max_encoder_seq_length = max([len(txt) for txt in input_texts])
encoder_input_data = np.zeros( (len(input_texts), max_encoder_seq_length, num_encoder_tokens), dtype='float32')
要指定dtype=float32
