benjamintli/dfine-xl_sku110k
Full fine-tune vs ustc-community/dfine-xlarge-obj365. Fine-tune: 100% of the weights changed, across 6 of 6 layers.
Tensors changed: other 303/305, Norms 470/970, Embeddings 41/61, Attention 80/86, MLP 28/28
Full fine-tune vs ustc-community/dfine-xlarge-obj365. Fine-tune: 100% of the weights changed, across 6 of 6 layers.
Tensors changed: other 303/305, Norms 470/970, Embeddings 41/61, Attention 80/86, MLP 28/28