Applied-Innovation-Center/Karnak-40B-v1.0
Full fine-tune vs Qwen/Qwen3-30B-A3B-Instruct-2507. Fine-tune: 100% of the weights changed, across 48 of 48 layers.
Tensors changed: Norms 193/193, MLP 1871/1871, Attention 192/192
2 derivatives checked
- mradermacher/Karnak-GGUF: Quantization
- mradermacher/Karnak-i1-GGUF: Quantization
1 other repos hold byte-for-byte identical weight files.