def load_model(cls, ensemble_path):
"""Load up ensembled models given a folder location."""
json_file = "{}_metadata.json".format(
os.path.join(ensemble_path, os.path.basename(ensemble_path)))
with open(json_file, 'r') as file:
config = json.load(file)
networks = []
for model_file in sorted(os.listdir(ensemble_path)):
if model_file.endswith('.network'):
file = os.path.join(ensemble_path, model_file)
networks += [Network.load_model(file)]
snap = Snapshot(name='snap1',
template_network=networks[0],
n_snapshots=config[ensemble_path]['n_snapshots'])
snap.networks = networks
return snap
if __name__ == "__main__":
args = parse_args()
update_config(cfg, args)
test_set = eval(cfg.DATASET.DATASET)("valid", cfg)
num_classes = test_set.get_num_classes()
device = torch.device("cpu" if cfg.CPU_MODE else "cuda")
model = Network(cfg, mode="test", num_classes=num_classes)
model_dir = os.path.join(cfg.OUTPUT_DIR, cfg.NAME, "models")
model_file = cfg.TEST.MODEL_FILE
if "/" in model_file:
model_path = model_file
else:
model_path = os.path.join(model_dir, model_file)
model.load_model(model_path)
if cfg.CPU_MODE:
model = model.to(device)
else:
model = torch.nn.DataParallel(model).cuda()
testLoader = DataLoader(
test_set,
batch_size=cfg.TEST.BATCH_SIZE,
shuffle=False,
num_workers=cfg.TEST.NUM_WORKERS,
pin_memory=cfg.PIN_MEMORY,
)
valid_model(testLoader, model, cfg, device, num_classes)
