ianpan/mammoscreen
Full fine-tune (fewer layers) vs timm/tf_efficientnetv2_s.in21k_ft_in1k. timm/tf_efficientnetv2_s.in21k_ft_in1k matches part of ianpan/mammoscreen (779 of its 2346 tensors; the rest of ianpan/mammoscreen has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 6 of 6 layers.
Tensors changed: other 779/779