* Allow molt annotation on preview datasets, independent of analysis workflow * Separate `create_selection` to run independently in a plain napari installation * Point to workflow config directly to the best checkpoint of the trained model (path on tachyon) * Use `git-lfs` version control for trained model * Migrate segmentation quality plots from `ggrossha_GRH1_NHR23` (`s03_measure`)
create_selectionto run independently in a plain napari installationgit-lfsversion control for trained modelggrossha_GRH1_NHR23(s03_measure)