🚀 Describe the improvement or the new tutorial

In current tutorial DQN is not performing well even after 600 episodes. I would like to update hyper parameters, so it reaches optimal policy by the end of training.

Following parameters needs to change and need to replace the Result image.

EPS_START = 1.0
EPS_END = 0
EPS_DECAY = 1500
LR = 5e-4

Existing tutorials on this topic

No response

Additional context

cc @vmoens @nairbv