metrics of pre-training (loss and learning rate) and fine-tuning (loss, learning rate, MAE, RMSE);
images of predictions produced by the forward pass of the pre-training every t steps (original, masked, recovered);
images of exemplars and predictions produced by the forward pass of the fine-tuning every t steps (GT, predicted density map, overlay of input + prediction);
model weights every n epochs for pre-training and fine-tuning, and the model with minimum MAE for fine-tuning.
New bash scriptrun.sh to reproduce results and/or run paper's pre-training and fine-tuning.
Parametric paths in FSC147.py, FSC_pretrain.py, FSC_finetune_cross.py.
Some examples of images logged during pre-training
and fine-tuning
Added:
run.shto reproduce results and/or run paper's pre-training and fine-tuning.FSC147.py,FSC_pretrain.py,FSC_finetune_cross.py.Some examples of images logged during pre-training
and fine-tuning
