Evaluating RL in Continual Learning

This project evaluates the behaviour of RL algorithms in continual learning settings.

Dependencies

1. Create and activate virtual environment:

   • With conda:
     • conda create -n cl-parking python=3.10
     • conda activate cl-parking
   • Alternatively with venv (requires python 3.10):
     • python -m venv .venv
     • source .venv/bin/activate

2. Install tools needed to build and install dependencies:

   • pip install --upgrade pip
   • pip install packaging
   • pip install setuptools==69.5.1 wheel

3. Install dependencies in the following order:

   • pip install numpy==1.24.3
   • pip install torch==2.2.1 torchvision==0.17.1 torchaudio==2.2.1 --index-url https://download.pytorch.org/whl/cu121
   • pip install causal-conv1d==1.5.0.post8 --no-build-isolation
   • pip install mamba-ssm==1.2.0.post1 --no-build-isolation
   • pip install -r requirements.txt