Bigger isn’t always better: Microsoft is also launching a ‘mini-AI’ model

In news: The Phi-3 Mini will be the first of those small AI models that Microsoft will launch.

• According to Eric Boyd, vice president of Azure, Microsoft’s AI platform, Phi-3 Mini will be “as powerful as Major Language Frameworks such as ChatGPT-3.5 (the current free version of ChatGPT, red.), but on a smaller scale,” he explains Until.
• In addition to the Mini, there will also be small and medium models. Each model has a different number of ‘parameters’, which indicate how many complex instructions it can handle. For the mini version this is limited to 3.8 billion variables. Phi-3 Mini is available for download through Azure, Hugging Face and Ollama.
• Phi types were put together by a certain “curriculum,” according to Boyd. Thus, they were trained to make children’s stories. “There aren’t enough children’s books, so we took a list of over 3,000 words and asked the LLM to create ‘children’s books’ for teaching Phi.”

Zoom in: Why are tech companies bringing small AI models to market?

• Boyd believes that smaller models like Phi-3 work better for specific business applications because their internal data sets are smaller. And often cheaper, because these models use less computer power.
• Other companies preceded Microsoft. Google launched Gemma, a lower version of Gemini. Meta also released a small version of its Llama and Claude 3 Haiku design that can quickly summarize scientific documents.
• In this way, technology companies are changing their AI models, and the perception that AI is difficult and heavy is being removed. The reach is increasing, because as an end user you know what you can use a particular example of AI.