Skip to content

Latest commit

 

History

History
65 lines (45 loc) · 2.57 KB

code.md

File metadata and controls

65 lines (45 loc) · 2.57 KB
Raw

Overview

This document explain the structure of this codebase and hyperparameters of experiments.

File organization

Step 1.

Please refer to the instruction of dopamine structure in here

Step 2.

We add variants of RPG agents in this folder and we explain each agent as follows:

Folder Exploration Supervision
rpg epsilon-greedy RPG (Hinge loss)
lpg epsilon-greedy LPG (Cross-Entropy)
epg EPG LPG (Cross-Entropy)
repg EPG RPG (Hinge loss)
implicit_quantilerpg implicit_quantile RPG (Hinge loss)

  • EPG: EPG is the stochastic listwise policy gradient with off-policy supervised learning, which is the vanilla policy gradient trained with off-policy supervised learning. The exploration and supervision agent is parameterized by the same neural network. The supervision agent minimizes the cross-entropy loss over the near-optimal trajectories collected in an online fashion.

  • LPG: LPG is the deterministic listwise policy gradient with off-policy supervised learning. We choose an action greedily based on the value of logits during the evaluation, and it stochastically explores the environment as EPG.

  • RPG: RPG explores the environment using a separate agent: epsilon-greedy, EPG in Pong and Implicit Quantile in other games. Then rpg conducts supervised learning by minimizing the hinge loss.

In this codebase, the folder rpg contain the code of RPG with epsilon-greedy exploration, and similarly repg for EPG exploration, implicit_quantilerpg for implicit quantile network exploration.

The agents with relatively simple exploration strategy (rpg, lpg, epg, repg) perform well on Pong, comparing to the state-of-the-arts, since there are higher chance to hit the good trajectories with in Pong. For more complicated games, we adopt implicit quantile network as the exploration agent.

Hyperparameters

The hyperparameters of networks, optimizers, etc., are same as the baselines in dopamine. The trajectory reward threshold c (see Def 5 in the paper) for each game is given as follows:

game c
Boxing 100
Breakout 400
Bowling 80
BankHeist 1100
DoubleDunk 18
Pitfall 0
Pong 1
Robotank 65