def __init__()

in vihds/config.py [0:0]

• 31 lines of code
• 9 McCabe index (conditional complexity)

    def __init__(self, args):
        args = _tidy_args(args)
        if args.yaml is None:
            return None
        with open(args.yaml, "r") as stream:
            config = munchify(yaml.safe_load(stream))
            # return Munch.fromYAML(stream)
        self.data = apply_defaults_data(config.data)
        # self.models = config.models
        # self.experiments = {}
        # for node, data_settings in config.experiments.items():
        #     self.experiments[node] = apply_defaults_data(data_settings)
        self.params = apply_defaults_params(config.params)
        if args.precision_hidden_layers is not None:
            self.params.n_hidden_decoder_precisions = args.precision_hidden_layers
        self.model = config.model
        self.seed = args.seed
        if (args.gpu is not None) & torch.cuda.is_available():
            print("- GPU mode computation")
            self.device = torch.device("cuda:" + str(args.gpu))
            if self.data.dtype == "float32":
                torch.set_default_tensor_type("torch.cuda.FloatTensor")
            elif self.data.dtype == "float64":
                torch.set_default_tensor_type("torch.cuda.DoubleTensor")
            else:
                raise Exception("Unknown dtype %s" % self.data.dtype)
        else:
            print("- CPU mode computation")
            self.device = torch.device("cpu")
            if self.data.dtype == "float32":
                torch.set_default_tensor_type("torch.FloatTensor")
            elif self.data.dtype == "float64":
                torch.set_default_tensor_type("torch.DoubleTensor")
            else:
                raise Exception("Unknown dtype %s" % self.data.dtype)
        self.trainer = None  # Trainer(args, log_dir=log_dir, add_timestamp=True)