per
per copied to clipboard
Prioritized Experience Replay (PER) implementation in PyTorch
First of all, thanks for this implementation. One question however: I see that the weights for the importance sampling are calculated and returned when a batch is sampled. However, the...
Print this line. It's always zero, rendering the PER useless. 
Thank you for this great code! Just one question regarding the line 16 of `prioritized_memory.py`: Should not **the absolute value of the TD-error** be taken in `return (error + self.e)...
Hello, thank you for the work. I am facing the issue of dividing by zero error in the line below when calling the sample function to sample memory. Any idea...
Why does my code always throw an error: File "D:\UAV\re_allocation\DQN.py", line 127, in sample state, action, reward, next_state, done = zip(*batch) TypeError: 'int' object is not iterable