It seems that DQN performs the worst if trained for 1e6 timesteps. But it did train quicker, taking about 17 min, as opposed to 20-22.

3f14e63 over 3 years ago

201 Bytes

	OS: Linux-5.4.0-135-generic-x86_64-with-glibc2.31 #152-Ubuntu SMP Wed Nov 23 20:19:22 UTC 2022
	Python: 3.9.16
	Stable-Baselines3: 1.6.2
	PyTorch: 1.13.0+cu117
	GPU Enabled: True
	Numpy: 1.23.5
	Gym: 0.21.0