Sam Stites

add a testable "convergence" criteria

2

This would allow us to start doing some convergence testing for #10

break the dependency on the gym-http-api server -- call the gym directly

8

one option might be to call the gym directly with call-python-via-msgpack: https://github.com/nh2/call-python-via-msgpack this would speed up performance considerably and would make me feel more comfortable about uploading to hackage.

should be as simple as a gym call, but I'm not clear on the details.

I'm not sure what this looks like, but being able to call a universe environment opens up a lot of sweet environments and it would be nice to be able...

This is going to be a meta branch to discuss any deep reinforcement learning implementations of algorithms to write, as well as any structural changes to the folder structure if...

- [ ] Actor-critic methods - [ ] REINFORCE

Add eligibility trace variants in algorithms

6

If you're unfamiliar with eligibility traces, they basically unify temporal-difference learning with monte carlo methods -- essentially you hold a buffer in memory of an agent's experience and perform reward...