Fly paddle PaddlePaddle carefully prepared seven video lessons, from today to the National Day holiday, will be released.
Today we bring to you the course is: "Flying paddle PaddlePaddle reinforcement learning framework Parl resolved."
Parl framework is based on the depth of learning to fly paddle, with high flexibility and scalability, support for customizable parallel expansion, covering DQN, DDPG, PPO, IMPALA, A2C, GA3C other mainstream reinforcement learning algorithm. Nearly 20,000 pulling node operation by CPU 8 GPU, nearly five hours in an iterative algorithm PPO accelerated to 1 minute.
The following video from the field to fly paddle depth study of the open class AI 2019 Baidu Developers Conference.
Acquiring speech PPT, click here .