MyCaffe.trainers.pg.mt Namespace Reference

Classes
class	Agent
	The Agent both builds episodes from the envrionment and trains on them using the Brain. More...
class	Brain
	The Brain uses the instance of MyCaffe (e.g. the open project) to run new actions and train the network. More...
class	Memory
	Specifies a single Memory (e.g. an episode). More...
class	MemoryCache
	Contains the best memory episodes (best by highest total rewards) More...
class	MemoryItem
	The MemoryItem stores the information for one step in an episode. More...
class	Optimizer
	The Optimizer manages a single thread used to apply updates to the primary instance of MyCaffe. Once applied, the new weights are then copied back to the worker who just applied its gradients to the primary MyCaffe. More...
class	TrainerPG
	The TrainerPG implements a simple Policy Gradient trainer inspired by Andrej Karpathy's blog posed referenced. More...
class	Worker
	The Worker class provides the base class for both the Environment and Optimizer and provides the basic threading functionality used by both. More...
class	WorkerStartArgs
	The WorkerStartArgs provides the arguments used when starting the agent thread. More...