This is source code for the final project of course Reinforcement Learning

Training Instruction

We train all the algorithms on NTY Greene HPC.

sbatch scripts/ppo_train.sh ppo_lr05_ent15_cp2 0

For example, if you want to train the PPO algorithm in NYU HPC with the above command. ppo_lr05_ent15_cp2 means the learning rate $5 \cdot 1e^{-5}$, entropy coefficient $15 \cdot 0.001$, clip range $2\cdot 0.1$. The last parameter 0 means seed is 0.