def test_denoising_build(self):
with self.subTest(model_type='denoising'):
ae = building.build_ae('denoising', (1, 32, 32), anomaly=False)
self.assertNotEqual(0., ae.noise_ratio)
with self.subTest(model_type='not denoising'):
ae = building.build_ae('vanilla', (1, 32, 32), anomaly=False)
self.assertEqual(0., ae.noise_ratio)
def test_anomaly_build(self):
with self.subTest(anomaly=False):
ae = building.build_ae('vanilla', (1, 32, 32), anomaly=False)
self.assertEqual(20, ae.encoder.latent_dim)
self.assertEqual(20, ae.bottleneck.latent_dim)
self.assertEqual(20, ae.decoder.latent_dim)
with self.subTest(anomaly=True):
ae = building.build_ae('vanilla', (1, 32, 32), anomaly=True)
self.assertEqual(2, ae.encoder.latent_dim)
self.assertEqual(2, ae.bottleneck.latent_dim)
self.assertEqual(2, ae.decoder.latent_dim)
def run(model_type, dataset, batch_size, gpu, anomaly=False):
assert model_type in AUTOENCODERS
task = 'anomaly' if anomaly else None
pl.seed_everything(42)
datamodule = build_datamodule(dataset, model_type, batch_size, anomaly)
ae = build_ae(model_type, datamodule.dims, anomaly)
logger = build_logger(model_type, dataset, task)
checkpoint_path = _train(model_type, ae, datamodule, logger, gpu)
return checkpoint_path
def setUp(self):
self.data = MNISTDataModule(data_dir='../data')
self.data.prepare_data()
self.data.setup()
self.net = build_ae('vae', self.data.dims)
self.latent_viz = Latent(self.net)
def setUp(self):
self.data = MNISTDataModule(data_dir='../data', exclude=9)
self.data.prepare_data()
self.data.setup()
self.net = build_ae('vanilla', self.data.dims)
self.anomaly_detector = AnomalyDetection(self.net)
