Training and visualizing of diffusion models from Reasoning with Latent Diffusion in Offline Reinforcement Learning (NeurIPS 2023).
pip install --upgrade pip
pip install -r requirement.txt
tqdm
matplotlib
wandb
ipdb
arcle == 0.2.5
추가하기!
Training Code for ARCLE Environment
cd training
- Training skill with:
./gpu0_train_1_skill_model.sh
- Collect data to train diffusion model with:
./gpu0_train_2_collect_diffusion_data.sh
- Training diffusion model with:
./gpu0_train_3_diffusion.sh
- Collect data to train offline Q-learning with:
./gpu0_train_4_collect_q_learning.sh
- Training Q-network with:
./gpu0_train_5_q_learning.sh
cd ../eval/
./gpu0_test_ARCLE.sh
@inproceedings{ldcq,
title = {Reasoning with Latent Diffusion in Offline Reinforcement Learning},
author = {Siddarth Venkatraman, Shivesh Khaitan, Ravi Tej Akella, John Dolan, Jeff Schneider, Glen Berseth},
booktitle = {Conference on Neural Information Processing Systems},
year = {2023},
}