run_dipo

#!/bin/bash

# Script to reproduce results

envs=(Hopper-v3 Walker2d-v3 Ant-v3 HalfCheetah-v3 Humanoid-v3)
steps=(1000000 1000000 3000000 3000000 10000000)
cnt=0
i=3
n_timesteps=100
for ((j=0;j<5;j+=1))
do
	nohup python -u main.py \
	--env_name ${envs[i]} \
	--num_steps 1000000 \
	--policy_type 'MLP' \
	--beta_schedule 'cosine' \
	--n_timesteps  ${n_timesteps}\
	--ratio 0.08 \
	--ac_grad_norm 2 \
	--action_gradient_steps 40 \
	--update_actor_target_every 2 \
	--seed $j \
	--cuda "cuda:${cnt}" \
	> "log/MLP-a_steps=40-%2-${envs[i]}-seed=${j}.log" 2>&1 &
done