PyTorch implementation of TD3 with spine-SNN actor network. The code based on the paper "Spine encoding improved spiking neural network for efficient reinforcement learning".
Method is tested on MuJoCo continuous control tasks in OpenAI gym. Networks are trained using PyTorch 1.2 and Python 3.7.
We use the code TD3 of Fujimoto2018 paper.
Experiments on single environments can be run by calling:
python main.py --env=HalfCheetah-v2 --network=EpspAdeepC
Hyper-parameters can be modified with different arguments to main.py. The configuration about the experiments is outline in the paper.
Experiment in the paper such as the influence of spine encoder dim and the comparison between different actor network could reproduce with this code.