Johnblick187/Nexus-Coder-5Q3
Partial fine-tune (fewer layers) vs Youssofal/Qwen3.6-35B-A3B-Abliterated-Heretic-BF16. Johnblick187/Nexus-Coder-5Q3 matches part of Youssofal/Qwen3.6-35B-A3B-Abliterated-Heretic-BF16 (693 of its 1026 tensors; the rest of Youssofal/Qwen3.6-35B-A3B-Abliterated-Heretic-BF16 has no counterpart). On the shared part: Partial change: 0% of the weights changed (norms), in 40 of 40 layers.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 50/131, Attention 0/280, MLP 0/280
2 derivatives checked
- mradermacher/Heretic-5Q3-Nexus-Coder-i1-GGUF: Quantization
- mradermacher/Heretic-5Q3-Nexus-Coder-GGUF: Quantization