cyberagent/ca-reward-3b-ja
Full fine-tune vs sbintuitions/sarashina2.2-3b-instruct-v0.1. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 1/1, Norms 1/65, MLP 96/96, Attention 128/128
Full fine-tune vs sbintuitions/sarashina2.2-3b-instruct-v0.1. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 1/1, Norms 1/65, MLP 96/96, Attention 128/128