Crimson AI NewsA CrimsonLingua Network service
EN ع
← Back to news
Research paper Hugging Face

Hugging Face Unveils LLaDA MoE v2: Scaling Laws for Diffusion Language Models

AI By Crimson AI Hugging Face Papers 5 August 2026 · 00:00 19 views
Share: X Telegram

A new paper from Hugging Face introduces LLaDA MoE v2, a 30B-A3B diffusion language model trained on 23.5T tokens, and reveals scaling laws for MoE dLLMs that differ from autoregressive models.

Hugging Face Unveils LLaDA MoE v2: Scaling Laws for Diffusion Language Models

Key points

Hugging Face researchers have published a paper detailing LLaDA MoE v2, a 30B-A3B diffusion language model (dLLM) trained from scratch on 23.5 trillion tokens. The work systematically characterizes how optimization hyperparameters, compute allocation, and architecture scale for Mixture-of-Experts (MoE) dLLMs, revealing quantitative differences from scaling trends previously observed in autoregressive (AR) models.

Key findings include that for optimization, the optimal nominal batch size grows faster, while the optimal learning rate decays more rapidly with compute. For model-data allocation, IsoFLOP analysis shows a slight data-side tilt: the optimal token budget grows faster than activated model-side computation. For MoE architecture, larger scales increasingly favor larger expert pools at fixed activated capacity, while moderate expert granularity remains effective and the preferred fraction of activated capacity assigned to shared experts stays stable.

Guided by these insights, the team trained LLaDA MoE v2 using approximately 65% as many pretraining tokens as Qwen3. Despite this, the model approaches Qwen3 on several knowledge, reasoning, and coding benchmarks. After supervised fine-tuning alone, it outperforms SDAR Chat on seven of eight reasoning and coding benchmarks and remains close to Qwen3 on several tasks.

The paper establishes practical scaling laws and design principles for MoE dLLMs, offering a roadmap for future development of diffusion-based language models. The full paper is available on arXiv.

ModelParametersPretraining TokensBenchmark Performance
LLaDA MoE v230B-A3B23.5TApproaches Qwen3; outperforms SDAR Chat on 7/8 reasoning/coding benchmarks after SFT
Qwen3Not specified~36T (inferred from 65% ratio)Baseline
Source
Hugging Face · Hugging Face Papers
Related news
Research paper
Hugging Face 31 Aug 2026

Hugging Face Unveils StepGuard: Step-Level Guardrails for Safer AI Agents

StepGuard, a new step-level guard model from Hugging Face, audits agent actions before execution, reducing attack success rates by...

1
Research paper
Hugging Face 31 Aug 2026

Hugging Face Researchers Unveil ABot-Recon for Stable Long-Horizon 3D Reconstruction

ABot-Recon, a new streaming 3D reconstruction model from Hugging Face, achieves stable long-horizon performance using only local t...

1
Research paper
Hugging Face 31 Aug 2026

ContextPilot: Teaching Agents Proactive Context Management via Fine-Grained RL

Hugging Face researchers introduce ContextPilot, a framework that enhances long-horizon agent reasoning by expanding context-editi...

1