Port DRL algo from SB3 on GYM to Ray RLlib
Budget: $250 – $750 USD
Working with github FinRL repo, the example works on training in a single core only.
Looking to port the trainer to utilize something with a more powerful GPU setup.
Looking to train with DD-PPO or Apex-DDPG
https://docs.ray.io/en/latest/rllib/rllib-algorithms.html#decentralized-distributed-proximal-policy-optimization-dd-ppo
https://docs.ray.io/en/latest/rllib/rllib-algorithms.html#distributed-prioritized-experience-replay-ape-x
The current trading environment of FinRL is written with the stablebaselines3, so the idea is to port over the gym.Env to use with Ray RLlib. The general condition of the gym env remains.
Just a code port.
Looking to port the trainer to utilize something with a more powerful GPU setup.
Looking to train with DD-PPO or Apex-DDPG
https://docs.ray.io/en/latest/rllib/rllib-algorithms.html#decentralized-distributed-proximal-policy-optimization-dd-ppo
https://docs.ray.io/en/latest/rllib/rllib-algorithms.html#distributed-prioritized-experience-replay-ape-x
The current trading environment of FinRL is written with the stablebaselines3, so the idea is to port over the gym.Env to use with Ray RLlib. The general condition of the gym env remains.
Just a code port.