California’s Nucleus AI Unveils 22B-Parameter LLM for Agriculture
California-based Nucleus AI has unveiled a groundbreaking 22-billion-parameter large language model (LLM). Developed by a team that hails experience from tech giants like Amazon and Samsung Research, this model promises to be a game-changer in AI.
Nucleus AI’s 22-billion-parameter Large Language Model (LLM) is accessible through both an open-source MIT license and a commercial license. It offers flexibility for various generative tasks, bridging the gap between 13-billion and 34-billion-parameter models.
The company is developing an intelligent operating system to address the agricultural sector’s critical challenges. It focuses on optimizing supply and demand, mitigating uncertainties for farmers, and addressing climate change-related issues. It aims to fill farmers’ knowledge gaps and optimize the agricultural supply chain by delivering timely information.
The development journey of the 22B LLM involved training on a vast dataset, including a trillion tokens of data sourced from various repositories, such as the web, Wikipedia, Stack Exchange, arXiv, and code. This comprehensive knowledge base forms the foundation for the model’s capabilities.
In the upcoming weeks, the company will launch its RetNet models, which are anticipated to deliver substantial cost savings and improved inference speeds.
“Right now, we’re not competing against anybody else’s algorithms. When we got access to compute, we were trying to build internal products to step into the farming landscape. But then we figured we need language models as the core of the marketplace itself and started building that with the contribution from the open-source community,” Gnandeep Moturi, the CEO shared.
