netcat420/MFANN3b-SFT

Partial fine-tune vs liminerity/Phigments12. Partial change: 23% of the weights changed (attention), in 32 of 32 layers.

Tensors changed: Norms 0/66, MLP 0/128, Attention 97/256, lm_head 0/2, Embeddings 0/1

1 derivatives checked