- torch
- torch_geometric
- rdkit
- pytorch_lightning
Run the script with python pretrain.py [DATA_FILE]
The [DATA_FILE]
has to be a csv file with column labeled smiles
, which will contain the SMILES representation of the molecules
Model data will be logged to tb_logs
folder (including checkpoints).
Run the script with python finetune.py [DATA_FILE] --pretrained_model_file [PRETRAIN_CHECKPOINT]
The [DATA_FILE]
should contain the smiles
column, as well as a y
column, which will label be used as the target value
In case of classification problem (y
denotes the class) use the --task class
argument, in case of regression --task reg