/aizynthfinder-deploy

modify original aizynthfinder code for end-user

Primary LanguagePythonMIT LicenseMIT

Modify Aizynthfinder for end-users

  • remove training module

  • remove other algorithms other than mcts

  • fix tying error and refactor context module, etc.

  • add script for load stock data to database to reduce memory usage.

How to use this package

  • python >= 3.10

  • download trained model from original code base

    • uspto_filter_model.onnx // model for filter policy
    • uspto_model.onnx // model for tree generator
    • uspto_templates.csv.gz // templates data
    • zinc_stock.hdf5 // stock data

  • install packages

    • numpy, pandas, matplotlib, rdkit, networkx

  • run usage.py to check results.

TODO

  • rewrite training module with pytorch
  • modify AI models