DeepChem Minutes
India/Asia/Pacific Call
Date: April 8th, 2021
Attendees: Bharath, Ninad
Summary: Ninad did his undergrad at IIT Bombay and is doing his masters at ANU working to predict new aluminum alloys. Ninad is interested in participating in GSoC this year.
Ninad had questions about his GSoC proposal and discussed it with Bharath.
Americas/Europe/Africa/Middle East
India/Asia/Pacific Call
Date: April 9th, 2021
Attendees: Bharath, Vignesh, Peter, Seyone, Nathan, Andrea
Summary: Bharath reviewed proposals and reviewed PRs this week.
Vignesh was busy this week but has been iterating on the deepchem GSoC proposal.
Peter finished up the PR to make github actions continue running even if one test fails. He has been updating the book to make the code match the latest version of DeepChem.
Stanley has gotten access to a anyscale cluster at work and has been working more with ray. Stanley also worked a bit with PyTorch Lightning and converted them to use multi-GPU
Seyone has been working on a sequence classification model for ChemBERTa and an improved fine-tuning script. Seyone has also been discussing with Ashwin on reaction modeling.
Nathan has been working on wrapping up the pdbbind benchmarking work and has also been working with applicants. There are some folks interested in differentiable programming
Andrea this week was busy with other work, but still working on coming up to speed on DeepChem.
David is working with protein sequences and protein 3D structures.
With the roundtable done, we moved to the general discussion.
Andrea pointed out that the installation instructions on the website currently point to 2.3.0 and we should get that updated to the latest DeepChem. Andrea pointed out that there are multiple installation differences that make things tricky and noted that the website currently only lists TensorFlow installation instructions and not not PyTorch. Bharath agreed and said we should get that integrated into the release notes
Andrea also asked about mlflow and deepchem integration and more generally about MLOps considerations when using DeepChem. Andrea suggested this could be the focus of a new DeepChem tutorial.
Joining the DeepChem Developer Calls
As a quick reminder to anyone reading along, the DeepChem developer calls are open to the public! If you’re interested in attending either or both of the calls, please send an email to X.Y@gmail.com, where X=bharath, Y=ramsundar.