https://github.com/alicerunsonfedora/abysima

A machine learning experiment with generating languages.
https://github.com/alicerunsonfedora/abysima

coreml generative-models keras lingustics recurrent-neural-networks swiftui

Last synced: 6 days ago
JSON representation

A machine learning experiment with generating languages.

• Host: GitHub
• URL: https://github.com/alicerunsonfedora/abysima
• Owner: alicerunsonfedora
• License: mpl-2.0
• Created: 2021-10-02T16:52:22.000Z (over 3 years ago)
• Default Branch: root
• Last Pushed: 2021-12-11T16:53:00.000Z (about 3 years ago)
• Last Synced: 2024-10-29T08:24:14.718Z (4 months ago)
• Topics: coreml, generative-models, keras, lingustics, recurrent-neural-networks, swiftui
• Language: Jupyter Notebook
• Homepage:
• Size: 8.66 MB
• Stars: 2
• Watchers: 2
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.txt

Awesome Lists containing this project

README

        
# Project Abysima

Welcome to the repository/notebook for **Project Abysima**. This notebook contains all of the notes and drafts for the project à la Markdown/[Obsidian](https://obsidian.md), as well as the Juptyer notebooks used to create the networks.

## 🏃🏻‍♂️ Quick Links

```ad-note

title: Obsidian Linking

The quick links below use Obsidian's wiki-style linking format; links may not work correctly when viewing other Markdown editors or viewing this document in GitHub. The paths to the links are included as sub-bullet points in this list.

```

- [[Linguistics Paper]]

	- `01 - Areas of Responsibility/Linguistics Paper`

- [[Annotated Bibliography]]

	- `03 - Resources/Annotated Bibliography`

[View the Jupyter Notebook →](./notebooks/validations.ipynb)

## ℹ️ What is Project Abysima?

**Project Abysima** is an attempt a creating a generative neural network that will devise its own language (not programming) based off of existing linguistic rules across languages. The main objective of the project is the following:

- Can we get a machine learning algorithm to generate a language?

- What linguistic properties can we use to improve these algorithms?

- Can we make linguistics and linguistic properties easy to understand for a neural network?

## 🗄 General Organization

The organization of this project is broken down into four domains:

- **Projects** contains series of tasks that are linked to a specific goal with a given deadline.

- **Areas of Responsibility** contain a sphere of activity that will be maintained over time.

- **Resources** contains themes, topics, and other notes of interest.

- **Archives** contain archived data from the previously mentioned areas.

More information on this approach to project organization can be found at https://fortelabs.co/blog/para/.