Accurate, Focused Research on Law, Technology and Knowledge Discovery With Daily Postings Since 2002
Via LLRX – When scientific citations go rogue: Uncovering ‘sneaked references’ – Reading and writing articles published in academic journals and presented at conferences is a central part of being a researcher. When researchers write a scholarly article, they must cite the work of peers to provide context, detail sources of inspiration and explain differences in …
Spreadsheet LLM – Encoding Spreadsheets for Large Language Models: “Spreadsheets are characterized by their extensive two-dimensional grids, flexible layouts, and varied formatting options, which pose significant challenges for large language models (LLMs). In response, we introduce SpreadsheetLLM, pioneering an efficient encoding method designed to unleash and optimize LLMs’ powerful understanding and reasoning capability on spreadsheets. …
Data is Plural: “Since launching in March 2022, homepages.news has archived millions of screenshots, performance audits, robots.txt files, accessibility trees, and hyperlink lists from the homepages of 1,100+ news sites. The open-source project, run by journalist Ben Welsh, provides bulk data for each of those assets. The screenshots themselves are stored on the Internet Archive; …
404 Media -“We’ve written a lot over the past year about how bad Google Search is getting, but it’s still incredibly useful for searching Reddit. Add “Reddit” to the end of any search query, and chances are you’ll find a useful thread. Due to a $60 million deal that allows Google to scrape Reddit for …
The Scholarly Kitchen: “There is broad consensus in scholarly publishing that AI tools will make the task of ensuring the integrity of the scientific record a Herculean task. However, it seems that many publishers are still struggling to figure out how to address the new issues and challenges that these AI tools present. Current publisher …
MIT Technology Review: “AI models work by training on huge swaths of data from the internet. But as AI is increasingly being used to pump out web pages filled with junk content, that process is in danger of being undermined. New research published in Nature shows that the quality of the model’s output gradually degrades …
Wired [unpaywalled]- This Machine Exposes Privacy Violations. A former Google engineer has built a search engine, WebXray, that aims to find illicit online data collection and tracking—with the goal of becoming “the Henry Ford of tech lawsuits.”…It’s a search engine for rooting out specific privacy violations anywhere on the web. By searching for a specific …
Via Mastodon – Ludo Waltman@[email protected] Scientific Director and Professor of Quantitative Science Studies @cwts, @universiteitleiden; Open Science Ambassador @universiteitleiden; Co-chair @RoRInstitute; President @ASAPbio; Editor-in-Chief MetaROR (MetaResearch Open Review); Former Editor-in-Chief @QSS_ISSI University libraries now have easy access to the information they need to include openness of publication metadata in their negotiations with publishers. Try it …
The Economist [unpaywalled] – All titles on this eclectic [fiction and non fiction] and wide ranging list were published in America or Britain in 2024. For more on the latest books, films, TV shows, albums and controversies, sign up to Plot Twist, a weekly subscriber-only newsletter.
Meta: “Llama 3.1 405B is the first openly available model that rivals the top AI models when it comes to state-of-the-art capabilities in general knowledge, steerability, math, tool use, and multilingual translation. With the release of the 405B model, we’re poised to supercharge innovation—with unprecedented opportunities for growth and exploration. We believe the latest generation …
TechCrunch: “Meta’s latest open source AI model is its biggest yet. Today, Meta said it is releasing Llama 3.1 405B, a model containing 405 billion parameters. Parameters roughly correspond to a model’s problem-solving skills, and models with more parameters generally perform better than those with fewer parameters. At 405 billion parameters, Llama 3.1 405B isn’t …
“…To date, NewsGuard’s team has identified and is tracking 14 misinformation narratives relating to the 2024 Paris Olympics in 13 languages: Dutch, English, French, German, Greek, Hungarian, Italian, Polish, Portuguese, Romanian, Russian, Spanish, and Turkish. The claims have spread on social media as well as on 31 unreliable news and information websites. Sixteen of the 31 sites have …
