Topic: "gated-attention"

Sep 11
Qwen3-Next-80B-A3B-Base: Towards Ultimate Training & Inference Efficiency
Show details
qwen3-next qwen3 mixtral-8x7b gemini-2.5-pro alibaba mistral-ai deepseek snowflake hugging-face baseten nvidia mixture-of-experts model-sparsity gated-attention hybrid-architecture rmsnorm model-stability model-training inference-optimization multi-token-prediction model-deployment justinlin610 teortaxestex yuchenj_uw
MoE (Mixture of Experts) models have become essential in frontier AI models, with Qwen3-Next pushing sparsity further by activating only 3.7% of parameters (3B out of 80B) using a hybrid architecture combining Gated DeltaNet and Gated Attention. This new design includes 512 total experts (10 routed + 1 shared), Zero-Centered RMSNorm for stability, and improved MoE router initialization, resulting in ~10× cheaper training and 10× faster inference compared to previous models. Alibaba's Qwen3-Next reportedly outperforms Gemini-2.5-Flash-Thinking and approaches the flagship 235B model's performance, with deployments on Hugging Face, Baseten, and native vLLM support for efficient inference.