BabyBench

Datasets

The Decompose, Plan, and Predict datasets are available in the datasets folder, along with a clear description.

This dataset is generated using the scripts in the generate folder.

Create and Activate Virtual Environment

To begin, create a Python virtual environment and activate it.

1. Create the virtual environment:

python -m venv babybench

2. Activate the environment:

source babybench/bin/activate

3. Install the requirements:

pip install -r requirements.txt

Install Minigrid

Next, clone the Minigrid repository and install the necessary dependencies.

1. Clone the Minigrid repository:

git clone https://github.com/Farama-Foundation/Minigrid.git

2. Checkout the specific commit:

After cloning the repository, navigate into the project directory and checkout the commit hash:

cd Minigrid
git checkout 6e713afef8d23d5280ebf28fb3fcf635d40d6a7f

This ensures that you are using the exact version of this repository.

3. Install Minigrid in editable mode:

In the same directory, run:

python3 -m pip install -e .