himel7/multilingual-bias-detector
Full fine-tune vs himel7/bias-detector. Fine-tune: 100% of the weights changed, across 12 of 12 layers.
Tensors changed: other 52/52, Embeddings 2/2, Attention 96/96
Full fine-tune vs himel7/bias-detector. Fine-tune: 100% of the weights changed, across 12 of 12 layers.
Tensors changed: other 52/52, Embeddings 2/2, Attention 96/96