A Recurrent Neural Network to generate High Energy Theory abstracts (https://dlvp.github.io/Physics-snnarxiv/)
I uploade all the notebook to train the RNN and use it to generate text. On the other hand neither the corpus, nor the trained parameters for the RNN are attached below. For those, contact me privately.
arxiv_abstracts.ipynb: notebook for preprocessing of titles and abstracts and extraction of formulas
abstract_word_rnn.ipynb: notebook that contains the TensorFlow code to train the RNN that will generate abstracts
title_word_rnn.ipynb: notebook that contains the TensorFlow code to train the RNN that will generate titles
abstract_word_rnn_gen.ipynb: notebook which used the trained RNN to generate abstracts
title_word_rnn_gen.ipynb: notebook which used the trained RNN to generate titles
abstracts_out: sample of generated abstracts
titles_out: sample of generated titles
generate_tex.py: randomly picks an abstract from abstracts_out and a title from titles_out and output a compilable tex file