: seven UTC calendar dates, including today, through the current instant. .

Browse latest papers · Browse the archive

Fresh core papers from the last 7 days

2026-09-07T07:55:40+00:00 · Post-training · Source

In-Place Instruction Following in Diffusion Language Models

Diffusion Large Language Models (dLLMs) generate text via bidirectional iterative denoising, naturally supporting user-specified constraints anchored at arbitrary output positions, a paradigm known as In-place Prompting (IPP). We formalize this as the In-place Instruction Following (IIF) task and construct IIF-Bench, a hierarchical benchmark spanning literal, style, and discourse-function constraints, paired with a rubric-based local-global evaluation protocol. An inference-time attention-bias probe suggests that vanilla dLLMs often under-prioritize constraint spans during denoising. We then p

Useful for: The approach applies to diffusion large language models supporting user-specified constraints at arbitrary output positions. · Limitation: Vanilla dLLMs often under-prioritize constraint spans during denoising.

2026-09-06T01:09:14+00:00 · unclassified · Source

A Ticket from Marginals to Joints: Coupled-Noise Distillation for One-Step Block Generation in Diffusion Language Models

Autoregressive language models commit one token per forward pass; diffusion language models commit a block of tokens over several steps. We ask whether a block can be committed in a single forward pass. We study this with a noise-conditioned masked denoiser: a data-independent Gaussian noise field is added to the mask embeddings so that, in principle, each sampled field selects one joint mode of the block. The established way of training such a model is to sample several fields per example and let them compete for the data, by winner-take-all or importance weighting. This gives the noise only

Useful for: The method applies to one-step block generation with one forward pass per block. · Limitation: The source does not provide quantitative results or specify the tested model sizes.

Public snapshot dates
Collection recorded in this snapshot
Newest publication recorded in this snapshot
Snapshot generated

These are saved observations, not the last collector attempt or a guarantee of corpus completeness.

What changed this week

2026-09-05 – 2026-09-11 UTC · Snapshot: 2026-09-11T06:19:44.013112+00:00

Recorded core-corpus events. Dates show when this site observed a change, not when the paper or code was released. This is not a complete account of the field.

  • Added to core corpus: 103
  • Source changed: 5
  • Code observed: 0

Open weekly summary

Learning paths

Changed in your topics

Topics are optional. Without an account, your choices stay in this browser.

Your topics

Choose up to three topics, or leave the selection empty.