#model.freeze_encoder()
if args.generate:
if args.tokens_path:
model.do_generate_from_tokens(paths, args.tokens_path, verbose=True)
else:
model.do_generate(paths,
data_path,
index,
args.test_speakers,
args.test_utts_per_speaker,
use_half=use_half,
verbose=True,
only_discrete=args.only_gen_discrete)
else:
logger.set_logfile(paths.logfile_path())
logger.log('------------------------------------------------------------')
logger.log('-- New training session starts here ------------------------')
logger.log(time.strftime('%c UTC', time.gmtime()))
logger.log('beta={}'.format(args.beta))
logger.log('num_group={}'.format(args.num_group))
logger.log('test_speakers ={}'.format(args.test_speakers))
logger.log('num_sample={}'.format(args.num_sample))
writer = SummaryWriter(paths.logfile_path() + '_tensorboard')
writer.add_scalars('Params/Train', {'beta': args.beta})
writer.add_scalars('Params/Train', {'num_group': args.num_group})
writer.add_scalars('Params/Train', {'num_sample': args.num_sample})
#model.do_train(paths, dataset, optimiser, writer, epochs=args.epochs, test_epochs=args.test_epochs, batch_size=args.batch_size, step=step, epoch=epoch, use_half=use_half, valid_index=test_index, beta=args.beta)
if model_type[:5] == 'vqvae':
model.do_train(paths,
dataset,
ssvae_parser.add_argument('--continue-from',
type=str,
help="model file path to make continued from")
ssvae_parser.add_argument('wav_files',
type=str,
nargs='+',
help="list of wav_files for prediction")
args = parser.parse_args()
if args.model is None:
parser.print_help()
sys.exit(1)
# some assertions to make sure that batching math assumptions are met
assert parse_torch_version() >= (0, 2,
1), "you need pytorch 0.2.1 or later"
set_logfile(Path(args.log_dir, "predict.log"))
logger.info(f"Prediction started with command: {' '.join(sys.argv)}")
args_str = [f"{k}={v}" for (k, v) in vars(args).items()]
logger.info(f"args: {' '.join(args_str)}")
if args.use_cuda:
torch.set_default_tensor_type("torch.cuda.FloatTensor")
# run prediction
predict = Predict(args)
predict(args.wav_files, logging=True)
ssvae_parser.add_argument('--continue-from',
default=None,
type=str,
help="model file path to make continued from")
args = parser.parse_args()
if args.model is None:
parser.print_help()
sys.exit(1)
# some assertions to make sure that batching math assumptions are met
assert parse_torch_version() >= (0, 2,
1), "you need pytorch 0.2.1 or later"
set_logfile(Path(args.log_dir, "train.log"))
logger.info(f"Training started with command: {' '.join(sys.argv)}")
args_str = [f"{k}={v}" for (k, v) in vars(args).items()]
logger.info(f"args: {' '.join(args_str)}")
if args.use_cuda:
logger.info("using cuda")
torch.set_default_tensor_type("torch.cuda.FloatTensor")
if args.seed is not None:
torch.manual_seed(args.seed)
np.random.seed(args.seed)
if args.use_cuda:
torch.cuda.manual_seed(args.seed)
if model_answer in KNOWN_TIMEOUT_PS:
runner.timed_out.append(model_answer)
runner.timed_out_count += 1
continue
print(model_answer)
synth.set_specs(examples)
# synth.synthesize(examples, problem['prediction'])
# runner.solve(synth, examples, problem['prediction'], model_answer)
# return runner.statistics()
return dict()
def run(p_size, data_set):
runner = TimedSolver()
annotated_problems = get_annotated_problems(p_size, data_set)
dsl = Dsl()
return solve_problems(dsl, annotated_problems, runner)
if __name__ == "__main__":
FLAGS = parse_args()
logger.logger = logger.initialize()
if FLAGS.logfile:
logger.logger = logger.set_logfile(get_log_filename(FLAGS.logfile))
logger.logger.debug(
'Starting script python -m run --program-size=%s --data-set=%s --synthesizer=%s'
% (FLAGS.p_size, FLAGS.set, FLAGS.synthesizer))
statistics = run(FLAGS.p_size, FLAGS.set)
# save_run_statistics(statistics, synthesizer=FLAGS.synthesizer, p_size=FLAGS.p_size)