Tom-the-AI - A compound artificial intelligence software for Linux systems.

Overview

Tom the AI (version 0.82)

WARNING: This software is not yet ready to use, I'm still setting up the GitHub repository. Should be ready in a few days.

Tom is an open source AI desktop assistant for Linux systems, built using a series of independent response modules to generate replies to any input.

Tom uses natural language processing to determine which response module is best suited to generate a response for each input, thus avoiding the need for precise syntax.

Tom the AI

By Analogy

Tom the AI is designed as a Linux alternative to software such as Apple's Siri, or Microsoft's Cortana.

Set Up

Step 1 - Update repositories:

Update apt package repositories using sudo apt update to ensure that the apt package manager has access to the latest versions of the below dependencies.

Step 2 - Install APT dependencies:

First, install python by running sudo apt install python3.9 in a terminal. Tom is tested on python 3.9, but any newer version should (probably) also work just fine.

Next, install the latest version of VLC Media player using sudo apt install vlc.

Step 3 - Download Tom:

Download Tom by cloning the GitHub repository into your home folder using git clone https://github.com/Mblizzard/Tom-the-AI.

Step 4 - Install Python dependencies:

Open a terminal inside Tom's application folder, or navigate using cd ~/Tom-the-AI/. Now run sudo pip3 install requirements.txt. Some systems may use pip in place of pip3.

Next, we need to download the required NLTK libraries by running the following code in a python shell:

>>> import nltk
>>> nltk.download('all')

Step 5 - Running Tom:

Go ahead and run python3.9 ~/Tom-the-AI/tom.py. Tom will boot up, and after a minute or so of loading, you'll be ready to go! If you feel inclined, go ahead and make a desktop launcher of this command, link Tom into your Application Menu, or create a dock shortcut.

Mission

The mission of Tom is to provide an open source compound AI for which anyone can program and contribute response modules, expanding Tom's capabilities to create a useful and entertaining artificial intelligence software.

Examples

Tom generates outputs to any input by using natural language processing to determine the most suitable response module from which to source the reply.

Give Tom natural language input, either via voice recognition or text input, for instance Hey Tom, what is petrichor?, and he'll respond in the most appropriate way. Note that the 'Hey Tom' activation phrase is only required of voice inputs.

The following is a non-exhaustive list of things you can do:

Hey Tom, I'm in an optimistic mood. I'm not sure if this is a good thing or not. Emotions (Using sentiment analysis + NLTK chatbots): ~> Hey Tom, you are a brilliant individual! I am but one, you are but one more. ~> Hey Tom, thou art a fool. Become more interesting before I die of fatal boredom. Fact Memory & Recall: ~> Hey Tom, the answer to life, the universe, and everything is 42. Ok. ~> Hey Tom, what is the answer to life, the universe, and everything?. The answer to life, the universe, and everything is 42. Playing music (From device or web, includes UI controls for the former): ~> Hey Tom, play up the shard. Playing /home/murray/Music/Dr Who/Up The Shard.webm. ~> Hey Tom, stop the music. Media stopped. *NOTE: File names do not have to match exactly.* ~> Hey Tom, open my English essay. Alright. *NOTE: File names do not have to match exactly.* Opening websites: ~> Hey Tom, open Reddit. Alright. Jokes (From PyJokes): ~> Hey Tom, tell me a joke. I went to a street where the houses were numbered 8k, 16k, 32k, 64k, 128k, 256k and 512k. It was a trip down Memory Lane. Trivia: ~> Hey Tom, ask me a trivia question. Question: What is "Sealed crustless sandwich"? 1) The part of Yellowstone National Park in Idaho, where any crime can technically be committed without punishment – but don't tempt fate! 2) I got a fever, and the only prescription... is more cowbell! 3) The only nuclear reactor in a 17th-century building. 4) A patented peanut butter and jelly sandwich. ~> 4. Correct! Colossal Cave Adventure (Willie Crowther's ADVENT-350): ~> Hey Tom, let's go on an adventure! Welcome to adventure!! would you like instructions? Fun facts: ~> Hey Tom, make me smarter. Spices were not used to mask the flavor of rotting meat before refrigeration. Spices were an expensive luxury item; those who could afford them could afford good meat, and there are no contemporaneous documents calling for spices to disguise the taste of bad meat. Dice Rolls (great for D&D): ~> Hey Tom, roll me a d20. I rolled a 14. Word generation (great for Articulate) ~> Hey Tom, give me a random action word. Your word is 'winning'. Complex Mathematics (using SymPy): ~> Hey Tom, integrate (tan(x))^1/2 ∫f(x) = -ln(cos(x))/2 + c Code generation (using howdoi): ~> Hey Tom, write a hello world script in C++. #include <\iostream> int main() { std::cout << "Hello World!" << std::endl; return 0; } Most of Betty's functionality (From https://github.com/pickhardt/betty): ~> Hey Tom, what time is it? Running date +"%r (%T)" ... 02:34:46 PM (14:34:46). ~> Hey Tom, what day is it? Running date +"%A" ... Saturday. ~> Hey Tom, whats my username? Running whoami ... murray ~> Hey Tom, what is my ip address? Wlo1: flags=4163 mtu 1500 inet 192.168.43.9 netmask 255.255.255.0 broadcast 192.168.43.255 inet6 fe80::5c61:caf:5614:7b82 prefixlen 64 scopeid 0x20 ether 54:35:30:60:a8:b9 txqueuelen 1000 (Ethernet) RX packets 401121 bytes 523184185 (523.1 MB) RX errors 0 dropped 0 overruns 0 frame 0 TX packets 235650 bytes 23471151 (23.4 MB) TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0.">
Objective Response (From anywhere on the internet):
~> Hey Tom, what is petrichor?
According to en.wikipedia.org... Petrichor is the earthy scent produced when rain falls on dry soil. The word is constructed from the Greek petra, "rock", or petros, "stone", and ichor, the fluid that flows in the veins of the gods in Greek mythology.

