Meta's Layoffs Leave Supernatural Fitness Users in Mourning Users of the VR fitness service are distraught that Supernatural has had its staff cut and won't receive any more content updates. I hear a stranger's heavy breathing through the rollicking dude-bro anthem blasting my eardrums, courtesy of the pop-rock band Imagine Dragons. Me and two people I just met are punching digital blocks that fly at our heads in the VR workout platform Supernatural . My new friends have nameplates floating above their heads that say Chip and Alisa. That's all I know about them.

The Transformer architecture has significantly advanced natural language processing (NLP) and has been foundational in developing large language models (LLMs) such as LLaMA and OPT, which have come to dominate a broad range of NLP tasks. Despite their superior accuracy, LLMs present unique challenges in practical inference, concerning the compute and memory-intensive nature. Thanks to the autoregressive characteristic of LLM inference, KV caching for the attention layers in Transformers can effectively accelerate LLM inference by substituting quadratic-complexity computation with linear-complexity memory accesses. Yet, this approach requires increasing memory as demand grows for processing longer sequences. The overhead leads to reduced throughput due to I/O bottlenecks and even out-of-memory errors, particularly on resource-constrained systems like a single commodity GPU. In this paper, we propose ALISA, a novel algorithm-system co-design solution to address the challenges imposed by KV caching. On the algorithm level, ALISA prioritizes tokens that are most important in generating a new token via a Sparse Window Attention (SWA) algorithm. SWA introduces high sparsity in attention layers and reduces the memory footprint of KV caching at negligible accuracy loss. On the system level, ALISA employs three-phase token-level dynamical scheduling and optimizes the trade-off between caching and recomputation, thus maximizing the overall performance in resource-constrained systems. In a single GPU-CPU system, we demonstrate that under varying workloads, ALISA improves the throughput of baseline systems such as FlexGen and vLLM by up to 3X and 1.9X, respectively.

Moscow-based search giant Yandex has revealed a Russian speaking smart speaker to take on Amazon. The $160 device will work with its digital assistant'Alice', becoming the latest challenger to take on the leading voice-activated home helpers from Silicon Valley. Yandex is targeting the Russian-speaking market with its Yandex.Station speaker, as well as a platform on which third-party developers can programme Alice to order pizza, check mobile phone bills or buy plane tickets. The $160 device will work with its digital assistant'Alice', becoming the latest challenger to take on the leading voice-activated home helpers from Silicon Valley. The Yandex Station has Bluetooth capabilities and an HDMI port for streaming television on connected displays, and has including two 10W audio drivers and a 30W woofer.

Alisa, a virtual reality assistant developed by Russia's tech giant Yandex, has been nominated to become the next president of Russia by thousands of supporters across the country. The favorite for elections scheduled for March 2018, President Vladimir Putin, announced his candidacy on Wednesday. Putin is seeking a fourth term in office which would extend his tenure into 2024. Other contenders for the presidency will likely include familiar political figures from the pro-Kremlin Communist Party and the nationalist Liberal Democratic Party, as well as new faces such as the former reality TV star Ksenia Sobchak and the business-oriented Party of Growth leader Boris Titov. It appears they will now be challenged by Alisa's progressive promise to bring "the political system of the future, built exclusively on rational decisions made on the basis of clear algorithms," the Lenta news website reported.

A virtual assistant, Alisa, is throwing its name in the hat to run against incumbent Vladmir Putin in the 2018 Russian presidential elections. First, let's just dismiss this idea as stupid. Okay, now let's give it a second look – because it actually makes a lot of sense. Maybe it's time to, academically at least, consider alternative political systems based on more logical and rational thought processes – like one that could elect Russia's Alisa. So far, the virtual assistant has over 80,000 "votes" from citizens requesting a place for it on the ballot next year.

Russia's next president could be an artificially intelligent robot that claims'enemies of the people will be shot'. Forty thousand Russians have nominated a piece of AI software on their phones to stand against Vladimir Putin for the 2018 Russian presidency. The AI assistant known as Alisa, similar to the Apple's voice-activated Siri, was created by Russian technology company Yandex. Russia's next president could be an artificially intelligent robot that claims'enemies of the people will be shot'. Since the AI's launch in September, Alisa has stirred controversy on social media, with users sharing a series of contentious statements from the software.