hugpadhugpad

z-lab / Qwen3.5-4B-DFlash

Text Generation634M13.3K44Updated Jun 19, 2026
Hugging Face

Downloads

13.3K

Likes

44

Last updated

Jun 19, 2026

License

apache-2.0

This DFlash draft model is a joint retrain from Z-Lab and Modal, trained with 40k sequence length and sliding-window attention for improved long-context performance. It is mirrored across the following Hugging Face repositories:

Tasks & library

  • text generation
  • transformers

Tags

  • transformers
  • qwen3
  • feature-extraction
  • dflash
  • speculative-decoding
  • speculative-decoding-draft

Technical specifications

Parameters634M
Librarytransformers
LanguagesNot available
Base modelQwen/Qwen3.5-4B
Licenseapache-2.0