- Status: Under Development
- First Beta Testing Release Target: June 2017
- Vesion 1.0 Release Target: Sept 2017

I began work on the implementation of the CommAI-env based on the release incorporated within the code provided in a commercial AI contest. After some initial exploration and refactoring, I confirmed my theory it would more efficient to fork farther upstream. On April 15th I set aside my work to date, and initiated this repository as a refactor of the Github facebookreseacrh/CommAI-env Beta-Test release.

I was motivated to start publicly coding instead of just contemplating the project privately by: Nat Friedman's AIGrant: Get $5,000 for your open source AI project:

https://nat.org/aigrant-get-5-000-for-your-open-source-ai-project-1118dd7db083

I am including the initial development from forking as a consequence as part of my entry.

I subscribe to Nat's feed via Medium: https://medium.com Nat is one a number of writers whose work is available FREE of third party advertising, published at Medium. I make time to read when I see a new post.

Type: Github personal accounts Purpose: Enable features on top of Github's FREE features

From amount $9.91 CAD To amount $7.00 USD Exchange rate: 1 CAD = 0.706704 USD

May 3, 2017

• 9.91 CAD

Nat's contest gained a sponsor, the prize pool has been extended from five $5000 to ten $5000 AI Grants.

https://nat.org/ai-grant-gets-a-sponsor-thanks-floodgate-a945afc79470

http://floodgate.com/

I'm very excited at the rapid pace of development as both the Project Lead and as the sole contributor to the effort. I hit the ground running to speed the process of refactoring my code and docstrings to a level where more than one contributor can check out files.

All Python source files have been refactored into Python style pseudo code, and tagged with notations where conversions are required to use the code blocks individually as functional code.

The development style here, in regards to the scope of this project, is:

Any discussion which adds to the progress towards a working implementation of general intelligence must be done in the form of a logical statement. The form factor does not matter, as an equation, symbolic logic, psuedo code, a structured language like Python, they all represent the same thing. Discussions in other media have great value for other purposes. However for the purposes outlined here are simply not relevant to progress in the implementation of general AI. Anything outside of those constraints, however well-reasoned, are examples of speculative fiction.

Creating an Issue for each file containing #TODOs or #FIXME, completion: Week 2.

Everything else. One of the core of elements of my development process is: Don't spend time preparing for description of future work when completing the work itself is possible within the given constraints. Also if the task is too complex to communicate in a few sentences to a peer, it needs to be reduced further.

As the initial week has no baseline to provide metrics, first metrics reports will be provided on May 5, 2017.

- Codeclimate rating from 1.04 - 3.46, 700+ issues, 
  (click Code Climate at top of page for details)

- Excluding merges, I have pushed 365 commits to master and 365 commits to all branches. 
- On master, 115 files have changed and there have been 14,070 additions and 69 deletions.

Raw reports of lines added & deleted, commits, and other activity can be viewed via the Pulse and other tabs at the top of the page.

The facebookresearch Team began the implementation of the CommAI-env in summer of 2016. Their work to date has provided me with an entry point into the ongoing development of the concepts detailed by Marco Baroni et al. in their paper CommAI-env. The abstract is available in /docs/CommAI_PDF_1701.08954.pdf.

After a line-by-line examination of the facebookresearch implementation, I was inspired by the possibilities a fresh perspective could bring to the table, and ultimately contribute to the existing body of work.

The approach I propose to further the development of the source code and documentation are a radical departure from the conventions I've observed in my limited reviews of existing development work-flows and wfm (workforce management) strategies.

I am still exploring how best to communicate the work flow information. A few stubs were added to the Github Wiki here. https://github.com/stephenbhope/Revised-CommAI-env/wiki

I will be documenting my techniques, tools, and strategies during the process from idea to release 1.0 in the hope the information will be of use to the community at large to 'cherry pick' any material relevant to their requirements, work style, and working environment. Or simply as another data point in their set.

The GNU 3.0+ style LICENSE is the most well suited to my objectives. In essence, it grants everyone permission to use the intellectual property contained in this body (* outside of the actual facebookresearch source code which is governed by the CommAI-env_software_LICENSE.txt in the root directory). Without charge, forever.

In addition, the GNU license makes this work available without imposing constraints on users or contributors of future Licensing or Patent enforcement by third parties and opens doors to people bound by other agreements.

• You may open a discussion by clicking the issue tab at the top of the page, New Issue.
• Service access, financial donations, or any form of sponsorship would be very welcome.
• Become a Patron!

Ottawa, Canada stephenbhope@gmail.com Tel [00] 1-613-594-4727

By Friday May 5th more structured methods for communicating and sharing will be found in docs/Contributions.md file.

Marco Baroni, Armand Joulin, Allan Jabri, Germàn Kruszewski, Angeliki Lazaridou, Klemen Simonic, Tomas Mikolov

With machine learning succesfully applied to new daunting problems almost every day, general AI starts looking like an attainable goal. However, most current research focuses instead on important but narrow applications, such as image classification or machine translation. We believe this to be largely due to the lack of objective ways to measure progress towards broad machine intelligence. In order to fill this gap, we propose here a set of concrete desiderata for general AI, together with a platform to test machines on how well they satisfy such desiderata, while keeping all further complexities to a minimum.

arXiv:1701.08954v2 [cs.LG] Cornell University Library

CommAI-env, https://arxiv.org/abs/1511.08130 The Github repository, https://github.com/facebookresearch/CommAI-env

Copyright (c) 2017, Stephen B. Hope, see LICENSE for details.