Code and dataset for paper "Modular Tree Network for Source Code Representation Learning".
For the code classification task:
1.Download the 293-classes code classification dataset at https://drive.google.com/file/d/1e5gX6ulVVMKC3g1tjk_7JdOEdnTn13F6/view?usp=sharing and unzip it.
2.Run mtn_b.py.
Currently we only provide the code of the MTN-b model. For MTN-a model, simply share the weights between different Tree-LSTM cells.