hugpadhugpad

ukisai / Swift-Qwen3.8-27B-NVFP4

Vision18B3.4K43Updated Sep 24, 2026
Hugging Face

Downloads

3.4K

Likes

43

Last updated

Sep 24, 2026

License

swift-open-license-1.0

NVFP4/FP8 mixed-precision quantization of Swift-Qwen3.8-27B, UkisAI's reasoning-efficient derivative of Qwen3.8-27B. This update was created directly from the original BF16 source weights using NVIDIA Model Optimizer, following the verified precision mapping of NVIDIA's Qwen3.…

Tasks & library

  • image text to text
  • vllm

Tags

  • vllm
  • qwen3_5
  • qwen3_8
  • efficient-thinking
  • reasoning
  • token-efficient

Technical specifications

Parameters18B
Libraryvllm
LanguagesNot available
Base modelukisai/Swift-Qwen3.8-27b
Licenseswift-open-license-1.0