prismdata/Perdix-1.1B-Instruct
Full fine-tune vs prismdata/Perdix-1.1B-Base. Fine-tune: 100% of the weights changed, across 20 of 20 layers.
Tensors changed: Attention 180/180, Norms 41/41, MLP 80/80
Full fine-tune vs prismdata/Perdix-1.1B-Base. Fine-tune: 100% of the weights changed, across 20 of 20 layers.
Tensors changed: Attention 180/180, Norms 41/41, MLP 80/80