split_map={'train': 0.7, 'eval': 0.3}))
# Add a preprocessing unit
training_pipeline.add_preprocesser(
StandardPreprocesser(
features=['times_pregnant', 'pgc', 'dbp', 'tst', 'insulin', 'bmi',
'pedigree', 'age'],
labels=['has_diabetes'],
overwrite={'has_diabetes': {
'transform': [{'method': 'no_transform', 'parameters': {}}]}}
))
# Add a trainer
training_pipeline.add_trainer(FeedForwardTrainer(
loss='binary_crossentropy',
last_activation='sigmoid',
output_units=1,
metrics=['accuracy'],
epochs=20))
# Add an evaluator
training_pipeline.add_evaluator(
TFMAEvaluator(slices=[['has_diabetes']],
metrics={'has_diabetes': ['binary_crossentropy',
'binary_accuracy']}))
# Run the pipeline on a Google Cloud VM and train on GCP as well
# In order for this to work, the orchestrator and the backend should be in the
# same GCP project. Also, the metadata store and artifact store should be
# accessible by the orchestrator VM and the GCAIP worker VM.
# Note: If you are using a custom Trainer, then you need
ds = Repository.get_instance().get_datasource_by_name(
'Pima Indians Diabetes')
training_pipeline.add_datasource(ds)
# Add a split
training_pipeline.add_split(RandomSplit(
split_map={'train': 0.7, 'eval': 0.3}))
# Add a preprocessing unit
training_pipeline.add_preprocesser(
StandardPreprocesser(
features=['times_pregnant', 'pgc', 'dbp', 'tst', 'insulin', 'bmi',
'pedigree', 'age'],
labels=['has_diabetes'],
overwrite={'has_diabetes': {
'transform': [{'method': 'no_transform', 'parameters': {}}]}}
))
# Add a trainer
training_pipeline.add_trainer(MyScikitTrainer(
C=1.0,
kernel='rbf',
))
# Run the pipeline locally
training_pipeline.run()
# See statistics of train and eval
training_pipeline.view_statistics()
path='gs://zenml_quickstart/diabetes.csv')
except AlreadyExistsException:
ds = Repository.get_instance().get_datasource_by_name(
'Pima Indians Diabetes')
training_pipeline.add_datasource(ds)
# Add a split
training_pipeline.add_split(RandomSplit(split_map={'eval': 0.3, 'train': 0.7}))
# Add a preprocessing unit
training_pipeline.add_preprocesser(
StandardPreprocesser(features=[
'times_pregnant', 'pgc', 'dbp', 'tst', 'insulin', 'bmi', 'pedigree',
'age'
],
labels=['has_diabetes'],
overwrite={
'has_diabetes': {
'transform': [{
'method': 'no_transform',
'parameters': {}
}]
}
}))
# Add a trainer
training_pipeline.add_trainer(MyPyTorchLightningTrainer(epoch=100))
# Run the pipeline locally
training_pipeline.run()
from examples.gan.preprocessing import GANPreprocessor
repo: Repository = Repository().get_instance()
gan_pipeline = TrainingPipeline(name="whynotletitfly", enable_cache=False)
try:
ds = ImageDatasource(
name="gan_images",
base_path="/Users/nicholasjunge/workspaces/maiot/ce_project/images_mini"
)
except:
ds = repo.get_datasource_by_name('gan_images')
gan_pipeline.add_datasource(ds)
gan_pipeline.add_split(
CategoricalDomainSplit(categorical_column="label",
split_map={
"train": [0],
"eval": [1]
}))
gan_pipeline.add_preprocesser(GANPreprocessor())
# gan_pipeline.add_preprocesser(transform_step)
gan_pipeline.add_trainer(CycleGANTrainer(epochs=5))
gan_pipeline.run()