hugpadhugpad

amazon / GKA-primed-HQwen3-8B-Instruct

Text Generation8.5B1K2Updated Sep 19, 2026
Hugging Face

Downloads

1K

Likes

2

Last updated

Sep 19, 2026

License

apache-2.0

GKA-primed-HQwen3-8B-Instruct is a Hybrid language model consisting of 50% Attention layers and 50% Gated KalmaNet (GKA) layers, primed from Qwen3-8B using the Hybrid Model Factory Priming pipeline. The model is instruction-tuned and supports context lengths up to 128K tokens.

Tasks & library

  • text generation
  • transformers

Tags

  • transformers
  • hybrid_qwen3
  • text-generation
  • hybrid
  • ssm
  • state-space-model

Technical specifications

Parameters8.5B
Librarytransformers
LanguagesNot available
Base modelQwen/Qwen3-8B
Licenseapache-2.0