Deep Reinforcement Learning in Large Discrete 'Action Spaces (2016) put forward Wolpertinger Training with DDPG to handle Large Discrete' Action space problem, is now I also want to use this algorithm to find related resources on making (https://github.com/ChangyWen/wolpertinger_ddpg), but not too miscellaneous content can be used to distinguish between ourselves, the resource is a gym to write, individual research inventory management, hope that through this method to study the problem, my problem, for example, the state is 3 d said inventory, the Action is a vector in 3 d, including an order pricing, etc. Here is the Action of Discrete space a total of 800, his writing environment to simulate,
The
After
Want to ask what a great god can change on the above code resources to adapt to the inventory problem, thank you!

CodePudding user response:

The building Lord problem solved, please