Skip to content

Latest commit

 

History

3 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

VAE-Transformer Retrosynthesis

A PyTorch-based VAE-Transformer model for predicting reactants from products in chemical reactions. Includes SMILES tokenization, latent-space encoding, and autoregressive decoding. Supports sampling, checkpointing, and experiment tracking with WandB.


Features

  • VAE-Transformer architecture with continuous latent space.
  • Autoregressive sampling with argmax decoding for retrosynthesis.
  • SMILES tokenization with halogen simplification ([Cl] → L, [Br] → R) and restoration.
  • WandB integration for training logs, learning rate, and model checkpoints.
  • Checkpoint loading for fine-tuning or inference.
  • Supports long SMILES sequences (up to 1300 tokens).

Note

The dataset has been compressed for easier upload.

  • Dataset: uspto.zip (contains uspto.csv)

About

A VAE-Transformer-based retrosynthesis prediction model that generates reactant SMILES sequences from a given product SMILES.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages