Delta-Vector/Mag-Picaro-72B
Full fine-tune vs Qwen/Qwen2-72B-Instruct. Fine-tune: 100% of the weights changed, across 80 of 80 layers.
Tensors changed: MLP 240/240, Attention 560/560, Norms 122/161
2 derivatives checked
- mradermacher/Mag-Picaro-72B-i1-GGUF: Quantization
- mradermacher/MagPicaro-Big-qwen-GGUF: Quantization