AI

6 posts

·3 min·Deep RL with Unity 3/3

DDPG on Reacher: Twenty Arms Chasing a Balloon

Solving Unity's Reacher with DDPG: twenty robot arms sharing one replay buffer, the actor-critic setup, and why it trained so much easier than Tennis.

·3 min·Deep RL with Unity 2/3

Two DDPG Agents Learning to Play Tennis

Two DDPG agents learning to rally in Unity's Tennis environment: continuous actions, sparse shared rewards, what didn't work, and the settings that solved it.

·2 min·Deep RL with Unity 1/3

Training a DQN Agent to Collect Bananas in Unity

Training a DQN agent on Unity's Banana Collector: the Q-network, replay buffer and target network settings, and how it solved the task (+13 over 100 episodes).