Conversation
| # torch-geometric | ||
| # torch_scatter | ||
| "black", | ||
| "descriptastorus", |
There was a problem hiding this comment.
descriptastorus is needed for creating datasets, right? I would put that under dependencies (as it is not just used for development)
There was a problem hiding this comment.
The main dependencies should have only packages required to inference of the model. Hence, "descriptastorus" is placed under optional dependency as its not needed for inference but used during dataset creation.
There was a problem hiding this comment.
Then we should follow the approach from #114 and differentiate between training and linters
|
I made some unrelated changes to the readme, descriptatorus is added in PR #26 |
changes as per ChEB-AI/python-chebai#114 (comment)