fyuuki0jp/gemma-2-2b-jpn-it_finetuning_sft
Partial fine-tune vs google/gemma-2-2b-jpn-it. Partial change: 27% of the weights changed (attention, embeddings), in 26 of 26 layers.
Tensors changed: Embeddings 1/1, Norms 0/105, MLP 0/78, Attention 52/104