I'm working on adding reinforcement learning support for this library making use of it's powerful evolution capabilites to dynamically evolve a network through reinforcement learning. In the DQN algorithm, I need to train off a replay buffer and then do things afterwards but currently the evolve function is async so I cannot use it in a sync function without everything exploding. Can you guys add an evolveSync function to get around this?
I'm working on adding reinforcement learning support for this library making use of it's powerful evolution capabilites to dynamically evolve a network through reinforcement learning. In the DQN algorithm, I need to train off a replay buffer and then do things afterwards but currently the
evolvefunction is async so I cannot use it in a sync function without everything exploding. Can you guys add anevolveSyncfunction to get around this?