Subjective Response (From Cleverbot):
~> Hey Tom, I'm in an optimistic mood.
I'm not sure if this is a good thing or not.

Emotions (Using sentiment analysis + NLTK chatbots):
~> Hey Tom, you are a brilliant individual!
I am but one, you are but one more.
~> Hey Tom, thou art a fool.
Become more interesting before I die of fatal boredom.

Fact Memory & Recall:
~> Hey Tom, the answer to life, the universe, and everything is 42.
Ok.
~> Hey Tom, what is the answer to life, the universe, and everything?.
The answer to life, the universe, and everything is 42.

Playing music (From device or web, includes UI controls for the former):
~> Hey Tom, play up the shard.
Playing /home/murray/Music/Dr Who/Up The Shard.webm.
~> Hey Tom, stop the music.
Media stopped.
*NOTE: File names do not have to match exactly.*

~> Hey Tom, open my English essay.
Alright.
*NOTE: File names do not have to match exactly.*

Opening websites:
~> Hey Tom, open Reddit.
Alright.

Jokes (From PyJokes):
~> Hey Tom, tell me a joke.
I went to a street where the houses were numbered 8k, 16k, 32k, 64k, 128k, 256k and 512k. It was a trip down Memory Lane.

Trivia:
~> Hey Tom, ask me a trivia question.
Question: What is "Sealed crustless sandwich"?
1) The part of Yellowstone National Park in Idaho, where any crime can technically be committed without punishment – but don't tempt fate!
2) I got a fever, and the only prescription... is more cowbell!
3) The only nuclear reactor in a 17th-century building.
4) A patented peanut butter and jelly sandwich.
~> 4.
Correct!

Colossal Cave Adventure (Willie Crowther's ADVENT-350):
~> Hey Tom, let's go on an adventure!
Welcome to adventure!! would you like instructions?

Fun facts:
~> Hey Tom, make me smarter.
Spices were not used to mask the flavor of rotting meat before refrigeration. Spices were an expensive luxury item; those who could afford them could afford good meat, and there are no contemporaneous documents calling for spices to disguise the taste of bad meat.

Dice Rolls (great for D&D):
~> Hey Tom, roll me a d20.
I rolled a 14.

Word generation (great for Articulate)
~> Hey Tom, give me a random action word.
Your word is 'winning'.

