Skip to content

We propose a method to integrate Soft Actor-Critic Algorithm with expert offline demonstrations and sample with Prioritized experience replay to achieve faster learning and a more stable training process.

Notifications You must be signed in to change notification settings

sai-aneesh/Enhancing-SAC

Repository files navigation

Enhancing SAC with Offline demonstrations and Prioritized Experience Replay

Result

Environment setup

Use the following commands to setup the environment in your local system.

conda create -n cs276f python=3.10.12
conda activate cs276f
pip install --upgrade mani_skill
conda install pytorch==2.2.1 torchvision==0.17.1 torchaudio==2.2.1 pytorch-cuda=12.1 -c pytorch -c nvidia
pip install stable-baselines3[extra]
pip install tyro wandb chardet

To test the setup run the following command.

python -m mani_skill.examples.demo_random_action

To Run

Head to the main.ipynb and run each cell to reproduce the results.

About

We propose a method to integrate Soft Actor-Critic Algorithm with expert offline demonstrations and sample with Prioritized experience replay to achieve faster learning and a more stable training process.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published