Miwa-Keita/zenz-v2.5-medium
Full fine-tune vs ku-nlp/gpt2-medium-japanese-char. Fine-tune: 100% of the weights changed, across 24 of 24 layers.
Tensors changed: Attention 96/96, Norms 98/98, MLP 96/96, Embeddings 1/2
Full fine-tune vs ku-nlp/gpt2-medium-japanese-char. Fine-tune: 100% of the weights changed, across 24 of 24 layers.
Tensors changed: Attention 96/96, Norms 98/98, MLP 96/96, Embeddings 1/2