How to use LSTM middleware with a DQN agent

IntelLabs / coach

Reinforcement Learning Coach by Intel AI Lab enables easy experimentation with state of the art Reinforcement Learning algorithms

https://intellabs.github.io/coach/

Apache License 2.0

2.32k stars 460 forks source link

How to use LSTM middleware with a DQN agent #352

Open nitsanluke opened 5 years ago

nitsanluke commented 5 years ago

Is it possible to use an LSTM layer with a DQN agent? If so how can specify the sequnece length (number of transitions) to use?

galnov commented 5 years ago

There was a PR a while ago adding support for Clipped PPO with LSTM middleware, including sequence length. We haven't merged it as no response was received from the requester of the feature. You can use it as a reference, and if you get it to work well and would like to contribute to the community, we'd be happy to merge your pull request.