LDLMIn construction

Meshchaninov et al. · May 2026Diffusion language model

Self-attention

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

Training Generation

y_attn[B, T, 768]Q, K, V projection

y_{\mathrm{attn}}\,W_{QKV}

Split query heads Split key heads Split value heads Rotary tableR R RoPE RoPE Q[B, 12, T, 64]K[B, 12, T, 64]V[B, 12, T, 64]

Attend Open →

O[B, 12, T, 64]Merge heads Output projection

\Delta_{\mathrm{attn}}=oW_{O}

Δ_attn[B, T, 768]

Token decoder › 3 blocks › Self-attention

Queries

tensor

A rotated 64-feature query for every head and position.

shape: $Q$ [B batch, 12 heads, T positions, 64 head features]

Sources

Contextual evidence from Rotate queries

The rotary table: angles from positions and inverse frequencies of base 10,000, computed in float32; queries and keys are rotated and values take the identity.

The implementation, pinned at revision e18c42acdfb7, not yet public

Contextual evidence from Rotate queries

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

Contextual evidence from Attention over all positions

The token decoder’s blocks: no softcap, no time conditioning, the gated MLP with the configured multiplier.

The implementation, pinned at revision e18c42acdfb7, not yet public