On Tuesday, Microsoft introduced a lightweight Artificial Intelligence model since the tech giant is looking forward to attract a wider audience or client base by incorporating cost effective options.
Get to know more about Microsoft’s lightweight AI model
The new version of the lightweight AI model is called Phi-3 mini and it is the first of the three small language models (SLM) that is to be launched by the company. Through these small language models, the company aims to stake it’s future on a technology that expects to have a wide-ranging impact on the way the world and the people work.
Sébastien Bubeck, Microsoft’s vice president of GenAI research, said, “Phi-3 is not slightly cheaper, it’s dramatically cheaper, we’re talking about a 10x cost difference compared to the other models out there with similar capabilities.”
According to reports, the company said that small language models are designed to perform easier and simpler tasks smoothly. They make things easier for certain companies that have limited resources.
Where will Phi-3 mini be available?
The company reported that Microsoft’s lightweight AI model, Phi-3 will be available immediately on Microsoft’s cloud service platform, Azure’s AI model catalog, machine learning model platform Hugging Face and Ollama which is a framework for running models on a local machine.
Besides, the small language model will also be available on Nvidia Inference Microservices (NIM), Nvidia’s software tool. The SLM has also been optimized for its GPUs ( graphics processing units)
Phi-3 mini will be extremely useful in regulating tasks in a smoother way for companies with limited resources. Besides, the best part about the Phi-3 mini is that the SLM is an affordable and economically -friendly option.
Last week, Microsoft invested $1.5 billion in G42 which is a UAE -based Artificial Intelligence firm. The company had also tied up with Mistral AI, a French startup to help the models be available through the Azure cloud computing platform.
For more information please keep reading techinnews
