Kaoeiri/MS-MagpantheonselRP-22B-12.22-Fictional
Full fine-tune vs ArliAI/Mistral-Small-22B-ArliAI-RPMax-v1.1. Fine-tune: 100% of the weights changed, across 56 of 56 layers.
Tensors changed: Embeddings 1/1, Norms 64/113, MLP 168/168, Attention 224/224, lm_head 1/1