HyperGAI, an applied AI research company innovating in multimodal generative AI technology and solutions, emerges from stealth mode today and releases a new open multimodal large language model which is among the best-in-class models for its size to date and will be made available to public through open-source.
The company has been operating in stealth since the last several months, progressively building their in-house multimodal foundation models, and aiming to develop the leading multimodal generative AI system that emulates the digital brain for hyper general intelligent machines. HyperGAI believes that true general intelligence requires more than just text-based understanding.
“Our goal is to develop a ‘full stack’ multimodal generative AI system that can understand all kinds of inputs and generate any type of output, which becomes the digital brain for consumer and enterprise products that empower individuals to unlock their creativity and improve workplace productivity,” said Steven Hoi, Ph.D., Founder and CEO of HyperGAI. “Unlike other AI companies focused on applications, we are distinctive in our approach of driving the development of the multimodal foundation models, and we’ll leverage this to continue delivering breakthrough AI products and solutions.”
The company is approaching their goal in two phases. The first includes building two types of multimodal foundation models:
- Multimodal LLM models that understand multiple types of sensory inputs, including text, images, videos; and
- Multimodal diffusion models that leverage the understanding ability of multimodal LLM to train multimodal generation models for achieving personalised content creation, including photos, videos, and more.
During the second phase the focus will center on building the end-to-end multimodal generative AI system by integrating both types of multimodal foundation models into a holistic framework, which will be the key to achieving the goal of general artificial intelligence.
The company believes that an open approach to advancing multimodal generative AI is necessary, and it will be more effective, sustainable, and responsible to accelerate the revolution of frontier general artificial intelligence technology through a joint community-based effort.
As part of HyperGAI’s ongoing progress, they are also releasing HPT (Hyper-Pretrained Transformers), their new family of the most capable multimodal large language model (LLM) for its size to date. It’s interesting to note that the company is now open sourcing their efficient HPT Air model for both research and commercial use.
HyperGAI was founded by a group of visionary leaders with more than 20 years of proven expertise in multimodal AI, computer vision, natural language processing, and deep learning, and who together have published more than 300 technical articles.
The post HyperGAI exits stealth mode: releases multimodal LLM first appeared on AI-TechPark.
#AI [Source: AI Techpark]