NYUAD-ComNets/FaceScanPaliGemma_Race
Partial fine-tune vs google/paligemma-3b-pt-224. Partial change: 82% of the weights changed (attention, embeddings, mlp, norms, other), in 27 of 27 layers.
Tensors changed: Embeddings 3/4, Norms 113/147, MLP 157/162, Attention 288/288, other 2/2