Exploring the heterogeneity-span tradeoff
The Heterogeneity Project explores the role of heterogeneity in neural network architectures by exploring diverse activation functions and optimisation algorithms. We want to understand why biology often come with tin, highly heterogeneous networks solutions that require big population for their homogeneous counterparts to solve.
- Python 3.11
- NEAT-Python (@Finebouche branch)
- Farama Gymnasium
-
Clone the repository:
git clone https://github.com/Finebouche/neat-gymnasium.git cd heterogeneous -
Install the required packages:
conda env create -f environment.yml
Run heterogeneity using the following command:
Using conda
❯ conda activate het_envThis project is protected under the MIT License License.
- List any resources, contributors, inspiration, etc. here.