NFL playoffs: Patriots went 3-0 with AFC Championship referee in 2025; Rams are 7-0 with NFC title game ref
https://www.cbssports.com/nfl/news/nfl-playoffs-patriots-rams-refer…
DOJ must release Epstein files by Friday or risk repercussions, law's co-author says (NBC News)
https://www.nbcnews.com/politics/justice-department/doj-must-release-epstein-files-friday-risk-repercussions-laws-co-autho-rcna249574
http://www.memeorandum.com/251217/p75#a251217p75
Exploiting ID-Text Complementarity via Ensembling for Sequential Recommendation
Liam Collins, Bhuvesh Kumar, Clark Mingxuan Ju, Tong Zhao, Donald Loveland, Leonardo Neves, Neil Shah
https://arxiv.org/abs/2512.17820 https://arxiv.org/pdf/2512.17820 https://arxiv.org/html/2512.17820
arXiv:2512.17820v1 Announce Type: new
Abstract: Modern Sequential Recommendation (SR) models commonly utilize modality features to represent items, motivated in large part by recent advancements in language and vision modeling. To do so, several works completely replace ID embeddings with modality embeddings, claiming that modality embeddings render ID embeddings unnecessary because they can match or even exceed ID embedding performance. On the other hand, many works jointly utilize ID and modality features, but posit that complex fusion strategies, such as multi-stage training and/or intricate alignment architectures, are necessary for this joint utilization. However, underlying both these lines of work is a lack of understanding of the complementarity of ID and modality features. In this work, we address this gap by studying the complementarity of ID- and text-based SR models. We show that these models do learn complementary signals, meaning that either should provide performance gain when used properly alongside the other. Motivated by this, we propose a new SR method that preserves ID-text complementarity through independent model training, then harnesses it through a simple ensembling strategy. Despite this method's simplicity, we show it outperforms several competitive SR baselines, implying that both ID and text features are necessary to achieve state-of-the-art SR performance but complex fusion architectures are not.
toXiv_bot_toot
Scientists at NeurIPS, which drew a record 26,000 attendees this year, say key questions about how AI models work and how to measure them remain unresolved (Jared Perlo/NBC News)
https://www.nbcnews.com/tech/tech-news/ai-progress-surges-resea…
Trump says he will stay out of the Netflix-Paramount battle over WBD, a reversal from his December statement that he would be involved in the approval process (Allie Canal/NBC News)
https://www.nbcnews.com/business/media/trump-netflix-paramount-fig…
An excellent review of #StarTrek #StarFleetAcademy which fits my viewing experience except I've been watching Star Trek since it first came out on NBC. https://infosec.exchange/@j4yc33/115948886318982523
Towards precise synthetic neural codes: high-dimensional stimulation with flexible electrodes #BCI
NFL-leading eight Cowboys to play in 2026 Pro Bowl Games https://www.dallascowboys.com/news/nfl-leading-eight-cowboys-to-play-in-2026-pro-bowl-games
MS NOW, formerly MSNBC, plans to launch membership subscriptions in the summer of 2026, offering events, curated insights, and moderated community spaces (Sara Fischer/Axios)
https://www.axios.com/2025/12/04/ms-now-msnbc-subscription-plan-members
49ers vs. Colts takeaways: Brock Purdy's 5 TDs put Philip Rivers, Indy on the brink https://www.nytimes.com/athletic/6912776/2025/12/22/49ers-colts-score-result-takeaways-mnf-nfl/