The training oder

Prefix for Cluster

srun --partition=ad_ap --gres=gpu:1 -n1 --ntasks-per-node=1

Step 0 Generating the data for VAE training

python -m Generating_data.carracing --dir datasets/carracing --rollouts 1000 --policy brown

Step 1 Training VAE

python trainvae.py --logdir exp_dir

Step 2 Training MDN

python trainmdrnn.py

Step3 Using PPO to train the model end to end

python train.py --algo PPO --env-id cCarRacing-v0 \
--num-envs 12 \
--lr 1e-4 \
--action-repeat 1 \
--entropy 0.0

for Cluster

srun --partition=ad_ap --gres=gpu:1 -n1 --ntasks-per-node=1 python train.py --algo PPO --env-id cCarRacing-v0 \
--num-envs 12 \
--lr 1e-4 \
--action-repeat 1 \
--entropy 0.0