Fix

src/test.py

@@ -20,7 +20,11 @@ def test(data, model, checkpoint_info, dataset_id):
  generatedB = os.path.join(path_to_dataset, 'generatedB' + os.sep)
  genA2B = model['genA2B']
  genB2A = model['genB2A']
+<<<<<<< Updated upstream
 
+=======
+
+>>>>>>> Stashed changes
  checkpoint, checkpoint_dir = checkpoint_info
  restore_from_checkpoint(checkpoint, checkpoint_dir)
  test_datasetA, test_datasetB, testA_size, testB_size = data
@@ -57,6 +61,10 @@ def test(data, model, checkpoint_info, dataset_id):
  with tf.device("/cpu:0"): # Preprocess data on CPU for significant performance gains.
  data = load_test_data(dataset_id, project_dir)
  with tf.device("/gpu:0"):
+<<<<<<< Updated upstream
  model = define_model(training=False)
+=======
+ model = define_model(initial_learning_rate, training=False)
+>>>>>>> Stashed changes
  checkpoint_info = initialize_checkpoint(checkpoint_dir, model, training=False)
  test(data, model, checkpoint_info, dataset_id)

src/train.py

@@ -25,7 +25,13 @@
 batch_size = 1 # Set batch size to 4 or 16 if training multigpu
 img_size = 256
 cyc_lambda = 10
+<<<<<<< Updated upstream
 identity_lambda = 0.5
+=======
+identity_lambda = 0
+if dataset_id == 'facades':
+ identity_lambda = 0.5
+>>>>>>> Stashed changes
 epochs = 15
 save_epoch_freq = 5
 batches_per_epoch = models.get_batches_per_epoch(dataset_id, project_dir)
@@ -137,14 +143,23 @@ def train(data, model, checkpoint_info, epochs):
  discA_real = discA(trainA)
  discB_real = discB(trainB)
 
- discA_fake = discA(genB2A_output)
- discB_fake = discB(genA2B_output)
+ discA_fake_refined = discA(genB2A_output)
+ discB_fake_refined = discB(genA2B_output)
  # Sample from history buffer of 50 images:
- discA_fake_refined = discA_buffer.query(discA_fake)
- discB_fake_refined = discB_buffer.query(discB_fake)
+ #discA_fake_refined = discA_buffer.query(discA_fake)
+ #discB_fake_refined = discB_buffer.query(discB_fake)
 
+<<<<<<< Updated upstream
  identityA = genB2A(trainA)
  identityB = genA2B(trainB)
+=======
+ reconstructedA = genB2A(genA2B_output)
+ reconstructedB = genA2B(genB2A_output)
+ identityA = genB2A(trainA)
+ identityB = genA2B(trainB)
+
+ cyc_loss = cyc_lambda * cycle_consistency_loss(trainA, trainB, reconstructedA, reconstructedB)
+>>>>>>> Stashed changes
  id_loss = identity_lambda * cyc_lambda * identity_loss(trainA, trainB, identityA, identityB)
 
  genA2B_loss_basic = generator_loss(discB_fake_refined)
@@ -193,7 +208,7 @@ def train(data, model, checkpoint_info, epochs):
  checkpoint_dir = os.path.join(project_dir, 'saved_models', 'checkpoints')
  with tf.device("/cpu:0"): # Preprocess data on CPU for significant performance gains.
  data = load_train_data(dataset_id, project_dir)
- #with tf.device("/gpu:0"):
+ with tf.device("/gpu:0"):
  model = define_model(initial_learning_rate, training=True)
  checkpoint_info = initialize_checkpoint(checkpoint_dir, model, training=True)
  train(data, model, checkpoint_info, epochs=epochs)