def plot_generated(trainer):
# Get a batch from the train set
train_set_one_epoch = ArrayIterator(dataset.train, batch_size, shuffle=False)
gen_series = trainer.predict(train_set_one_epoch, num_batches=1)
train_set_one_epoch.reset()
# Get an example from the batch
gen_series = gen_series[4]
if args.backward:
# If args.backward is set, the autoencoder would have produced the input sequence in reverse.
# We flip it again to match the true series
gen_series = gen_series[::-1, :]
true_series = next(train_set_one_epoch)['X'][4]
# Plot the true and generated values of each series
ncols = int(np.ceil((dataset.n_sensors + dataset.n_operating_modes) * 1.0 // 3))
fig, ax = plt.subplots(ncols, 3)
fig.set_figheight(20)
fig.set_figwidth(10)
for i in range(dataset.n_operating_modes):
plt.subplot(ncols, 3, i + 1)
if i == 0:
plt.plot(true_series[:, i], label="true", color="blue")
else:
plt.plot(true_series[:, i], color="blue")
if i == 0:
plt.plot(gen_series[:, i], label="gen", color="red")
else:
plt.plot(gen_series[:, i], color="red")
plt.title("Operating mode {}".format(i + 1))
for i in range(dataset.n_sensors):
plt.subplot(ncols, 3, dataset.n_operating_modes + i + 1)
plt.plot(true_series[:, dataset.n_operating_modes + i], color="blue")
plt.plot(gen_series[:, dataset.n_operating_modes + i], color="red")
plt.title("Sensor {}".format(i + 1))
fig.legend()
plt.tight_layout()
fig.savefig(os.path.join(args.results_dir, "generated_series.png"))
error = np.mean(data['answer'].argmax(axis=1) != preds)
test_error.append(error)
val_cost_str = "Epoch {}: validation_cost {}, validation_error {}".format(
e, np.mean(test_loss), np.mean(test_error))
print(val_cost_str)
if args.save_log:
with open(log_file, 'a') as f:
f.write(val_cost_str + '\n')
# Shuffle training set and reset the others
shuf_idx = np.random.permutation(
range(train_set.data_arrays['memory'].shape[0]))
train_set.data_arrays = {k: v[shuf_idx]
for k, v in train_set.data_arrays.items()}
train_set.reset()
dev_set.reset()
print('Training Complete.')
if args.interactive:
interactive_loop(interactive_computation, babi)
if args.test:
# Final evaluation on test set
test_loss = []
test_error = []
for idx, data in enumerate(test_set):
test_output = loss_computation(data)
test_loss.append(np.sum(test_output['test_cross_ent_loss']))
preds = np.argmax(test_output['test_preds'], axis=1)
for idx, data in enumerate(test_set):
test_output = eval_computation(data)
test_loss.append(np.sum(test_output['test_cross_ent_loss']))
preds = np.argmax(test_output['test_preds'], axis=0)
error = np.mean(data['answer'] != preds)
test_error.append(error)
print("Epoch {}, Test_loss {}, test_batch_error {}".format(
e, np.mean(test_loss), np.mean(test_error)))
# Shuffle training set and reset the others
shuf_idx = np.random.permutation(
range(train_set.data_arrays['query'].shape[0]))
train_set.data_arrays = {
k: v[shuf_idx]
for k, v in train_set.data_arrays.items()
}
train_set.reset()
test_set.reset()
if (model_file is not None and e % 50 == 0):
print('Saving model to: ', model_file)
weight_saver.save(filename=model_file)
else:
print('Loading saved model')
with closing(ngt.make_transformer()) as transformer:
eval_computation = make_bound_computation(transformer, eval_outputs,
inputs)
if args.interactive:
interactive_computation = make_bound_computation(
transformer, interactive_outputs, inputs)
weight_saver.setup_restore(transformer=transformer,
computation=eval_outputs,
