win10/EVE-27B-XENO-HAT

Full fine-tune vs OrionLLM/GRM-2.6-Plus-0628. Fine-tune: 100% of the weights changed, across 64 of 64 layers.

Tensors changed: Embeddings 5/5, MLP 307/307, Attention 560/560, lm_head 1/1, Norms 325/325, other 1/1

2 derivatives checked