:param l: list of all the parameters value (changeable and not changeable ones) :param name: the name of the dimension :retu

Get the tuple from the string given by the user :param separator: separator of the value in the string :param t: float, int or other type

:param name: :param value: :param exp_format: :param first_printed: :return: string which is pretty to print (to show current hp

:param start_time: :param on_time_step_begin_callbacks: :param on_time_step_end_callbacks: :param on_step_begin_call

:param inputs_shape: (batch, nb_steps, size) :return:

Add Categorical dimension to the dimensions :param m_tuple: :param name: :return:

Add the Real dimension to the dimensions But if the parameter is fixed, it only adds a categorical dimension with 1 choice :p

:param argtype: :param t: :return:

Used to set the weights of a sublayers in a layers Since Tensorflow 2.0 it works without this function :param args:

:param input_shape: (batch, nb_steps, input_size) :return:

to filter messy output if there are some :param x: :param args: :param kwargs: :return:

:param params: :return:

Add to the parser a store true action :param self: :param name: :param argtype: :param help: :return:

Used to reset the weights of a sublayers in a layers Since Tensorflow 2.0 it works without this function :return:

:param weight: :param sum_axis: To sum through time. If is None, no summation is done :param kwargs:

Used to add the weights of a sublayers in a layers Since Tensorflow 2.0 it works without this function :param args:

This function fix the issue which comes from some note's duration. For instance some note has duration like 14/3 or 7/3. :param duration:

:param path: :return:

Used to evaluate the boolean written in the string :param string: :return:

:param args: :return:

:param path: :return:

Get an attribute off the `socket` module, or use a default.

:param bidirectional: :param size: :param args: :param dropout: :param return_sequences: :param kwar

:param input_shape: (batch, nb_steps, input_size) :return:

:param input_shape: (batch, nb_steps, size) :return:

:param x: :param env: :return:

:param name: The index of the parameters in the list of changeable parameters :return:

:param encoder_param: { 'conv': [[4, 8], [16, 8], [4, 8]], 'dense': [24, 12] } :param dropout_c: flo

:param axis: No batch in axis

:param encoder_param: { 'conv': [[4, 8], [16, 8], [4, 8]], 'dense': [24, 12] } :param dropout_c: flo

:param num_or_size_to_split: Union[int, List[int]]: :param axis: axis to split the tensor : int: ⚠ axis=0 correspond to

:param filters: int: the size of the filters :param strides: tuple<int>: (3,): :param dropout: float:

:param input_shape: List(2)[ (batch, modalities, nb_steps, size) ] + List[1](batch, modal

Before TF2, it was necessary to fit the model before fit_generator But now it works :param generator: :param args:

:param array: (instruments, size, length, channels) :param n: :return: (length,)

:param length: :return:

:param tensor: activation, (batch, nb_instruments, nb_steps, step_size, input_size) (no mono silent note) :param interval: :return:

:param note: :param notes_range: :return:

:param path: :return:

:param layer: a Keras layer :param args: :param kwargs:

:param array: :return:

:param step_length: :param notes_range: To add the piano on the left of the image :param colors: :param array: (nb_instruments, inpu

if batch dim is not in the shape, then put it :param final_shape: tuple<int>: (4,) or (5,): (a, b, c, d) or (?, a, b, c, d)

:param sequence_name: :param kwargs: :return:

:return: