Machine learning on graphs has recently achieved impressive progress in various domains, including molecular property prediction and chip design. However, benchmarking practices remain fragmented, often relying on narrow, task-specific datasets and inconsistent evaluation protocols, which hampers reproducibility and broader progress. To address this, we introduce GraphBench, a comprehensive benchmarking suite that spans diverse domains and prediction tasks, including node-level, edge-level, graph-level, and generative settings. GraphBench provides standardized evaluation protocols -- with consistent dataset splits and performance metrics that account for out-of-distribution generalization -- as well as a unified hyperparameter tuning framework. Additionally, we benchmark GraphBench using message-passing neural networks and graph transformer models, providing principled baselines and establishing a reference performance. See www.graphbench.io for further details.

We study the problem of excess risk evaluation for empirical risk minimization (ERM) under general convex loss functions. Our contribution is an efficient refitting procedure that computes the excess risk and provides high-probability upper bounds under the fixed-design setting. Assuming only black-box access to the training algorithm and a single dataset, we begin by generating two sets of artificially modified pseudo-outcomes--termed wild responses--created by stochastically perturbing the gradient vectors with carefully chosen scaling. Using these two pseudo-labeled datasets, we then refit the black-box procedure twice to obtain two corresponding wild predictors. Finally, leveraging the original predictor, the two wild predictors, and the constructed wild responses, we derive an efficient excess-risk upper bound. A key feature of our analysis is that it requires no prior knowledge of the complexity of the underlying function class. As a result, the method is essentially model-free and holds significant promise for theoretically evaluating modern opaque machine learning systems--such as deep neural networks and generative models--where traditional capacity-based learning theory becomes infeasible due to the extreme complexity of the hypothesis class.

This material may not be published, broadcast, rewritten, or redistributed. Quotes displayed in real-time or delayed by at least 15 minutes. Market data provided by Factset . Powered and implemented by FactSet Digital Solutions . Mutual Fund and ETF data provided by Refinitiv Lipper .

You can get a year of the ad-free Fan subscription for $67. If you're looking for something to watch during a holiday break from work or school, you can't go wrong with anime, and right now you can sign-up for for as low as $65. Holiday discounts running through December 29 bring the anime streaming service's year-long Fan subscription from $80 down to $67 and its Mega Fan subscription from $120 down to $100. While both the Fan and Mega Fan plans are ad-free, you do get some different benefits depending on which you decide to pay for. Fan subscribers get full access to Crunchyroll's library, new episodes shortly after they air in Japan and five percent off select purchases in the Crunchyroll Store.

Former chancellor George Osborne is joining artificial intelligence (AI) giant OpenAI. He will lead its OpenAI for Countries programme, which is aimed at helping governments increase their AI capacity. Announcing his new London-based role, Osborne said it was a privilege to be joining the company. I recently asked myself the question: what's the most exciting and promising company in the world right now? The answer I believe is OpenAI, he said on X.

Grindr Goes'AI-First' as It Strives to Be an'Everything App for the Gay Guy' After controlling shareholders failed to take Grindr private and controversies over data and the banning of the phrase "No Zionists," Grindr's CEO opens up about AI, privacy, and big expansion plans. Every Grindr user is unique. South Koreans prefer open relationships. The highest percentage of self-proclaimed "daddies" call the US home, and Switzerland is overrun with twinks. Delivered by annual trend report Grindr Unwrapped, those critical insights offer the type of information that will help usher the company into its "AI-first" era where it's "the everything app for the gay guy," CEO George Arison tells WIRED. Grindr was the first to leverage geo-location tech when it burst onto the scene in 2009. Arison arrived at the company in 2022 from the world of automotive ecommerce.

Science Space Solar System Mars NASA's Mars Reconnaissance Orbiter snaps 100,000th image A high school student suggested the steep sand dunes of Syrtis Major for the milestone image. Breakthroughs, discoveries, and DIY tips sent every weekday. NASA's Mars Reconnaissance Orbiter (MRO) officially went into service above the Red Planet in November 2006. The spacecraft has since spent nearly 20 years circling Earth's closest neighbor, studying its geology and identifying icy evidence of a once watery world . After already sending back more than 450 terabits of data over the course of its ongoing mission, the orbiter recently passed a major milestone: its 100,000th image of the Martian surface.

Expect big announcements from the likes of Samsung, Sony, NVIDIA and many more. CES doesn't start until January, but whispers of the products and announcements that could be in store for tech's biggest annual conference have already started to take shape. The CES 2026 show floor is officially open from January 6 through 9, although the show kicks off with events on Sunday January 4 and a host of press conferences on Monday. As always, product demos, announcements and networking will be happening at the Las Vegas Convention Center and surrounding hotels all over the city. As usual, Engadget will be covering the event in-person and remotely, bringing you news and hands-ons straight from the show floor.

Chrome for Android now uses Google Gemini to convert webpages into podcast-style summaries with virtual hosts for hands-free browsing during commutes.

As OpenAI scrambles to improve ChatGPT, it's ditching a feature in its free tier that contributed to last summer's user revolt. OpenAI has quietly reversed a major change to how hundreds of millions of people use ChatGPT . On a low-profile blog that tracks product changes, the company said that it rolled back ChatGPT's model router--an automated system that sends complicated user questions to more advanced "reasoning" models--for users on its Free and $5-a-month Go tiers. Instead, those users will now default to GPT-5.2 Instant, the fastest and cheapest-to-serve version of OpenAI's new model series. Free and Go users will still be able to access reasoning models, but they will have to select them manually.