xnn.dnn.common.runner.load_runner_model#

xnn.dnn.common.runner.load_runner_model(directory, input_nn='input.nn', dtype=torch.float64)[source]#

Build an HDNNP from a RuNNer model directory.

Parameters:

directory (str or pathlib.Path) – Directory with input.nn, the weights and the scaling files (one committee member).
input_nn (str, optional) – Name of the settings file, by default "input.nn".
dtype (torch.dtype, optional) – Parameter dtype, by default float64.

Returns:

The model in xnn units (Angstrom, eV), in eval mode.

Return type:

HDNNP