anthracite-org/magnum-v3-27b-kto
Full fine-tune vs IntervitensInc/gemma-2-27b-chatml. Fine-tune: 100% of the weights changed, across 46 of 46 layers.
Tensors changed: Embeddings 1/1, Attention 184/184, Norms 185/185, MLP 138/138
4 derivatives checked
- bartowski/magnum-v3-27b-kto-GGUF: Same weights, converted
- mradermacher/magnum-v3-27b-kto-i1-GGUF: Quantization
- mradermacher/magnum-v3-27b-kto-GGUF: Quantization
- djuna/G2-Noranum-27B: Full fine-tune