--The increasing adoption of large language models (LLMs) with extended context windows necessitates efficient Key-V alue Cache (KVC) management to optimize inference performance. We analyze real-world KVC access patterns using publicly available traces and evaluate commercial key-value stores like Redis and state-of-the-art RDMA-based systems (CHIME [1] and Sherman [2]) for KVC metadata management. Our work demonstrates the lack of tailored storage solution for KVC prefilling, underscores the need for an efficient distributed caching system with optimized metadata management for LLM workloads, and provides insights into designing improved KVC management systems for scalable, low-latency inference. Large Language Models (LLMs) have shown remarkable ability in tasks like text generation, translation, and question-answering, but their attention architecture introduces significant challenges. The use of key-value caches (KVC) in attention layer of transformer models, while essential for efficient token generation, requires substantial memory resources.

The world's richest man, Elon Musk, has vowed to oversee a radical hollowing out of government agencies, asserting this week that some should be "deleted entirely" as he defunds public programs and lays off federal workers. While the immense cuts are framed as a means of removing waste, they may also become a boon to private companies – including Musk's own businesses – that the government increasingly relies on for many of its key initiatives. Musk and his allies in the "department of government efficiency" (Doge), the unofficial committee acting as the operations arm of his cost-cutting efforts, have targeted a range of major government departments. They have moved to close the United States Agency for International Development, slashed the Department of Education and taken over the General Services Administration that controls federal IT structures. Doge staffers have also gained access to the treasury department, as well as set their sights on the Department of Defense, energy department, Environmental Protection Agency and at least a dozen others.

B.T. had proven easy enough to find. When he went dark, Lily figured he was in one of the world's three gambling capitals--Vegas, Monte Carlo, or Macau. It only became a matter of checking with a handful of five-star hotels in each, something Sherman was happy to handle for her. In the days since Castro's death, it was Sherman who'd stoked Lily's concern for B.T. Each morning, Sherman came in parroting another conspiracy theory as to who or what was behind the president's untimely demise. He'd even gone so far as to place a #TRUTHNOTDREAMS sticker adjacent to the US Marine Corps sticker on his wheelchair.

In elementary school, I remember seeing a Scholastic visual of what the world would look like by 2010. While we still are working on flying cars, the artist illustrated artificial intelligence (AI) somewhere in that futuristic world. Many marketers today shy away from digging deeper into AI use cases because their brand is still working out how to use essential media and basic website tools effectively. I am here to tell you that it's simpler than it sounds and can significantly impact your brand's line, especially with social media use declining. If a client asked me whether to build new social media channels or integrate AI into their existing strategy, I would lean towards AI development.

The Defense Department must become a digital and artificial intelligence-enabled enterprise capable of operating at the speed and scale necessary to preserve its military advantage, according to a memorandum issued by Deputy Secretary of Defense Kathleen H. Hicks. The memorandum, published on defense.gov, John Sherman, DOD chief information officer, will serve as the acting chief digital and artificial intelligence officer until the position is filled permanently. "[It's] an honor to be able to help get this organization stood up while performing my chief information officer duties," Sherman said today in a Pentagon media roundtable, adding that he has worked closely with several organizations to make sure the CDAO effort is launched on a solid footing. "This is a key milestone for the department to become a digital AI-enabled enterprise," a senior DOD official said in the roundtable.

The Pentagon's chief information officer will also serve as the head of a new organization overseeing the Defense Department's various digital and artificial intelligence efforts, the department announced Feb. 2. DoD Chief Information Officer John Sherman will serve as the acting chief digital and artificial intelligence officer, or CDAO, a newly created office designed to oversee the Defense Digital Service, the Joint Artificial Intelligence Center and the CIO office he was already leading. The new office was established to better align a number of data, analytics, digital solutions and AI efforts across the DoD. Previously, all three of those offices reported directly to the deputy defense secretary. Sherman will serve as DoD CIO and CDAO as the Pentagon continues to look for a director. "I'm honored to be able to help get this organization stood up, again, while performing my chief information officer duties and also serving as the acting CDAO," Sherman said.

The Defense Department's Chief Digital and Artificial Intelligence Office, a new hub to align disparate AI-centered pursuits across the vast enterprise, officially reached initial operating capacity this week--but much must still be puzzled out before it's totally realized this summer. John Sherman, DOD's recently Senate-confirmed chief information officer, will play a major role in seeing it through. He's taking the office's lead as acting chief digital and AI officer until the department completes its search for the right person to fill this first-of-a-kind position. "In addition to getting the OCDAO up and running for [full operational capacity], rest assured we'll remain laser-focused on our CIO duties--cybersecurity, digital modernization and other areas the department relies on us for," Sherman told reporters during a press call on Wednesday. He and two other senior defense officials shared fresh details about the new unit's establishment and what it's ultimately meant to accomplish.

Technology investment company and incubator Red Cell Partners announced today the launch of Zephyr AI, a company that leverages large data sets to inform both clinical care and the development of new targeted precision therapies. The management team of the new company consists of CEO Yisroel Brumer, formerly of the office of the Secretary of Defense; Executive Chairman Grant Verstandig, who most recently served Chief Digital Officer at UnitedHealth Group; and Chief Technology Officer Jeff Sherman, who was the machine learning architect at Rally Health, which was acquired in 2017 by UnitedHealth's Optum unit. According to a press release announcing its launch, Zephyr AI will look to improve patient outcomes while lowering costs by integrating "artificial intelligence with extensive datasets to upend traditional'guess and test' drug development and personalized medicine processes to unearth novel therapeutics, new applications for existing therapeutics, and advanced biomarkers for individualized treatments." The potential new company gave a hint at its direction earlier in the year via the publication of two papers by the founders in the journal Oncogene that detailed the company's technology and it's performance. "These findings demonstrate that Zephyr AI can already identify novel-use cases for existing therapeutics in cancer," company CTO Sherman.

Late in 2019, the Pentagon chose Microsoft for a $10 billion contract called JEDI that aimed to use the cloud to modernize US military computing infrastructure. Tuesday, the agency ripped up that deal. The Pentagon said it will start over with a new contract that will seek technology from both Amazon and Microsoft, and that offers better support to data-intensive projects, such as enhancing military decisionmaking with artificial intelligence. The new contract will be called the Joint Warfighter Cloud Capability. It attempts to dodge a legal and political mess that had formed around JEDI. Microsoft competitors Amazon and Oracle both claimed in lawsuits that the award process had been skewed.

Now that this attention-grabbing headline has drawn you in, let me clarify. Data scientists should not partake in illegal drugs. Data scientists should participate in pharmacological research, as artificial intelligence and machine learning can add value, even when the data scientist does not have a background or training in physics, biology, chemistry, or medicine. The CAIA Association and FDP Institute had a recent conversation with Woody Sherman, the CSO of Silicon Therapeutics. While many of us can be left behind in a discussion of computational drug discovery, it seems that almost everyone today is a budding epidemiologist trying to better understand the prevention and spread of COVID-19, so let's continue.