def test_mlp_model_output(self):
folder_name, dataset_name, args = generate_dataset_and_parser()
f = io.StringIO()
with redirect_stdout(f):
execute_train(encoder_class=MLPEncoder, decoder_class=MLPDecoder,
encoder_args=dict(hidden_size=5,
layers=2),
decoder_args=dict(hidden_size=5,
layers=2),
args=args)
out = f.getvalue()
# check correct output
assert "Cost multiple" in out and "Final multiple val" in out and "Final loss_val" in out \
and "Final loss_train" in out, 'Wrong output format'
remove_files(folder_name, dataset_name)
def test_cnn_model_output(self):
folder_name, dataset_name, args = generate_dataset_and_parser()
f = io.StringIO()
with redirect_stdout(f):
execute_train(encoder_class=CNNEncoder, decoder_class=CNNDecoder,
encoder_args=dict(readout_layers=1,
channels=4,
layers=2,
kernel_size=3,
non_linearity=True),
decoder_args=dict(readout_layers=1,
channels=4,
layers=2,
kernel_size=3,
non_linearity=True),
args=args)
out = f.getvalue()
# check correct output
assert "Cost multiple" in out and "Final multiple val" in out and "Final loss_val" in out \
and "Final loss_train" in out, 'Wrong output format'
remove_files(folder_name, dataset_name)
from multiple_alignment.steiner_string.models.convolutional.model import CNNEncoder, CNNDecoder
from multiple_alignment.steiner_string.train import execute_train
from multiple_alignment.steiner_string.parser import general_arg_parser
parser = general_arg_parser()
parser.add_argument('--readout_layers', type=int, default=2, help='')
parser.add_argument('--channels', type=int, default=20, help='')
parser.add_argument('--layers', type=int, default=2, help='')
parser.add_argument('--kernel_size', type=int, default=3, help='')
parser.add_argument('--non_linearity', type=bool, default=False, help='')
args = parser.parse_args()
execute_train(encoder_class=CNNEncoder, decoder_class=CNNDecoder,
encoder_args=dict(readout_layers=args.readout_layers,
channels=args.channels,
layers=args.layers,
kernel_size=args.kernel_size,
non_linearity=args.non_linearity),
decoder_args=dict(readout_layers=args.readout_layers,
channels=args.channels,
layers=args.layers,
kernel_size=args.kernel_size,
non_linearity=args.non_linearity),
args=args)