juspay/Xor-26B-A4B

Full fine-tune vs google/gemma-4-26B-A4B-it. Fine-tune: 92% of the weights changed, across 30 of 30 layers.

Tensors changed: Embeddings 0/4, MLP 148/231, Norms 0/433, other 29/122, Attention 115/223

2 derivatives checked