hiebo/Qwen3.6-34B-80L-Fable-5-Heretic
Partial fine-tune vs Qwen/Qwen3.6-27B. Partial change: 55% of the weights changed (attention, mlp, norms), in 59 of 64 layers.
Tensors changed: Embeddings 0/5, MLP 123/307, Attention 246/560, lm_head 0/1, Norms 104/325, other 0/1
2 derivatives checked
- mradermacher/Qwen3.6-34B-80L-Fable-5-Heretic-GGUF: Quantization
- mradermacher/Qwen3.6-34B-80L-Fable-5-Heretic-i1-GGUF: Quantization
2 other repos hold byte-for-byte identical weight files.