Complex Mathematics (using SymPy):
~> Hey Tom, integrate (tan(x))^1/2
∫f(x) = -ln(cos(x))/2 + c

Code generation (using howdoi):
~> Hey Tom, write a hello world script in C++.
#include <\iostream>
int main()
{
std::cout << "Hello World!" << std::endl;
return 0;
}

Most of Betty's functionality (From https://github.com/pickhardt/betty):
~> Hey Tom, what time is it?
Running date +"%r (%T)" ...
02:34:46 PM (14:34:46).
~> Hey Tom, what day is it?
Running date +"%A" ...
Saturday.
~> Hey Tom, whats my username?
Running whoami ...
murray
~> Hey Tom, what is my ip address?
Wlo1: flags=4163
    
      mtu 1500
    inet 192.168.43.9 netmask 255.255.255.0 broadcast 192.168.43.255
    inet6 fe80::5c61:caf:5614:7b82 prefixlen 64 scopeid 0x20
     
    ether 54:35:30:60:a8:b9 txqueuelen 1000 (Ethernet)
    RX packets 401121 bytes 523184185 (523.1 MB)
    RX errors 0 dropped 0 overruns 0 frame 0
    TX packets 235650 bytes 23471151 (23.4 MB)
    TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0.

    

This is a fair representation of Tom's capabilities as they currently stand. See the following section on contributing for a guide of how to create your own response modules for Tom, and expand upon the above abilities.

Contributing

How to write a custom response module for Tom:

Step 1 - Understanding how Tom will treat your module:

Tom is programmed in Python. Response modules are imported into Tom using the python import statement, and the response is retrieved from the module using output = .respond( ) . The output is then returned to the user.

Step 2 - Programming the response module:

Go ahead and program your response. Your script should have a main function def respond(inp):, where inp is the user input parameter that will be passed to your function by Tom. Your function should provide it's output through a return statement (NOT a print() statement).

Step 3 - Testing your module:

Paste the following bit of code at the end of your python script, then run your program:

")))">
if __name__ == "__main__":
    while True:
        print(respond(input("~> ")))

If this works as expected, and you can type inputs on the ~> prompts and receive your output printed in the console, then continue to step 4.

Step 4 - Relative imports:

Rename your main response script to __init__.py, and make sure it's at the first level of your project folder (not nested in other folders). Next, rename the folder containing your script to the name of your module (no white-space or special characters). Now, if you are importing any functions from other scripts (does not include dependencies installed through pip), you will need to change the import statement by placing a '.' in front of the location. For example, from myOtherScript import customFunction becomes from .myOtherScript import customFunction, but import requests would remain unchanged.

Step 5 - Dependencies:

If your response module requires python packages from PyPi, make sure it includes a requirements.txt file. Any dependencies not available from PyPi should bundled with project, located in the project folder alongside __init__.py.

Step 6 - Using your module:

Paste the folder containing your response module into Tom's /responses directory. You will then need to activate the response module within Tom's modules interface, or by manually adding the name of your module to responseOrder.txt.

Step 7 - Creating a pull request:

If you feel inclined to share your module with the world, go ahead and create a pull request for your module on Tom's GitHub repository (https://github.com/Mblizzard/Tom-the-AI).

Planned Features

New response modules & capabilities to look forward to in future versions of Tom:

  • Timers & stopwatch capabilities.
  • Ability execute terminal commands.
  • Automated module installation.
  • Releases and updates available on the Ubuntu apt repositories.

Features I'm not currently planning to include in Tom, but that I'll consider adding if enough people are interested:

  • Windows support.

Versioning

Releases will follow a semantic versioning format:

. .

For more information on SemVer, visit http://semver.org/.

License

Tom the AI: A compound AI for Linux systems.
Copyright (C) 2021  Murray Jones

This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.

This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with this program. If not, see 
   .
naked is a Python tool which allows you to strip a model and only keep what matters for making predictions.

naked is a Python tool which allows you to strip a model and only keep what matters for making predictions. The result is a pure Python function with no third-party dependencies that you can simply c

Max Halford 24 Dec 20, 2022
🧮 Matrix Factorization for Collaborative Filtering is just Solving an Adjoint Latent Dirichlet Allocation Model after All

Accompanying source code to the paper "Matrix Factorization for Collaborative Filtering is just Solving an Adjoint Latent Dirichlet Allocation Model A

Florian Wilhelm 39 Dec 03, 2022
Deep Learning: Architectures & Methods Project: Deep Learning for Audio Super-Resolution

Deep Learning: Architectures & Methods Project: Deep Learning for Audio Super-Resolution Figure: Example visualization of the method and baseline as a

Oliver Hahn 16 Dec 23, 2022
Reference models and tools for Cloud TPUs.

