def generate_validation_data():
global val_inputs, val_shifts, val_targets
global tst_inputs, tst_shifts, tst_targets
print "Generating validation data..."
val_inputs, val_shifts, val_targets = generate_data_2d(cfg.width, cfg.height, cfg.n_val_samples, with_hot=with_hot,
force_id=use_id_data)
tst_inputs, tst_shifts, tst_targets = generate_data_2d(cfg.width, cfg.height, cfg.n_val_samples, binary=True,
with_hot=with_hot, force_id=use_id_data)
print "Done."
def generate_new_data():
global trn_inputs, trn_shifts, trn_targets
#print "Generating new data..."
trn_inputs, trn_shifts, trn_targets = \
generate_data_2d(cfg.width, cfg.height, cfg.n_batch, with_hot=with_hot, force_id=use_id_data)
if use_id_data:
trn_targets = trn_inputs.copy()
# final weights
plt.clf()
plot_all_weights(ps)
plt.suptitle("final iter=%d" % iteration)
plt.draw()
if ml.common.plot.headless:
plt.savefig(plot_dir + "/weights_final.pdf")
# final loss
plt.clf()
his.plot()
plt.savefig(plot_dir + "/loss.pdf")
# check with simple patterns
sim_inputs, sim_shifts, sim_targets = generate_data_2d(cfg.width, cfg.height, 3, binary=True, with_hot=with_hot,
force_id=use_id_data)
if use_id_data:
sim_targets = sim_inputs.copy()
sim_results = gather(f_output(ps.data, post(sim_inputs), post(sim_shifts)))
threshold = None
print "input: "
print format_sample(sim_inputs[:, 0], cfg.width, cfg.height, threshold=threshold)
print "shift: "
print sim_shifts[:, 0]
print "targets: "
print format_sample(sim_targets[:, 0], cfg.width, cfg.height, threshold=threshold)
print "results: "
print format_sample(sim_results[:, 0], cfg.width, cfg.height, threshold=threshold)
# test on global test set