Reinforcement-Learning

Fail Fast, Learn Faster

This repository contains the implementation and notes of different Reinforcement Learning Algorithms and techniques I have learned throughout time.

Each algorithm uses GYM environments gym environements.

Also now I have published a python package pb-deepRL - my own python package through which you can import my helper functions such as ReplayBuffer etc

package - pb-deepRL

!pip install pb-deepRL

usage -

• from deep_rl_impl.utils.record_game import record_game
• from deep_rl_impl.value_based.dqn.replay_buffer import ReplayBuffer etc

Algorithms

1. Model based learning - Dynamic Programming
   • Policy Iteration
   • Value Iteration
2. Model free learning
   • Monte Carlo
     • Monte Carlo
     • Online Monte Carlo
     • Monte Carlo with decay scheduler
   • SARSA
   • Q-learning
   • TD_N
3. Deep Reinforcement Learning
   • Value Based
     • Deep Q learning
   • Policy Based
     • REINFORCE
     • REINFORCE With Baseline

Steps to push a version to prod-

• login to pypi.org
• create a api token
• now open command line and go to the root folder of the dir
• install - pip install build twine
• bulid the package - python -m build
• dist folder created now upload the dist to pypi using twine
• twine upload --verbose -u token -p TOKEN dist/*
• and always while pushing a new build change the version number