LDLMIn construction

All models, with previews →

Meshchaninov et al. · May 2026Diffusion language model

Perceiver 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

TrainingGeneration

Latent encoder › 6 encoder blocks › Perceiver attention

Features with positions

×

supplied · no parameters

The standardized GPT-2 features with their positions added: read through keys and values only.

shape
hph^{p} [B batch, T positions, 768 d]
Sources

Contextual evidence from Add positions

The encoder adds positions to the features, normalizes the learned queries, applies the blocks, each attending to the features, and normalizes the result.

The implementation, pinned at revision e18c42acdfb7, not yet public

Contextual evidence from RMSNorm

RMSNorm divides by the vector’s norm times width^(−1/2), clamped from below at ε, and multiplies by a learned gain.

The implementation, pinned at revision e18c42acdfb7, not yet public

Contextual evidence from RMSNorm

Perceiver attention owns an RMSNorm for each of its two streams, a fused query–key–value projection of the query stream, a fused key–value projection of the other stream, RMSNorms for each head’s queries and keys, and an output projection; no bias.

The implementation, pinned at revision e18c42acdfb7, not yet public

Contextual evidence from RMSNorm

Both streams are normalized; queries come from one stream, keys and values from both, concatenated along positions; queries and keys are normalized per head; attention runs over all positions with softcap; the heads are merged and projected.

The implementation, pinned at revision e18c42acdfb7, not yet public