nurdich/Qwen3.8-9B-Distill-uncensored-heretic

Partial fine-tune vs Qwen/Qwen3.5-9B. Partial change: 72% of the weights changed (attention, mlp, norms), in 32 of 32 layers.

Tensors changed: lm_head 0/1, Embeddings 0/5, MLP 96/211, Attention 177/336, other 0/1, Norms 58/221

2 derivatives checked

• mradermacher/Qwen3.8-9B-Distill-uncensored-heretic-GGUF: Same weights, converted
• mradermacher/Qwen3.8-9B-Distill-uncensored-heretic-i1-GGUF: Quantization

On the Hub · Downloads on Model Pulse