error after executing initial training
Opened this issue · 0 comments
dzako commented
Hi, thanks for the nice implementation of DyNODE, however after executing the provided example
CUDA_VISIBLE_DEVICES=0 python train_dynode.py --domain_name cartpole --task_name swingup --seed -1 --critic_lr 1e-3 --actor_lr 1e-3 --eval_freq 5000 --batch_size 128 --num_train_steps 100001
I get the following error
Traceback (most recent call last):
File "train_dynode.py", line 203, in <module>
main()
File "train_dynode.py", line 183, in main
agent.update_model(replay_buffer, L, step)
File "/home/.../repos/DyNODE/dynode.py", line 76, in update_model
obs_m, action_m, reward_m, next_obs_m, _ = replay_buffer.sample_dynode()
File "/home/.../repos/DyNODE/utils.py", line 99, in sample_dynode
((i+1)*self.episode_length)-self.batch_length) for i in range(inter)])
ValueError: need at least one array to concatenate