amoudgl

# monte carlo policy gradient algorithm 
# use neural network to decide the policy
# from observations and rewards, update the parameters of the neural networks to optimize the policy

import numpy as np
import tensorflow as tf
import gym
from gym import wrappers

# initialize constants

amoudgl

# hill climbing algorithm 
# generate a random configuration of the parameters, add small amount of noise to the parameters and evaluate the new parameter configuration
# if new configuration is better than old one, discard the old one and accept the new one
# optimize it for weighted sum 

# returns the net episode reward 
def get_episode_reward(env, observation, params):
    t = 0
    net_reward = 0
    while (t < 1000):

amoudgl

# random guessing algorithm 
# generate 10000 random configurations of the model's parameters and pick the one that achieves the best cumulative reward. 
# optimize it for weighted sum 

import gym
from gym import wrappers
import numpy as np

env = gym.make('CartPole-v0')
env = wrappers.Monitor(env, '/tmp/cartpole-random-guessing', force=True)

amoudgl

# monte carlo policy gradient algorithm 
# use neural network to decide the policy
# from observations and rewards, update the parameters of the neural networks to optimize the policy

import numpy as np
import tensorflow as tf
import gym
from gym import wrappers

# initialize constants

amoudgl

# simple neural network implementation of qlearning
import gym
from gym import wrappers 
import numpy as np
import tensorflow as tf

# build environment
env = gym.make("FrozenLake-v0")
env = wrappers.Monitor(env, '/tmp/frozenlake-qlearning', force=True)
n_obv = env.observation_space.n

amoudgl

Checkout the lastest versions of our datasets!

Following are the higher quality versions of TLP and TinyTLP datasets. The resolution is still the same but images are much sharper and have higher quality.

TLP V2 (87.2 GB)

https://drive.google.com/open?id=1mv0ULctCzGn4gzum_Sb6kwXwB2QjZwu1

TLP V2 Sequences

amoudgl

Setup Visdom on Remote Server

Install visdom on your local system and remote server.

pip3 install visdom

On remote server, do:

amoudgl

amoudgl

%% This part goes in preamble
\newcommand{\dummyfig}[1]{
  \centering
  \fbox{
    \begin{minipage}[c][0.33\textheight][c]{0.5\textwidth}
      \centering{#1}
    \end{minipage}
  }
}

amoudgl

# Copyright 2021 DeepMind Technologies Limited. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,