SAC with PER

[xwen24]

Hello!

Thanks so much for sharing the codes!

I just have a question about sac_discrete_per.py: why the importance-sampling weights are not used to update the Q-function and the policy function?

Popular-RL-Algorithms/sac_discrete_per.py

Line 162 in 5926245

weight_update = [min(l1.item(), l2.item()) for l1, l2 in zip(q_value_loss1, q_value_loss2)]

I have seen other SAC-PER repositories that used the weights to calculate the q_value_loss and policy_loss.

Thanks.