zenlm/zen-family
The zenlm/zen-family is an 8 billion parameter language model fine-tuned by Hanzo AI and Zoo Labs Foundation from Qwen/Qwen3-8B. This model family focuses on identity training, agentic-data fine-tuning, and abliteration, building upon the robust Qwen3 architecture. It is designed to provide a specialized foundation for applications requiring nuanced identity and agentic capabilities. This specific checkpoint is an archived v1.0.1 family overview, with current models available under the zenlm organization.
Loading preview...
Zen AI Model Family (Archived v1.0.1)
This entry represents an archived checkpoint of the Zen AI Model Family, specifically version 1.0.1. Developed by Hanzo AI and Zoo Labs Foundation, this model is a fine-tuned iteration of the Qwen/Qwen3-8B base model, inheriting its 8 billion parameters and Apache-2.0 license.
Key Characteristics
- Base Model: Fine-tuned from Qwen/Qwen3-8B, a robust foundation from Alibaba Cloud.
- Fine-tuning Focus: Incorporates Hanzo's unique identity training, agentic-data fine-tuning, and abliteration techniques.
- Development Philosophy: Zen models are built by fine-tuning the best open-weight models of their respective eras, rather than being developed from scratch.
- Archived Status: This specific
zen-familyentry is an archived overview, with the current and canonical lineup available at huggingface.co/zenlm.
Good For
- Reproducibility: Provides access to the v1.0.1 weights for research and development requiring this specific version.
- Understanding Zen AI's Approach: Illustrates the initial fine-tuning methodology of Hanzo AI and Zoo Labs Foundation on a prominent base model.