win10/EVE-26b-XENO-HAT

Partial fine-tune vs win10/K2-nano-v1. Partial change: 9% of the weights changed (attention, mlp, other), in 30 of 30 layers.

Tensors changed: Embeddings 0/4, MLP 171/231, Norms 0/433, other 30/122, Attention 223/223

2 derivatives checked