ThomasComics/MS-ManciousWriter-22B-v0.2

Full fine-tune vs allura-org/MS-Meadowlark-22B. Fine-tune: 100% of the weights changed, across 56 of 56 layers.

Tensors changed: lm_head 1/1, Embeddings 1/1, Norms 64/113, MLP 168/168, Attention 224/224

2 derivatives checked