Israeli forces killed at least nine Palestinians in Gaza in strikes on Saturday, according to the Gaza health ministry, the latest in a string of Israeli attacks on the enclave that have kept up despite a roughly two-month-old truce with Hamas. Since the cease-fire went into effect in mid-January, the military has conducted constant strikes in Gaza. Israel has accused militants of threatening its forces by laying explosive devices, flying drones or by approaching where Israeli troops are deployed. Hamas has claimed those attacks have killed more than 150 people since the truce took effect, at least some of them civilians. And it has accused Israel of repeatedly violating the agreement by continuing military operations.

At least nine people, including three journalists, have been killed and several others wounded in an Israeli drone attack on Beit Lahiya in northern Gaza, according to Palestinian media. The attack on Saturday reportedly targeted a relief team that was accompanied by journalists and photographers. At least three local journalists are among the dead. The Palestinian Journalists' Protection Center said in a statement that "the journalists were documenting humanitarian relief efforts for those affected by Israel's genocidal war" and called on Gaza ceasefire mediators to pressure Israeli Prime Minister Benjamin Netanyahu to move forward with implementing the agreed truce and prisoner exchange. Israel has rejected opening talks on the second phase of the ceasefire between it and Hamas, which would require it to negotiate over a permanent end to the war, a key Hamas demand.

White House press secretary Karoline Leavitt discusses the U.S. inflation rate, peace talks between Russia and Ukraine and more on'America Reports.' "Real Time" host Bill Maher challenged one of President Trump's central arguments as he wages a tariff war among several countries. "I have one basic question: Why do we want to bring back manufacturing?" Maher asked his panel on Friday. "It's so 70s, you know? I mean, that ship has sailed. You know, there are countries that make jeans for 11. We're never going to be that country again."

DOGE leader Elon Musk opens up about his work in space on'Kudlow.' Elon Musk has confirmed that SpaceX's Starship will head to Mars at the end of 2026. The ship will be carrying Optimus, Tesla's humanoid robot. The tech billionaire said that if all goes well, humans could be on the red planet by 2029, although he admitted that 2031 is more likely. NASA's Perseverance Mars rover used its dual-camera Mastcam-Z imager to capture this image of "Santa Cruz," a hill within Jezero Crater, on April 29, 2021, the 68th Martian day, or sol, of the mission. The X account for Optimus replied to Musk's announcement with just two words: "Hold on."

British Prime Minister Keir Starmer stresses the United Kingdom's intention to provide "robust and credible" security guarantees to prevent Russia from attacking Ukraine again as Russian President Vladimir Putin has yet to sign up to a US-brokered truce proposal as fighting rages on the ground. "We will build up Ukraine's own defences and armed forces and be ready to deploy as a'coalition of the willing' in the event of a peace deal to help secure Ukraine on the land, at sea and in the sky," Starmer said on Saturday after a virtual meeting with 25 European Union and other world leaders as well as Ukrainian President Volodymyr Zelenskyy. "We agreed military planners would convene again in the UK this week to progress practical plans for how our militaries can support Ukraine's future security," Starmer added. The meeting was held after Putin on Friday agreed in principle to an immediate 30-day ceasefire put forward by Washington and already accepted by Ukraine. But on Friday, Putin also said there were issues to work out.

A team of charity workers has been killed in Israeli strikes in northern Gaza, the UK-registered Al Khair Foundation has told the BBC. The charity said eight workers - including volunteers and journalists documenting their activities - were killed when their vehicles were targeted on Saturday in what Hamas described as a "blatant violation" of the ceasefire agreement with Israel. The Israeli military has said it had struck "two terrorists who were identified operating a drone that posed a threat to Israeli troops", adding that it then targeted "additional terrorists" who arrived at the scene. The charity rejects the allegation that members of its team were terrorists.

A woman in Washington, D.C., views a manipulated video on January 24, 2019, that changes what is said by President Donald Trump and former President Barack Obama. This illustration photo taken on January 30, 2023 shows a phone screen displaying a statement from the head of security policy at META with a fake video of Ukrainian President Volodymyr Zelensky calling on his soldiers to lay down their weapons shown in the background, in Washington, D.C. (OLIVIER DOULIERY/AFP via Getty Images) NEW REALITY: Artificial intelligence (AI) is producing hyperrealistic "digital twins" of politicians, celebrities, pornographic material, and more – leaving victims of deepfake technology struggling to determine legal recourse. NO BOUNDARY: Scarlett Johansson has taken a vocal stand on artificial intelligence, after having her likeness and voice used without permission. Last year, Johansson said she had been asked to voice OpenAI's Chatbot by CEO Sam Altman, but turned down the job, only for people to notice that the feature, named "Sky," sounded almost exactly like the actress. It was like: If that can happen to me, how are we going to protect ourselves from this? There's no boundary here; we're setting ourselves up to be taken advantage of," the 40-year-old told InStyle Magazine earlier this month.

SpaceX said it would review data "to better understand [the] root cause" of the most recent explosion and noted it happened after the loss of "several" engines. The Federal Aviation Administration (FAA) said the company would be required to conduct an investigation before it could fly again. Nasa hopes to use a modified version of the spaceship as a human lunar lander for its Artemis missions to return to the Moon. The tech billionaire has grand designs that the rocket system will one day take humans to the Moon, and then on to Mars, making humans "multi-planetary". He said that the first Mars mission would carry the Tesla humanoid robot "Optimus", which was shown to the public last year.

We propose the notion of empirical privacy variance and study it in the context of differentially private fine-tuning of language models. Specifically, we show that models calibrated to the same $(\varepsilon, \delta)$-DP guarantee using DP-SGD with different hyperparameter configurations can exhibit significant variations in empirical privacy, which we quantify through the lens of memorization. We investigate the generality of this phenomenon across multiple dimensions and discuss why it is surprising and relevant. Through regression analysis, we examine how individual and composite hyperparameters influence empirical privacy. The results reveal a no-free-lunch trade-off: existing practices of hyperparameter tuning in DP-SGD, which focus on optimizing utility under a fixed privacy budget, often come at the expense of empirical privacy. To address this, we propose refined heuristics for hyperparameter selection that explicitly account for empirical privacy, showing that they are both precise and practically useful. Finally, we take preliminary steps to understand empirical privacy variance. We propose two hypotheses, identify limitations in existing techniques like privacy auditing, and outline open questions for future research.

This paper investigates whether sequence models can learn to perform numerical algorithms, e.g. gradient descent, on the fundamental problem of least squares. Our goal is to inherit two properties of standard algorithms from numerical analysis: (1) machine precision, i.e. we want to obtain solutions that are accurate to near floating point error, and (2) numerical generality, i.e. we want them to apply broadly across problem instances. We find that prior approaches using Transformers fail to meet these criteria, and identify limitations present in existing architectures and training procedures. First, we show that softmax Transformers struggle to perform high-precision multiplications, which prevents them from precisely learning numerical algorithms. Second, we identify an alternate class of architectures, comprised entirely of polynomials, that can efficiently represent high-precision gradient descent iterates. Finally, we investigate precision bottlenecks during training and address them via a high-precision training recipe that reduces stochastic gradient noise. Our recipe enables us to train two polynomial architectures, gated convolutions and linear attention, to perform gradient descent iterates on least squares problems. For the first time, we demonstrate the ability to train to near machine precision. Applied iteratively, our models obtain 100,000x lower MSE than standard Transformers trained end-to-end and they incur a 10,000x smaller generalization gap on out-of-distribution problems. We make progress towards end-to-end learning of numerical algorithms for least squares.