16 Mar 2022

Open-Source Natural Language Processing: EleutherAI’s GPT-J

OpenAI released GPT-3 in June of 2020  – and the results were instantly compelling. By 2021, the MIT Technology Review was proclaiming OpenAI’s GPT-3 “a big step toward AI that can understand and interact with the human world.” It has since commercialized in collaboration with Microsoft.   In response, EleutherAI – a self-described “grassroots collective of researchers working to open-source AI research”  launched GPT-J in July 2020 as a quest to replicate the OpenAI GPT collection of models. The goal is to “break the OpenAI-Microsoft monopoly” through broadening availability and the collective intelligence of open-source development of a competing class of GPT models.  

Read More
07 Feb 2022

The Current AI Innovation Hype Cycle: Large Language Models, OpenAI’s GPT-3 and DeepMind’s RETRO

For better or for worse, Large Language Models (LLMs) – used for natural language processing by commercial AI Platform-as-a-Service (PaaS) subscription offerings – have become one of the first “big data” applied technologies to become a crossover hit in the AI marketplace. OpenAI, a San Francisco-based research and deployment company, released GPT-3 in June of 2020 – and the results were instantly compelling. By 2021, the MIT Technology Review was proclaiming OpenAI’s GPT-3 a top 10 breakthrough technology and “a big step toward AI that can understand and interact with the human world.”

Read More
25 Jan 2022

TinyML is bringing deep learning models to microcontrollers

Deep learning models owe their initial success to large servers with large amounts of memory and clusters of GPUs. The promises of deep learning gave rise to an entire industry of cloud computing services for deep neural networks. Consequently, very large neural networks running on virtually unlimited cloud resources became

Read More
15 Mar 2019

Intel offers AI breakthrough in quantum computing

While applications of deep learning have produced impressive empirical results, the theory behind it is still poorly developed. However, a team of researchers from Intel and the Hebrew University are saying that they have made an important breakthrough that allows for a better theoretical understanding of the capabilities of deep

Read More