quettabit/pytorch_mtl

Multi-task learning of Abstractive Summarization with Entailment Generation implemented using PyTorch

Information and Instructions about the code

• To run the code, Python 3.6 and PyTorch 0.2 are needed
• The training code is present in the file mtl_learning.py whereas the testing code is present in the file mtl_testing.py
• The code in mtl_learning.py is pretty much self documented
• For the training, the datasets should be present as pickle files in the pickles/ directory. The embeddings file should be present in the data/ directory.
• In case of reloading the models and continuing the training, the models should be placed in reloads/ directory.
• When all the necessary files and data are present, simply run python mtl_learning.py for training and python mtl_testing.py for testing.
• The outputs of the training would be present in a file named outputs.txt and that of testing would be present in test_output.txt

Credits

The project is inspired from Pasunuru et al's (2017) work on "Towards Improving Abstractive Summarization via Entailment Generation" and Sean Robertson's tutorial on seq2seq translation.