Arun

> Can we add a small utility to so that it prints the graph more useful @neerajanand-coder , will you be interested in working on this?

@peastman I have a question on using `GraphData` with pytorch models. Since GraphData stores node_features and others as numpy arrays, to pass it to a pytorch model, should we convert...

> Do we want to create a mechanism where data can get preprocessed into a form that's more efficient for a particular model? Yes, because for each epoch we are...

> Is there a reason this is still a draft? Do you need help resolving the failing `doctest`? When we we made the PR, some of our required dependencies did...

We are having some package conflicts for 3.10 (conflicts [here](https://github.com/deepchem/deepchem/runs/6613218209?check_suite_focus=true#step:10:787)) @Seanny123 . I will try to take a look over the weekend or some time next week and find the...

@Seanny123 we were able to fix the conflicts for rdkit and mdtraj by using pip for installation but we have a similar issue of conda conflicts with hhsuite, openmm, pdbfixer...

> For `hhsuite` you may need to add the `bioconda` channel to `conda`? I looked at the recipe and there doesn't seem to be any version restrictions. We have the...

I updated the deepchem core CI system to use mamba. In [this](https://github.com/deepchem/deepchem/runs/6860385048?check_suite_focus=true#step:11:81) step of ci, it seems that hhsuite is causing a problem in dependency management. On a side note,...

> Is this being actively worked on? If not, I'd like to take a look. Tests in common CI for ubuntu py 3.10 is failing since all dependencies are not...

The `test_hindsight` is a flaky test - it passes sometimes, other time it fails. Some of the other flaky tests we have are `test_wgan_reload`, `test_atomic_conv_feat`, `test_fittransform_regression_overfit` + a few more....