nvidia/GR00T-N1.6-G1-PnPAppleToPlate
Partial fine-tune vs nvidia/GR00T-N1.6-3B. Partial change: 39% of the weights changed (attention, embeddings, mlp, norms, other), in 4 of 27 layers.
Tensors changed: Embeddings 4/9, Attention 272/540, other 137/149, Norms 80/241, lm_head 0/1, MLP 12/166