Chemprop is now available on PyPi: https://pypi.org/project/chemprop. Installation instructions are below.
1. `conda create -n chemprop python=3.8`
2. `conda activate chemprop`
3. `conda install -c conda-forge rdkit`
4.`pip install git+https://github.com/bp-kelley/descriptastorus`
5. `pip install chemprop`
After installing through PyPi, training and predicting are available via the `chemprop_train` and `chemprop_predict` commands, which are equivalent to `python train.py` and `python predict.py`. All the command line arguments for training and predicting apply as usual. Please see the [README](https://github.com/chemprop/chemprop) for more details.