Ray2333/GRM-gemma2-2B-rewardmodel-ft
Partial fine-tune vs Ray2333/GRM-Gemma2-2B-sftreg. Partial change: 77% of the weights changed (attention, mlp, norms), in 26 of 26 layers.
Tensors changed: Embeddings 0/1, Norms 92/105, MLP 78/78, Attention 104/104