Skip to content

Bayesian Approximate Reinforcement Learning (BARL)

License

Notifications You must be signed in to change notification settings

ai-nikolai/barl

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

40 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

BARL - Bayesian Approximate Reinforcement Learning

This package should serve as a collection of tools to do RL in general and in particular bayesian RL.

The Main Features(Jul 2019):

  1. estimators
  2. agents
  3. environments
  4. simulations & visualisation

Installation:

PIP:

pip3 install barl

Github:

git clone https://github.com/ai-nikolai/barl
cd barl
pip3 install -e .

Usage:

Testing

cd barl
pytest

Experiments:

cd barl
cd experiments
python3 experiments_mab.py

Scripts:

import barl

env = barl.environments.MultiArmedBandit(arms=4)

agent1 = barl.agents.baselines.RandomActionsSampler(numActions=4)

total, arlist, _ = barl.simulations.run_state_less_agent_and_env( environment=env, agent=agent1, N=100)

barl.utils.plotting.plot_reward_over_time_from_ar(arlist)

Copyright (C) - Nikolai Rozanov 2019-Present

About

Bayesian Approximate Reinforcement Learning (BARL)

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages