cl-nagoya/ruri-v3-pt-70m
Full fine-tune vs sbintuitions/modernbert-ja-70m. Fine-tune: 100% of the weights changed, across 13 of 13 layers.
Tensors changed: Embeddings 2/2, Norms 26/26, Attention 26/26, MLP 26/26
2 derivatives checked
- cl-nagoya/ruri-v3-70m: Full fine-tune
- hotchpotch/japanese-reranker-small-v2: Full fine-tune