LAW proposes dropping diffusion entirely: language models built on pure self-conditioning
ArnaudDoucet1 · x · 2026-10-23
Self-Conditioning (SC) in diffusion language models is polarizing: carrying two states during sampling is costly. While recent work tries to remove SC, the new LAW ("Loopholing All the Way") asks the opposite question — what if you drop diffusion instead, and build a language model purely on self-conditioning?
The work was shared in an 8-post thread by the authors and amplified by prominent diffusion researcher Arnaud Doucet.
Related event: LAW: A language model built purely on self-conditioning(3 posts)→
More from Models
- NVIDIA's sub-1B open-weight Nemotron 3.5 Parse takes #1 on LlamaIndex's ParseBench — lmoroney · 2026-10-23
- TypeSafe's Jev: a "Decision Model" LLM That Returns Floats at $0.042/M Input Tokens — ariG23498 · 2026-10-23
- Microsoft's MAI-Base-2 scaling ladder pegs the L42 run at ~1.8×10^23 FLOPs — teortaxesTex · 2026-10-23
- Local model agentic coding benchmark: 27B matches 125B, tool-calling is the lever — Joralk_ · 2026-10-23
- NVIDIA open-sources PersonaPlex-7B, a full-duplex voice AI model that listens and talks simultaneously — solyarisoftware · 2026-10-23
- NVIDIA's KumoRFM-2: a foundation model that predicts churn, fraud and demand straight from relational databases — solyarisoftware · 2026-10-23