Microsoft has launched the next version of its lightweight AI model Phi-3-mini, the first of three small models the company plans to release.
The Phi-3-mini measures 3.8 billion parameters and is trained on a data set that is smaller relative to large language models like GPT-4.
According to The Verge, it is now available on Azure, Hugging Face, and Ollama. Microsoft plans to release Phi-3 Small (7bn parameters) and Phi-3 Medium (14bn parameters). (Parameters refer to how many complex instructions a model can understand)
The company Phi-2, released in December, performed just as well as bigger models like Llama 2.
Microsoft says Phi-3 performs better than the previous version and can provide responses close to how a model 10 times bigger than it can, The Verge reported.
Phi-3-mini is as capable as LLMs like GPT-3.5 “just in a smaller form factor,” Eric Boyd, corporate vice president of Microsoft Azure AI Platfom, told The Verge.
Small AI models are often less expensive to run and perform more better on personal devices like phones and laptops as juxtaposed to their larger counterparts.
Developers trained Phi-3 with a “curriculum”, says Boyd. They took inspiration from how children learned from bedtime tales, books with easy words as well as sentence structures about larger topics.
“There aren’t enough children’s books out there, so we took a list of more than 3,000 words and asked an LLM to make ‘children’s books’ to teach Phi,” Boyd says.
