add custom loss arg to supervised_training.py by stmartineau99 · Pull Request #161 · computational-cell-analytics/synapse-net

stmartineau99 · 2026-03-16T16:03:00Z

added custom loss_fn arg to supervised_training
if None, defaults to torch_em.Loss.DiceLoss
custom loss functions are expected to be compatible with ignore_label and mask_channel args (loss wrappers)

GOESTERN-1107078 added 2 commits March 16, 2026 17:02

add custom loss arg to supervised_training.py

70b8fcc

update supervised_training.py for custom loss

794e8d2

constantinpape approved these changes Apr 13, 2026

View reviewed changes

constantinpape merged commit 97acb6f into main Apr 13, 2026
3 checks passed

constantinpape deleted the cldice_supervised_training branch April 13, 2026 08:22

Provide feedback