Max weight should be calculated over whole buffer - Githubissues

waffoo / accel

accelerate reinforcement learning

MIT License

1 stars 1 forks source link

Max weight should be calculated over whole buffer #6

Closed waffoo closed 4 years ago

waffoo commented 4 years ago

Max weight should be calculated over all experience using the minimum probability like https://github.com/openai/baselines/blob/master/baselines/deepq/replay_buffer.py

https://github.com/waffoo/accel/blob/bf7f975729dc04e4ee2b0766ad12fa839698f0f6/accel/replay_buffers/prioritized_replay_buffer.py#L55-L62