metavar='S',
help='random seed (default: 42)')
parser.add_argument(
'--log-interval',
type=int,
default=10,
metavar='N',
help='how many batches to wait before logging training status (default: 10)'
)
args = parser.parse_args()
if args.classifier: args.idx_label = None
args.betas = tuple(np.array(args.betas, dtype=float))
nu.train_rnn(file_name=args.file_name,
hidden_size=args.hidden_size,
n_layers=args.n_layers,
bidirectional=args.bidirectional,
classifier=args.classifier,
idx_label=args.idx_label,
n_epochs_max=args.epochs,
train_ratio=args.train_ratio,
batch_size=args.batch_size,
n_workers=args.n_workers,
root_dir=args.root_dir,
lr=args.lr,
betas=args.betas,
opt_level=args.opt_level,
seed=args.seed,
log_interval=args.log_interval)
'''
>>>>>>> 18a8402a9b721b66ed28bcee5beabe6256a96076
# train RNNs for scat transformed data
for file_name_scat in file_names_scat:
meta = torch.load(os.path.join(root_dir, file_name_scat))
avg_len = meta['avg_len']
n_filter_octave = meta['n_filter_octave']
for hidden_size in hidden_sizes:
for n_layers in n_layerss:
for bidirectional in bidirectionals:
try:
print("training rnn for {}, avg_len:{}, n_filter_octave:{}, hidden_size:{}, n_layers:{}, bidirectional:{}"
.format(file_name_scat, avg_len, n_filter_octave, hidden_size, n_layers, bidirectional))
nu.train_rnn(file_name_scat, [hidden_size, hidden_size], n_layers, bidirectional, classifier=False,
n_epochs_max=n_epochs_max, train_ratio=train_ratio, batch_size=batch_size,
n_workers=n_workers, root_dir=root_dir)
except:
print("exception occurred for {}, avg_len:{}, n_filter_octave:{}, hidden_size:{}, n_layers:{}, bidirectional:{}"
.format(file_name_scat, avg_len, n_filter_octave, hidden_size, n_layers, bidirectional))
# train RNNs for raw data
for file_name_data in file_names_data:
for hidden_size in hidden_sizes:
for n_layers in n_layerss:
for bidirectional in bidirectionals:
try:
print("training rnn for {}, hidden_size:{}, n_layers:{}, bidirectional:{}".format(file_name_data, hidden_size, n_layers, bidirectional))
nu.train_rnn(file_name_data, [hidden_size, hidden_size], n_layers, bidirectional, classifier=False,
n_epochs_max=n_epochs_max, train_ratio=train_ratio, batch_size=batch_size,
n_workers=n_workers, root_dir=root_dir)
except:
print("exception for avg_len:{}, n_filter_octave:{}".format(avg_len, n_filter_octave))
# train RNNs for scat transformed data
for file_name_scat in file_names_scat:
meta = torch.load(os.path.join(root_dir, file_name_scat))
avg_len = meta['avg_len']
n_filter_octave = meta['n_filter_octave']
for hidden_size in hidden_sizes:
for n_layers in n_layerss:
for bidirectional in bidirectionals:
try:
print("training rnn for {}, avg_len:{}, n_filter_octave:{}, hidden_size:{}, n_layers:{}, bidirectional:{}"
.format(file_name_scat, avg_len, n_filter_octave, hidden_size, n_layers, bidirectional))
nu.train_rnn(file_name_scat, hidden_size, n_layers, bidirectional, classifier=True,
n_epochs_max=n_epochs_max, train_ratio=train_ratio, batch_size=batch_size,
n_workers=n_workers, root_dir=root_dir, lr=lr, betas=betas)
except:
print("exception for file_name_scat:{}, hidden_size:{}, n_layers:{}, bidirectional:{}".format(file_name_scat, hidden_size, n_layers, bidirectional))
# train RNNs for raw data
for hidden_size in hidden_sizes:
for n_layers in n_layerss:
for bidirectional in bidirectionals:
try:
print("training rnn for {}, hidden_size:{}, n_layers:{}, bidirectional:{}".format(file_name_data, hidden_size, n_layers, bidirectional))
nu.train_rnn(file_name_data, hidden_size, n_layers, bidirectional, classifier=True,
n_epochs_max=n_epochs_max, train_ratio=train_ratio, batch_size=batch_size,
n_workers=n_workers, root_dir=root_dir, lr=lr, betas=betas)
except:
print("exception for file_name_data:{}, hidden_size:{}, n_layers:{}, bidirectional:{}".format(file_name_data, hidden_size, n_layers, bidirectional))