# isolated=True,
# runs_path=Path(__file__).parent.parent / 'runs',
):
# load hyper-parameter settings
with (param_path / 'param2val.yaml').open('r') as f:
param2val = yaml.load(f, Loader=yaml.FullLoader)
params = Params.from_param2val(param2val)
# re-generate data the way it was during training
world = World(params)
dataset = DataSet(world.generate_sequences(
leftout_colors=params.leftout_colors,
leftout_shapes=params.leftout_shapes,
leftout_variants=params.leftout_variants,
leftout_positions=get_leftout_positions(params.leftout_half),
),
params,
name='re-generated')
# multiple models may exist for the same hyper-parameter configuration
for path_to_net in list(param_path.rglob('model.pt')):
print(f'Loading net from {path_to_net}')
# load net
net = Network(params)
state_dict = torch.load(path_to_net,
map_location=torch.device('cpu'))
net.load_state_dict(state_dict)
net.requires_grad_(False)
isolated=True,
runs_path=Path(__file__).parent.parent / 'runs',
):
# load hyper-parameter settings
with (param_path / 'param2val.yaml').open('r') as f:
param2val = yaml.load(f, Loader=yaml.FullLoader)
params = Params.from_param2val(param2val)
# use all locations, rotations, shapes, and colors - and filter later
world = World(params)
data = DataSet(world.generate_sequences(
leftout_colors=('', ),
leftout_shapes=('', ),
leftout_variants='',
leftout_positions=get_leftout_positions(''),
),
params,
name='re-generated')
if RGB_ID is not None:
name_of_color_channel = {0: 'red', 1: 'green', 2: 'blue'}[RGB_ID]
else:
name_of_color_channel = None
# multiple models may exist for the same hyper-parameter configuration - use first only
for path_to_net in list(param_path.rglob('model.pt'))[:1]:
print(f'Loading net from {path_to_net}')
# load net