Youssofal/Qwen3.6-27B-Abliterated-Heretic-Uncensored-BF16
LoRA-style vs Qwen/Qwen3.6-27B. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, o_proj, out_proj, up_proj changed, in 30 of 64 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/4, MLP 84/304, Attention 25/556, lm_head 0/1, Norms 0/319
5 derivatives checked
- Youssofal/Qwen3.6-27B-Abliterated-Heretic-Uncensored-MLX-4bit: Quantization
- Youssofal/Qwen3.6-27B-Abliterated-Heretic-Uncensored-MLX-8bit: Quantization
- Youssofal/Qwen3.6-27B-Abliterated-Heretic-Uncensored-MLX-3bit: Quantization
- Youssofal/Qwen3.6-27B-Abliterated-Heretic-Uncensored-MLX-6bit: Quantization
- underlotus/Qwen3.6-27B-Abliterated-Heretic-Uncensored-oQ4: Quantization