Actually, LSTM's binary format (*.npy) depends on TensorFlow. LSTM team suggests that we can create other format (see https://github.com/unicode-org/lstm_word_segmentation/issues/16). So I should use rust data type or any format instead of JSON via build.rs.
Actually, LSTM's binary format (*.npy) depends on TensorFlow. LSTM team suggests that we can create other format (see https://github.com/unicode-org/lstm_word_segmentation/issues/16). So I should use rust data type or any format instead of JSON via
build.rs
.