def class_process_params(output_size = 1, weights = None, biases = None): input_size = len(cf['alphabet']) hidden_size = cf['init_hidden'] size = hidden_size + output_size if weights is None: weights = np.mat(matnn.genentry((size,size+input_size))) if biases is None: biases = np.mat(matnn.genbias((size, 1))) return (output_size, weights, biases)
def gen_process_params(output_size = None, weights = None, biases = None, gen_input = None): if gen_input is None: gen_input = matnn.randomInput if output_size is None: output_size = len(cf['alphabet']) hidden_size = cf['init_hidden'] size = hidden_size + output_size if weights is None: weights = np.mat(matnn.genentry((size,size+cf['gen_insize']))) if biases is None: biases = np.mat(matnn.genbias((size, 1))) return (output_size, weights, biases, gen_input)