MiniTorch is a diy teaching library for machine learning engineers who wish to learn about the internal concepts underlying deep learning systems. It is a pure Python re-implementation of the Torch API designed to be simple, easy-to-read, tested, and incremental. The final library can run Torch code. The project was developed for the course Machine Learning Engineering at Cornell Tech.
Github: https://github.com/minitorch/
In this article, I will take you through an explanation and implementation of all Machine Learning algorithms with Python programming language.
Machine learning algorithms are a set of instructions for a computer on how to interact with, manipulate, and transform data. There are so many types of machine learning algorithms. Selecting the right algorithm is both science and art.
Jina is geared towards building search systems for any kind of data, including text, images, audio, video and many more. With the modular design & multi-layer abstraction, you can leverage the efficient patterns to build the system by parts, or chaining them into a Flow for an end-to-end experience. Large-scale indexing and querying of unstructured data: video, image, long/short text, music, source code, etc. Decentralized architecture from day one. Scalable & cloud-native by design: enjoy containerizing, distributing, sharding, async, REST/gRPC/WebSocket.
Self hostable?
PyCameraServer is a Flask video / image / Youtube / IP Camera frames online web-editor with live streaming preview for objects recognition, extraction, segmentation, resolution upscaling, styling, colorization, interpolation, using OpenCV with neural network models: YOLO, Mask R-CNN, Caffe, DAIN, EDSR, LapSRN, FSRCNN, ESRGAN.
CUDA enabled.
Free and Open Source Machine Translation API. 100% self-hosted, no limits, no ties to proprietary services. Run your own API server in just a few minutes. Playing with it a little, it seems like it might be interesting to experiment with. Supports a couple of languages right now, but at least they're useful ones.
API docs: https://libretranslate.com/docs/
Github: https://github.com/uav4geo/LibreTranslate
The back end appears to be written in Python. Examples of use are in Javascript, but that's not necessarily the way it has to be done.
Language models are kept in a different repo: https://github.com/uav4geo/LibreTranslate-Models
EleutherAI is a grassroots AI research group aimed at democratizing and open sourcing AI research. Multiple projects and usable training corpora. F/OSS model called GPT-Neo.
Several spinoff projects to investigate.
Script that will detect if a stranger is trying to use your laptop or if a stranger/authorized driver is trying to drive your car. This script will detect the face, and send you an email if new user is not identified.
A generative adversarial network that takes images and re-does them in various anime studios' styles. GPU enabled. Download the pre-trained model to get a jumpstart.
A framework that instantiates AI/ML models more or less automagickally - one line of code. Can read training data from URLs as well as locally. Has a GUI also but it's an optional install. Sits on top of PyTorch.
The description's kind of weird, but it seems to use a neural network to generate pictures of someone based upon a picture that you give it. "Generate your lover with your photo," whatever that means.
A face detector (not facial identification) deep learning system based upon OpenCV and Tensorflow. Optimized for CPU, not GPU operation but does have a tensorflow-gpu switch available. Can even identify faces that aren't edge-on or partially obscured.
Transformers (formerly known as pytorch-transformers and pytorch-pretrained-bert) provides state-of-the-art general-purpose architectures (BERT, GPT-2, RoBERTa, XLM, DistilBert, XLNet, CTRL...) for Natural Language Understanding (NLU) and Natural Language Generation (NLG) with over 32+ pretrained models in 100+ languages and deep interoperability between TensorFlow 2.0 and PyTorch.
Stanford has open sourced a self-hosted, personal assistant system. Designed with privacy in mind. Speech recognition, analysis, task execution. They want to make so that it's easy and highly useful for everyone to integrate into their stuff. Can monitor things and filter for certain things. Aims for composability. Services (skills) are also open source and crowdsourced.
Github: https://github.com/stanford-oval
An application that uses AI and ML to intelligently strip the vocals out of music tracks. Written in Python, uses Tensorflow. The pre-trained model is included in the distribution. Use as a CLI tool or a library for your own code.
textacy is a Python library for performing a variety of natural language processing (NLP) tasks, built on the high-performance spaCy library. With the fundamentals --- tokenization, part-of-speech tagging, dependency parsing, etc. --- delegated to another library, textacy focuses primarily on the tasks that come before and follow after. Abstracts away the boilerplate for the stuff you actually care about.
Quickstart: https://chartbeat-labs.github.io/textacy/getting_started/quickstart.html
A remarkably streamlined and simple to use system for using AI and ML models to interact with data. Build interactive data analyses with just a very little code. The demo shows most of everything you need to be productive. Hot reloading; change some of the Python code in your research script, the display updates. Don't need to mess with HTML and Javascript, just a text editor and a web browser.
A F/OSS natural language translation system that seems to want to give Google Translate a run for its money. The corpuses used for training appear to be crowdsourced, and I think you can download the trained models on their own. Aims to be self-hosted.
Github: https://github.com/apertium
Installation docs: http://wiki.apertium.org/wiki/Installation
An NLP deep learning toolkit for building training pipelines. Tries to minimize the effort for constructing the training and inference stages. Defines modular building blocks of neural network components, and a suite of NLP models. The end goal is to make building a neural network as easy as playing with Legos. Supports English and Chinese.
A three-stage deep learning system which can figure out how to imitate a person's voice from as little as five seconds of recorded speech. Speaks with the deepfaked voice in realtime. The sample is used to condition an existing TTS model to sound like someone.
Running it inside a Docker container: https://sean.lane.sh/posts/2019/07/Running-the-Real-Time-Voice-Cloning-project-in-Docker/
A basic server which implements a neural network and a REST API. Give it some basic training data, and then chat with it. It will slowly learn how to communicate with you.