DavidAU/Qwen3.6-12B-IQ-Ultra-Heretic-Uncensored-Thinking-V2-Hightop
Partial fine-tune (fewer layers) vs trohrbaugh/Qwen3.6-27B-heretic-ara. DavidAU/Qwen3.6-12B-IQ-Ultra-Heretic-Uncensored-Thinking-V2-Hightop matches part of trohrbaugh/Qwen3.6-27B-heretic-ara (654 of its 1199 tensors; the rest of trohrbaugh/Qwen3.6-27B-heretic-ara has no counterpart). On the shared part: Partial change: 75% of the weights changed (attention, mlp, norms), in 24 of 27 layers.
Tensors changed: lm_head 0/1, Embeddings 0/4, Norms 65/189, Attention 159/276, MLP 72/184
4 derivatives checked
- KevinJK51/Qwen3.6-12B-IQ-Ultra-Heretic-Uncensored-Thinking-V2-Hightop-GGUF: Quantization
- hellork/Qwen3.6-12B-IQ-Ultra-Heretic-Uncensored-Thinking-V2-Hightop-Q4_K_M-GGUF: Quantization
- larue316/Qwen3.6-12B-IQ-Ultra-Heretic-Uncensored-Thinking-V2-Hightop-GGUF: Quantization
- hellork/Qwen3.6-12B-IQ-Ultra-Heretic-Uncensored-Thinking-V2-Hightop-IQ3_M-GGUF: Quantization
2 other repos hold byte-for-byte identical weight files.