Microsoft has recently introduced a new small language model called Phi-3 mini, which is optimized for use on modern smartphones and offers similar performance to OpenAI’s GPT-3.5. This updated version of Microsoft’s lightweight language model was trained with 3.3 billion tokens from larger and more advanced datasets compared to its predecessor, Phi-2, which was trained with 1.4 billion tokens. The result is a model with 3.8 billion parameters that can fit on a modern smartphone, taking up approximately 1.8GB of memory and can be quantified to 4 bits.
The researchers tested the Phi-3 mini model on an iPhone 14 with an A16 Bionic chip, where it ran natively and offline at a speed of over 12 tokens per second. The overall performance of this model is comparable to larger models like Mixtral 8x7B and GPT-3.5, making it a versatile option for various applications. It uses a transformer decoder architecture that supports 4K text length and is based on a block structure similar to Meta’s Llama 2, supporting packages developed for Llama 2.
Phi-3 mini is designed for conversational chat formats and aligns with Microsoft’s values of robustness and security. Along with Phi-3 mini, Microsoft has also trained two other models in the same family: Phi-3 medium with 14 billion parameters and Phi-3 small with 7 billion parameters, both trained with 4.8 billion tokens. The company is focused on providing efficient and secure language models for a variety of applications.
Overall, the introduction of Phi-3 mini marks an exciting development in the field of natural language processing (NLP) technology, as it provides a powerful yet lightweight solution for mobile devices that can deliver high performance while still being easy to use on smaller screens.
In summary, Microsoft has launched a new small language model called Phi-3 mini that runs on modern smartphones at high speeds while offering similar performance to larger models like Mixtral 8x7B and GPT-3.5. With its focus on conversational chat formats and security measures in place, this updated version of Microsoft’s lightweight language model will provide developers with an efficient way to build robust NLP applications across various industries.
As AI continues to evolve at an unprecedented pace, it’s clear that companies like Microsoft are working hard to keep up with these advancements by creating cutting-edge solutions like Phi-