Toggle light / dark theme

Microsoft research releases Phi-2 and promptbase.

Phi-2 outperforms other existing small language models, yet it’s small enough to run on a laptop or mobile device.


Over the past few months, our Machine Learning Foundations team at Microsoft Research has released a suite of small language models (SLMs) called “Phi” that achieve remarkable performance on a variety of benchmarks. Our first model, the 1.3 billion parameter Phi-1 (opens in new tab), achieved state-of-the-art performance on Python coding among existing SLMs (specifically on the HumanEval and MBPP benchmarks). We then extended our focus to common sense reasoning and language understanding and created a new 1.3 billion parameter model named Phi-1.5 (opens in new tab), with performance comparable to models 5x larger.

We are now releasing Phi-2 (opens in new tab), a 2.7 billion-parameter language model that demonstrates outstanding reasoning and language understanding capabilities, showcasing state-of-the-art performance among base language models with less than 13 billion parameters. On complex benchmarks Phi-2 matches or outperforms models up to 25x larger, thanks to new innovations in model scaling and training data curation.

Ukraine’s largest mobile phone operator said Russia was most likely behind a “powerful” cyberattack that disrupted phone and internet services Tuesday for about 24 million people in the country.

The cyberattack, the largest to strike Ukraine since Russia’s invasion began nearly two years ago, took down ATMs and digital banking services, disabled some air raid sirens, and hampered broader cellular service across Ukraine.

“This is definitely a cyberattack and the probability that Russian entities are behind it is very close to 100%,” Oleksandr Komarov, chief executive officer of Kyivstar GSM JSC, told Bloomberg by phone.

Scientists from the Leibniz Institute for Astrophysics Potsdam (AIP) have discovered a new plasma instability that promises to revolutionize our understanding of the origin of cosmic rays and their dynamic impact on galaxies.

At the beginning of the last century, Victor Hess discovered a new phenomenon called that later on earned him the Nobel prize. He conducted high-altitude balloon flights to find that the Earth’s atmosphere is not ionized by the radioactivity of the ground. Instead, he confirmed that the origin of ionization was extra-terrestrial. Subsequently, it was determined that cosmic “rays” consist of charged from flying close to the speed of light rather than radiation. However, the name “cosmic rays” outlasted these findings.

In the new study, Dr. Mohamad Shalaby, scientists at AIP and the main author of this study, and his collaborators have performed to follow the trajectories of many cosmic ray particles and study how these interact with the surrounding plasma consisting of electrons and protons. The paper appears on the pre-print server arXiv.