Arun

Two pointers: - After training, the encoder and decoder loss should start reducing because the generator starts to learn a good mapping. - The loss for critic should be increasing...

Encoder and decoder loss should decrease as they learn a better mapping - I observed it in training log file. I am not sure what is wrong in your case,...

Hey, thanks for bringing this up. Your understanding that `fp_size` equals to `dense_layer_size` is correct. I think its a bug in the `deepchem.layers.GraphGather` layer. In the `call` method in GraphGather...

MPNNModel has been migrated to pytorch earlier([ref](https://github.com/deepchem/deepchem/blob/master/deepchem/models/torch_models/mpnn.py)).

Sorry for the delayed response but thanks for bringing this up @ismorphism @bananenpampe. The error can be reproduced in the following snippet: ``` import deepchem as dc import numpy as...

@ARY2260 is this ready for review?

The issue should not occur in the lightly version. Can you please try using the nightly version, `pip install --pre deepchem` and check import? @mcloughlin2 atom_init.json will be downloaded only...

Well, that's mysterious. Ideally, it should proceed fine because the download part is in the try except block, https://github.com/deepchem/deepchem/blob/f06d2d6b1add4ad9573b0b3779a22d3833e86614/deepchem/feat/material_featurizers/cgcnn_featurizer.py#L78-L82 I will take a look at it later this week!

`dataset` has to be NumpyDataset. ``` import numpy as np X = np.random.rand(10, 10) dataset = dc.data.NumpyDataset(X=X) dataset.to_csv('out.csv') ``` If you want to use `to_csv` in DiskDataset, you should have...

> It seems that you removed or didn't commit some of the needed generated files and that's why CI isn't passing. `ta` was not supported as a language in jinja...