12B-Suite/G4-Prototype01-12B-v0.1

Full fine-tune vs MuXodious/gemma-4-12B-it-QAT-SOMPOA-heresy. Fine-tune: 100% of the weights changed, across 48 of 48 layers.

Tensors changed: Embeddings 1/12, Norms 0/289, other 0/48, MLP 144/144, Attention 184/184

2 derivatives checked