ISTA-DASLab / Qwen3.8-Flash-Next-P48NVFP4-MoESQ

Text Generation135B1.2K14Updated Oct 5, 2026

Downloads

1.2K

Likes

14

Last updated

Oct 5, 2026

License

qwen-community-1.0

A W4A4 + paired-4:8 sparse compressed checkpoint of Qwen/Qwen3.8-Flash-Next, produced with MoESQ. The routed MoE expert weights are NVFP4 with paired-4:8 structured sparsity and are stored sparse: only the kept values plus a small mask are on disk. The target is NVIDIA Blackwe…

Tasks & library

  • text generation
  • transformers

Tags

  • transformers
  • qwen4_exp
  • image-text-to-text
  • qwen4-exp
  • moe
  • nvfp4

Technical specifications

Parameters135B
Librarytransformers
Languagesen
Licenseqwen-community-1.0