ARGA100/ru-reranker-modernbert-small
Full fine-tune vs deepvk/RuModernBERT-small. Fine-tune: 100% of the weights changed, across 12 of 12 layers.
Tensors changed: other 1/1, Norms 25/25, Embeddings 2/2, Attention 24/24, MLP 24/24
Full fine-tune vs deepvk/RuModernBERT-small. Fine-tune: 100% of the weights changed, across 12 of 12 layers.
Tensors changed: other 1/1, Norms 25/25, Embeddings 2/2, Attention 24/24, MLP 24/24