NbAiLab/borealis2-26b-a4b-preview

Full fine-tune vs google/gemma-4-26B-A4B-it. Fine-tune: 98% of the weights changed, across 30 of 30 layers.

Tensors changed: Embeddings 1/4, MLP 150/231, Norms 120/433, other 30/122, Attention 115/223

2 derivatives checked

1 other repos hold byte-for-byte identical weight files.