fix reward to one episode

[qrh1]

Rewards should not discounted across different episodes. maybe episodes and steps are confused here?

[titu1994]

for t in reversed(range(0, rewards.size)):
            if rewards[t] != 0:
                running_add = 0
            running_add = running_add * self.discount_factor + rewards[t]
            discounted_rewards[t] = running_add
        return discounted_rewards[-1]

This us the discounted reward, which returns your values anyway.

[qrh1]

hi Somshubra, I still don't get it, could you please expain more?
In my understanding, each action is a step, the 8 actions is a episode. In RL, we usually discout rewards for history steps, but for each independent episode, rewards are caculated independent.