Skip to main content
Blog

LF AI & Data Announces Three New Sandbox Projects

By July 21, 2022August 9th, 2022No Comments

LF AI & Data Foundation – the organization building an ecosystem to sustain open source innovation in artificial intelligence (AI) and data open source projects – is announcing three new Sandbox Projects; 1ChipML, BeyondML, and FlagAI to their portfolio of projects.

 

Open sourced in April 2022 by Ericsson, 1ChipML aims to create an open source library for basic numerical crunching and machine learning that can run on the various families of microcontrollers developed by AVR and ARM. These devices, especially the ARM Cortex-M family, have recently shown relatively advanced computing capabilities, both in terms of clock speed and memory access. This allows these extremely low power demanding microcontrollers to perform mathematical and machine learning algorithms that were practically impossible in the past. As we gradually enter the era of the Internet of Things (IoT) and of Edge Computing, a library like this is highly desirable to enable practical applications in real life. In addition, a variety of future AI applications will be designed to use compute distributed in a network of hardware, and this library will serve as the low-level enabler for such distributed machine learning capabilities. The value of this project lies with providing unique AI capabilities for very specific hardware that is becoming part of the AI revolution currently in progress, which contributes to a number of LF AI & Data goals and projects.

 

BeyondML is a framework for developing sparse neural networks that can perform multiple tasks across multiple data domains. This framework provides value to the community by: simplifying development and deployment of advanced machine learning capabilities for use on low-end devices and in dynamic environments characteristic of the resource-constrained edge, reducing complexity and costs of deploying ML models or systems of models to cloud platforms, and reducing carbon footprint of deployed ML models.

Contributed by AI Squared in June 2022, BeyondML provides the community with an open way to train and utilize ML models which are economical (due to the lessened computational requirements compared to the state-of-the-art in deep neural networks) and simple to deploy (due to the ability of BeyondML to package multiple models/tasks within a single neural network), making ML more accessible for all members of the community. 

 

 

FlagAI (Fast LArge-scale General AI models) is a fast, easy-to-use and extensible toolkit for large-scale models. This project was contributed by the Beijing Academy of Artificial Intelligence (BAAI) in June 2022. The goal of the project is to support training, fine-tuning, and deployment of large-scale models on various downstream tasks with multi-modality. 

Currently, the project is focusing on NLP models and tasks. In the near future, it will support other modalities. At this time, it supports GLM, BERT, RoBERTa, GPT2, T5, and models from Huggingface Transformers. It also provides APIs to quickly download and use those pre-trained models on a given text, and fine-tune them on individual datasets. These models can be applied to text for tasks like text classification, information extraction, question answering, summarization, and text generation. FlagAI is backed by the three most popular data/model parallel libraries — PyTorch/Deepspeed/Megatron-LM — with seamless integration between them. You can parallel your training/testing process with less than ten lines of code.

LF AI & Data supports projects via a wide range of services, and the first step is joining as a hosted Project. Learn more about these projects on their GitHub repositories and join their respective Announce Mailing Lists and Technical-Discuss Mailing Lists, listed below. 

A warm welcome to these new projects! We look forward to the projects’ continued growth and success as part of the LF AI & Data Foundation. To learn about how to host an open source project with us, visit the LF AI & Data website.

1ChipML Key Links

BeyondML Key Links

FlagAI Key Links

LF AI & Data Resources

Access other resources on LF AI & Data’s GitHub or Wiki

Erin Thacker