Visual novelty datasets for the Polycraft domain.
If you do not have Pipenv installed, run the following:
pip install pipenv
The Pipenv dependencies can be installed within a Pipenv with the following commands:
pipenv install
Pytorch requires different versions depending on the machine it is running on. Therefore it is not included in the Pipenv by default. To install Pytorch, generate the appropriate Pip command on the Pytorch website then run it within the Pipenv by prepending pipenv run
to it. For example:
pipenv run pip install torch==1.7.1+cpu torchvision==0.8.2+cpu torchaudio===0.7.2 -f https://download.pytorch.org/whl/torch_stable.html
The following instructions are not recommended unless you are unable to install a Python version compatible with the Pipenv.
For this installation Pytorch will be installed in the Conda environment using the appropriate command according to the Pytorch website. For example:
conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch
Then the Pipenv must be configured for the Conda environment before install:
pipenv --python=$(conda run which python) --site-packages
pipenv install
To run unit tests, run the following command:
pipenv run python -m pytest
-
Create a folder within either the
normal_data
ornovel_data
folder in thedatasets
folder of the SAIL-ON Box. Ensure the folder has a name that clearly describes how it is different from other data. -
Upload the data to the Box folder.
-
Create a folder in the
normal_data
ornovel_data
folder of this repository with the same name as the Box folder. -
Copy
data_readme_template.md
to the new folder asREADME.md
. ModifyREADME.md
with a title, description, and reproducable steps to generate the data.