gyx
gyx copied to clipboard
doctorcorral
Reinforcement Learning environment for Elixir
Setting up ApeX should be straightforward See https://github.com/sorentwo/oban
This implementation would handle an enumerable discrete set of observations and actions. Set sizes would be the unique parameter required.
This environment is useful for understanding how to solve finite MDP with perfect information using dynamic programming. Solving this environment requires an iterative policy optimization followed directly from Bellman´s expectation...
An environment render allows seeing the environment representation of its current state. The render can be in the terminal for games like FrozenLake, but atari like environments need a richer...
Check https://github.com/deepmind/bsuite for the original project. Paper with details: https://arxiv.org/abs/1908.03568
An environment render allows seeing the environment representation of its current state. The render can be in the terminal for games like FrozenLake, but atari like environments need a richer...
See OTP [Agent]([https://elixir-lang.org/getting-started/mix-otp/agent.html]) documentation. Implement [Gyx.Core.ReplayMemory](https://github.com/doctorcorral/gyx/blob/master/lib/core/replay_memory.ex) behaviour.
Usually, actions or environment states are represented as single points in the action and environment state spaces. For full operability between no Elixir components, a serialization method is needed. This...