Sulstice
commented
3 months ago @thedevp
We are going to make a DVC pipeline for the training of the data.
1.) Take the directory narcotics and place it into it's own repository under Global-Chem: https://github.com/Global-Chem/private-workers/tree/master/machine_learning_operations/narcotics
2.) Edit the https://github.com/Global-Chem/private-workers/blob/master/machine_learning_operations/narcotics/mol.smi to include molesules only from Pihkal.
We will be using REINVENT4: https://github.com/MolecularAI/REINVENT4 on the narcotics pihkal list.
1.) Train the model with the Pihkal book as the inputs 2.) Check the training using a visual output of generated compounds 3.) Integrate it into discord.
Read these two papers:
1.) https://arxiv.org/abs/1704.07555 2.) https://link.springer.com/article/10.1186/s13321-024-00812-5?utm_source=rct_congratemailt&utm_medium=email&utm_campaign=oa_20240221&utm_content=10.1186/s13321-024-00812-5