print_flush("... getting the pre-training functions")
pretraining_fns = stacked_autoencoder.pretraining_functions(training_set=training_set, batch_size=batch_size)
if ENABLE_FINE_TUNING:
print_flush("... getting the fine-tune function")
if fine_tune_supervised:
finetune_fn, validate_model = stacked_autoencoder.finetune_functions(training_set=training_set,
training_labels=training_labels,
test_set=test_set, test_labels=test_labels,
batch_size=batch_size,
learning_rate=fine_tune_learning_rate)
else:
finetune_fn, validate_model = stacked_autoencoder.finetune_functions_unsupervised(training_set=training_set,
test_set=test_set,
batch_size=batch_size,
learning_rate=fine_tune_learning_rate)
print_flush("... pre-training the model")
start_time = clock()
# Pre-train layer-wise
for i in range(stacked_autoencoder.n_layers):
# go through pretraining epochs
learning_rate = pretraining_learning_rates[i]
corruption_level = corruption_levels[i]
for epoch in range(pretraining_epochs):
# go through the training set
c = []
for batch_index in xrange(n_train_batches):
cur_cost = pretraining_fns[i](index=batch_index, corruption=corruption_level, lr=learning_rate)
