Major players introduce small language models to make natural language processing more accessible

From VentureBeat: 2024-07-19 17:08:44

Three major players in AI, including Hugging Face, Nvidia with Mistral AI, and OpenAI, have introduced small language models (SLMs) aiming to make advanced natural language processing more accessible. This shift challenges the focus on large neural networks and could revolutionize AI implementation.

Hugging Face’s SmolLM models are designed to run directly on mobile devices, offering sizes from 135 million to 1.7 billion parameters. This move towards edge computing addresses privacy and latency concerns, paving the way for new applications with minimal latency and enhanced privacy.

Nvidia and Mistral AI’s Mistral-Nemo model, with 12 billion parameters and a 128,000 token context window, targets desktop computers. By democratizing access to advanced AI capabilities, Mistral-Nemo could lead to a surge in AI-powered applications across industries, from customer service to data analysis tools.

OpenAI’s GPT-4o Mini is a cost-efficient SLM, priced at just 15 cents per million tokens for input and 60 cents per million for output. This pricing strategy could drive AI innovation among startups and small businesses, lowering barriers to AI adoption and accelerating technological advancements.



Read more at VentureBeat: AI’s new frontier: Hugging Face, Nvidia, and OpenAI lead charge in small language models