megnet.config module

Data types

class DataType[source]

Bases: object

Data types for tensorflow. This enables users to choose from 32-bit float and int, and 16-bit float and int

np_float

alias of numpy.float32

np_int

alias of numpy.int32

classmethod set_dtype(data_type: str) None[source]

Class method to set the data types :param data_type: ‘16’ or ‘32’ :type data_type: str

tf_float = tf.float32
tf_int = tf.int32
set_global_dtypes(data_type) None[source]

Function to set the data types :param data_type: ‘16’ or ‘32’ :type data_type: str

Returns: