/molecule_grammar_rnn

Primary LanguagePython

  1. Install rdkit (chemical informatics, http://www.rdkit.org/docs/Install.html) using Anaconda.

  2. Install pytorch. The project uses pytorch 0.3.1, later versions are not backward compatible and will not run.

  3. First prepare data by running python -m data data/zinc.smi (Generate samples from pretrained VAE for testing the grammar model)

  4. To train your own VAE. python -m vae --restore vae-pretrained --generate-samples

  5. Train grammar_rnn model, the main paper contribution. This takes approximately ten minutes. python -m grammar_model