amoudgl

"""
Adapted from modula's Hello GPT tutorial:
https://github.com/modula-systems/modula/blob/ede2ba72a1b9de3e1f44156db058b5c32c682941/examples/hello-gpt.ipynb

This script simply exposes dataloader seed as command line argument to test
training sensitivity to seed.

Usage:

python hello-gpt.py --seed 0

amoudgl

Input shape: (12, 64)
Target shape: (12, 64)
First input sequence: [41 53 50 42  1 40 50 53 53 42] ...
First target sequence: [53 50 42  1 40 50 53 53 42  1] ...

Decoded input: cold blood no spark of honour bides.

NORTHUMBERLAND:
Be thou a 

amoudgl

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,

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

amoudgl

Setup Visdom on Remote Server

Install visdom on your local system and remote server.

pip3 install visdom

On remote server, do:

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

# 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

# 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