hugpadhugpad

shisa-ai / Qwen3.8-27B-FP8-BLOCK

Vision28B1381Updated Aug 17, 2026
Hugging Face

Downloads

138

Likes

1

Last updated

Aug 17, 2026

License

apache-2.0

Data-free (RTN) compressed-tensors FP8BLOCK quantization of Qwen/Qwen3.8-27B @ 1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0, with all Gated DeltaNet linear-attention projections kept in BF16. Built 2026-08-17 with llmcompressor modelfreeptq, scheme FP8BLOCK (static 128x128 E4M3 we…

Tasks & library

  • image text to text
  • transformers

Tags

  • transformers
  • qwen3_5
  • image-text-to-text
  • qwen3.8
  • fp8
  • compressed-tensors

Technical specifications

Parameters28B
Librarytransformers
LanguagesNot available
Base modelQwen/Qwen3.8-27B
Licenseapache-2.0