config.restart = True
# copy over new num_epochs and lr schedule
config.num_epochs = given_config.num_epochs
config.lr_schedule = given_config.lr_schedule
if not hasattr(config, "kmeans_on_features"):
config.kmeans_on_features = False
else:
print("Config: %s" % config_to_str(config))
# Data, nets, optimisers -------------------------------------------------------
dataloader_original, dataloader_positive, dataloader_negative, \
dataloader_test = make_triplets_data(config)
train_dataloaders = [
dataloader_original, dataloader_positive, dataloader_negative
]
net = archs.__dict__[config.arch](config)
if config.restart:
model_path = os.path.join(config.out_dir, "latest_net.pytorch")
taking_best = not os.path.exists(model_path)
if taking_best:
print("using best instead of latest")
model_path = os.path.join(config.out_dir, "best_net.pytorch")
net.load_state_dict(
torch.load(model_path, map_location=lambda storage, loc: storage))
assert config.model_ind == given_config.model_ind
config.restart = True
# copy over new num_epochs and lr schedule
config.num_epochs = given_config.num_epochs
config.lr_schedule = given_config.lr_schedule
if not hasattr(config, "kmeans_on_features"):
config.kmeans_on_features = False
else:
print("Config: %s" % config_to_str(config))
# Data, nets, optimisers -------------------------------------------------------
dataloader_original, dataloader_positive, dataloader_negative, dataloader_test = make_triplets_data(
config)
train_dataloaders = [
dataloader_original, dataloader_positive, dataloader_negative
]
net = archs.__dict__[config.arch](config)
taking_best = None
if config.restart:
model_path = os.path.join(config.out_dir, "latest_net.pytorch")
taking_best = not os.path.exists(model_path)
if taking_best:
print("using best instead of latest")
model_path = os.path.join(config.out_dir, "best_net.pytorch")
net.load_state_dict(