Cloud TPUs This repository is a collection of reference models and tools used with Cloud TPUs. The fastest way to get started training a model on a Cl

5k Jan 05, 2023
Lenia - Mathematical Life Forms

For full version list, see Timeline in Lenia portal [2020-10-13] Update Python version with multi-kernel and multi-channel extensions (v3.4 LeniaNDK.p

Bert Chan 3.1k Dec 28, 2022
novel deep learning research works with PaddlePaddle

Research 发布基于飞桨的前沿研究工作,包括CV、NLP、KG、STDM等领域的顶会论文和比赛冠军模型。 目录 计算机视觉(Computer Vision) 自然语言处理(Natrual Language Processing) 知识图谱(Knowledge Graph) 时空数据挖掘(Spa

1.5k Dec 29, 2022
A motion detection system with RaspberryPi, OpenCV, Python

Human Detection System using Raspberry Pi Functionality Activates a relay on detecting motion. You may need following components to get the expected R

Omal Perera 55 Dec 04, 2022
Code for "Adversarial attack by dropping information." (ICCV 2021)

AdvDrop Code for "AdvDrop: Adversarial Attack to DNNs by Dropping Information(ICCV 2021)." Human can easily recognize visual objects with lost informa

Ranjie Duan 52 Nov 10, 2022
Benchmarking Pipeline for Prediction of Protein-Protein Interactions

B4PPI Benchmarking Pipeline for the Prediction of Protein-Protein Interactions How this benchmarking pipeline has been built, and how to use it, is de

Loïc Lannelongue 4 Jun 27, 2022
EigenGAN Tensorflow, EigenGAN: Layer-Wise Eigen-Learning for GANs

Gender Bangs Body Side Pose (Yaw) Lighting Smile Face Shape Lipstick Color Painting Style Pose (Yaw) Pose (Pitch) Zoom & Rotate Flush & Eye Color Mout

Zhenliang He 321 Dec 01, 2022
DANet for Tabular data classification/ regression.

Deep Abstract Networks A pyTorch implementation for AAAI-2022 paper DANets: Deep Abstract Networks for Tabular Data Classification and Regression. Bri

Ronnie Rocket 55 Sep 14, 2022
Airborne Optical Sectioning (AOS) is a wide synthetic-aperture imaging technique

AOS: Airborne Optical Sectioning Airborne Optical Sectioning (AOS) is a wide synthetic-aperture imaging technique that employs manned or unmanned airc

JKU Linz, Institute of Computer Graphics 39 Dec 09, 2022
Ἀνατομή is a PyTorch library to analyze representation of neural networks

Ἀνατομή is a PyTorch library to analyze representation of neural networks

Ryuichiro Hataya 50 Dec 05, 2022
A Simple and Versatile Framework for Object Detection and Instance Recognition

SimpleDet - A Simple and Versatile Framework for Object Detection and Instance Recognition Major Features FP16 training for memory saving and up to 2.

TuSimple 3k Dec 12, 2022
Directed Greybox Fuzzing with AFL

AFLGo: Directed Greybox Fuzzing AFLGo is an extension of American Fuzzy Lop (AFL). Given a set of target locations (e.g., folder/file.c:582), AFLGo ge

380 Nov 24, 2022
SCALoss: Side and Corner Aligned Loss for Bounding Box Regression (AAAI2022).

SCALoss PyTorch implementation of the paper "SCALoss: Side and Corner Aligned Loss for Bounding Box Regression" (AAAI 2022). Introduction IoU-based lo

TuZheng 20 Sep 07, 2022
Efficient Conformer: Progressive Downsampling and Grouped Attention for Automatic Speech Recognition

Efficient Conformer: Progressive Downsampling and Grouped Attention for Automatic Speech Recognition Official implementation of the Efficient Conforme

Maxime Burchi 145 Dec 30, 2022
Model-based reinforcement learning in TensorFlow

Bellman Website | Twitter | Documentation (latest) What does Bellman do? Bellman is a package for model-based reinforcement learning (MBRL) in Python,

46 Nov 09, 2022
Rainbow is all you need! A step-by-step tutorial from DQN to Rainbow

Do you want a RL agent nicely moving on Atari? Rainbow is all you need! This is a step-by-step tutorial from DQN to Rainbow. Every chapter contains bo

Jinwoo Park (Curt) 1.4k Dec 29, 2022
A Demo server serving Bert through ONNX with GPU written in Rust with <3

Demo BERT ONNX server written in rust This demo showcase the use of onnxruntime-rs on BERT with a GPU on CUDA 11 served by actix-web and tokenized wit

Xavier Tao 28 Jan 01, 2023