Image Source : FILE Microsoft introduces advanced AI ‘Mini Language Model’ catering to the research community
Microsoft has launched its latest compact ‘small language mannequin’ which was titled ‘Phi-2’. The new compact mannequin will proceed to carry out at par and claims to be higher than sure bigger open-source Llama 2 fashions which have lower than 13 billion parameters.
What is Phi?
In the previous few months, the Machine Learning Foundations group at Microsoft Research has launched a set of small language fashions (SLMs) referred to as “Phi” which has achieved a outstanding efficiency on quite a lot of benchmarks.
About Phi-1
The first mannequin, the 1.3 billion parameter Phi-1 will obtain state-of-the-art efficiency on Python coding amongst current SLMs (particularly on the HumanEval and MBPP benchmarks).
In an replace, the firm mentioned, “We at the moment are releasing Phi-2, a 2.7 billion-parameter language mannequin that demonstrates excellent reasoning and language understanding capabilities, showcasing state-of-the-art efficiency amongst base language fashions with lower than 13 billion parameters.”
About Phi-2
Phi-2 is alleged to be a perfect playground for researchers, which incorporates exploration round mechanistic interpretability, fine-tuning experimentation and security enhancements on quite a lot of duties.
Microsoft mentioned, “We have made Phi-2 obtainable in the Azure AI Studio mannequin catalog to foster research and improvement on language fashions.”
It was additional mentioned, that the improve in the dimension of language fashions to tons of of billions of parameters has unlocked a number of rising capabilities- redefining the panorama of pure language processing.
However, a query stays whether or not such emergent talents might be achieved at a smaller scale utilizing strategic selections for coaching, e.g., information choice.
Phi fashions intention to reply this query by coaching SLMs
Microsoft additional said: “Our line of labor with the Phi fashions goals to reply this query by coaching SLMs that obtain efficiency on par with fashions of a lot larger scale (but nonetheless removed from the frontier fashions).”
The firm has carried out in depth testing on generally used prompts from the research community.
The tech big has additional mentioned, “We noticed a behaviour in accordance with the expectation we had given the benchmark outcomes.”
ALSO READ: Global good private audio market drops down by 3%, Apple leads
Inputs from IANS
Latest Technology News
https://www.indiatvnews.com/technology/news/microsoft-unveils-powerful-ai-compact-language-model-tailored-for-research-applications-2023-12-17-907665
