nasos10/MuToR-gemma-2B-GSM8K-dmax_4_a_03
Full fine-tune vs google/gemma-2b. Fine-tune: 100% of the weights changed, across 18 of 18 layers.
Tensors changed: Norms 31/37, MLP 54/54, Attention 72/72
Full fine-tune vs google/gemma-2b. Fine-tune: 100% of the weights changed, across 18 of 18 layers.
Tensors changed: Norms 31/37, MLP 54/54, Attention 72/72