Scientists have discovered a new brain glitch that is the exact opposite of deja vu. While deja vu is the unsettling sense that you've lived a moment before, jamais vu is when something familiar suddenly feels alien -- like encountering it for the very first time. You've likely felt it: walking through your hometown and suddenly feeling lost, or repeating a common word until it sounds strange and meaningless. Repetition is often the trigger. The brain, overloaded by familiarity, short-circuits, making the ordinary feel bizarre.

Large language models (LLMs) with hundreds of billions of parameters have sparked a new wave of exciting AI applications. However, they are computationally expensive at inference time. Sparsity is a natural approach to reduce this cost, but existing methods either require costly retraining, have to forgo LLM's in-context learning ability, or do not yield wall-clock time speedup on modern hardware. We hypothesize that contextual sparsity, which are small, input-dependent sets of attention heads and MLP parameters that yield approximately the same output as the dense model for a given input, can address these issues. We show that contextual sparsity exists, that it can be accurately predicted, and that we can exploit it to speed up LLM inference in wall-clock time without compromising LLM's quality or in-context learning ability. Based on these insights, we propose DejaVu, a system that uses a low-cost algorithm to predict contextual sparsity on the fly given inputs to each layer, along with an asynchronous and hardware-aware implementation that speeds up LLM inference. We validate that DejaVu can reduce the inference latency of OPT-175B by over 2X compared to the state-of-the-art FasterTransformer, and over 6X compared to the widely used Hugging Face implementation, without compromising model quality. The code is available at https://github.com/FMInference/DejaVu.

A record number of NFL players tested positive for COVID-19 last week, leading to postponed games. LAS VEGAS, NEVADA – Football fans got double the action Monday night and will again Tuesday night. That's because the NFL rearranged its Week 15 schedule as COVID-19 cases surge among the players, 100 of whom tested positive over three days last week. For the second year in a row, professional sports teams aren't just worried about their opponent -- they're worried about whether there are enough players to play the game. "Just like anywhere else, you have to think about can you operate your business if everybody is sick?" said Brian Labus, infectious disease epidemiologist and assistant professor at the University of Nevada-Las Vegas.

At first glance, the letter from the Delhi police commissioner's desk could have easily been dismissed as another routine laundry list of his department's "achievements" in the previous year. A closer look at the letter, written a little over two years ago, would have sprung a pleasant surprise in the context of the city police's technology prowess. The Delhi Police, according to the letter, had partnered with the Indian Space Research Organisation to implement CMAPS--Crime Mapping, Analytics and Predictive System--under the "Effective use of Space Technology-based Tools for Internal Security Scheme" initiated by Prime Minister Narendra Modi in 2014. CMAPS generates crime-reporting queries and has the capacity to identify crime hotspots by auto sweep on the Dial 100 database every 1-3 minutes, replacing a Delhi Police crime-mapping tool that involved manual gathering of data every 15 days. It performs trend analysis, compiles crime and criminal profiles and analyses the behaviour of suspected offenders--all with accompanying graphics.