DeepChem Minutes 4/03/2020

Date: 4/03/2020
Attendees: Bharath, Peter, Vignesh, Seyone
Summary: Bharath started working on cleaning up the examples (PR). This is blocked by the open TF2.x PR, which still has a few more issues that need to be resolved. Bharath is planning on making a push on that PR to try to get it through the door.

Peter’s been working on upgrading the book example code to DeepChem 2.4 (PR). This is basically done, with the exception of the graph convolutional models, which are blocked on the TF2.X PR

Vignesh will be working on submitting his thesis for the next two weeks, but will likely have a bit of time after. Bharath and Vignesh briefly chatted about using the transfer learning framework building off Vignesh’s GSoC project last summer (link) on other datasets. Vignesh said it should be possible to get it working, and that he might be able to take a look at it if there are any processing issues that crop up. Bharath is hoping to put together a transfer learning colab notebook over the next few week’s

Seyone has been continuing work on building NLP models. He’s got a colab notebook with a model that’s capable of doing mask token prediction on SMILES strings up and running. He’s also been looking at some assay data for some covid19 targets.

1 Like