This PR covers changes as a reflection to the 2nd Diamond Hackathon's discussion.
Feature Upgrades
[x] Refactored the Parameter Class for Pydantic BaseModel to lift the nested structure from last version, thanks to @dylanmcreynolds !
[x] I/O: Now both train.py and segment.py will only take one argument yaml_file_path.
[x] Tiled Dataset: The InferenceDataset has been taken out and TrainingDataset has been modified to a unified TiledDataset which will handle both training and inference.
[x] Testing: Example yaml files have been provided as a template under a new directory.
To Test:
Modify the tiled_api_key for all clients under each example yaml file.
This PR covers changes as a reflection to the 2nd Diamond Hackathon's discussion.
Feature Upgrades
train.py
andsegment.py
will only take one argumentyaml_file_path
.InferenceDataset
has been taken out andTrainingDataset
has been modified to a unifiedTiledDataset
which will handle both training and inference.To Test:
tiled_api_key
for all clients under each example yaml file.uid
if necessary.make train_tunet
make segment_tunet