3 Downloads Updated 10 hours ago
ollama run AndyZ/GFEI-Qwen14B-Tao
Updated 10 hours ago
10 hours ago
6088976be2bf · 9.0GB
A Qwen2.5-14B-Instruct model fine-tuned with QLoRA (r=32, α=64) on 1,704 Chinese samples distilled from three value systems:
The model internalizes four foundational values:
Researchers, practitioners, and curious users who want a model that holds a clear Daoist stance while remaining grounded, compassionate, and willing to push back against shallow takes.
Inherits the Qwen2.5-14B-Instruct base model license (Apache 2.0 + Qwen Research Use restrictions). Training data was synthesized from public-domain and original sources; no proprietary fiction or character material was used.