def _save(self, phase, epoch):
"""Save model state dict and generator's samples"""
if phase != 'ADV':
torch.save(self.gen.state_dict(), cfg.save_model_root + 'gen_{}_{:05d}.pt'.format(phase, epoch))
save_sample_path = cfg.save_samples_root + 'samples_{}_{:05d}.txt'.format(phase, epoch)
samples = self.gen.sample(cfg.batch_size, cfg.batch_size)
write_tensor(save_sample_path, samples)
def _save(self, phase, epoch):
torch.save(
self.gen.state_dict(),
cfg.save_model_root + 'gen_{}_{:05d}.pt'.format(phase, epoch))
save_sample_path = cfg.save_samples_root + 'samples_{}_{:05d}.txt'.format(
phase, epoch)
samples = self.gen.sample(cfg.batch_size, cfg.batch_size, self.dis)
write_tensor(save_sample_path, samples)
def _save(self, phase, epoch):
"""Save model state dict and generator's samples"""
for i in range(cfg.k_label):
torch.save(
self.gen_list[i].state_dict(), cfg.save_model_root +
'gen{}_{}_{:05d}.pt'.format(i, phase, epoch))
save_sample_path = cfg.save_samples_root + 'samples_d{}_{}_{:05d}.txt'.format(
i, phase, epoch)
samples = self.gen_list[i].sample(cfg.batch_size, cfg.batch_size)
write_tensor(save_sample_path, samples)
def _save(self, phase, epoch, label_i=None):
assert type(label_i) == int
torch.save(
self.gen.state_dict(),
cfg.save_model_root + 'gen_{}_{:05d}.pt'.format(phase, epoch))
save_sample_path = cfg.save_samples_root + 'samples_c{}_{}_{:05d}.txt'.format(
label_i, phase, epoch)
samples = self.gen.sample(cfg.batch_size,
cfg.batch_size,
label_i=label_i)
write_tensor(save_sample_path, samples)