LDLMIn construction

All models, with previews →

Meshchaninov et al. · May 2026Diffusion language model

Denoiser block

Pinned implementation and paper LDLM @ e18c42a, not yet public · arXiv:2605.07933v1 · OpenWebText · 1024 · checkpoint, tokenizer and saved statistics not verified · How to read

TrainingGeneration

Denoiser › 12 denoiser blocks

Add attention update

×

no parameters

Add the mechanism’s update to the residual stream element by element.

xn′=xn+Δ~attn{\htmlData{light=residual attention-add}{x'_n}}={\htmlData{light=block-input}{x_n}}+{\htmlData{light=attention-gate}{\tilde\Delta_{\mathrm{attn}}}}
in
xnx_n [B batch, T positions, 768 d]Δ~attn\tilde\Delta_{\mathrm{attn}} [B batch, T positions, 768 d]
out
xn′x'_n [B batch, T positions, 768 d]
Sources2

Direct evidence

The residual update is scale · dropout(x) added to the residual; modulation is x · (1 + scale) + shift.

The implementation, pinned at revision e18c42acdfb7, not yet public

Direct evidence

A block normalizes, modulates by the time condition’s shift and scale, projects to queries, keys and values, rotates queries and keys, attends over all positions with softcap, and adds the projected output scaled by a gate; the MLP branch does the same with its own shift, scale and gate.

The implementation, pinned at revision e18c42acdfb7, not yet public