US lawmakers questioned Attorney General Pam Bondi on Wednesday about the Justice Department's handling of documents related to the late sex offender Jeffrey Epstein, while the Environmental Protection Agency is expected to roll back climate change regulations on Thursday, sparking legal battles. Meanwhile, the El Paso International Airport in Texas briefly closed due to a military operation involving a new anti-drone laser, and SpaceX removed a crew access arm at Kennedy Space Center. In other news, researchers at MIT have developed a new method for fine-tuning large language models.
Attorney General Pam Bondi testified before the House Judiciary Committee for five hours on Wednesday, defending the DOJ's handling of the Epstein records, according to Al Jazeera. Bondi stated that there are pending investigations in the case. This hearing followed demands for justice from Epstein survivors.
The EPA's expected rollback of the endangerment finding, which underpins the US's ability to regulate greenhouse gases, is the result of over 15 years of work from right-wing special interest groups, according to Wired. This move represents the most aggressive action against climate regulation to date and is expected to lead to a lengthy legal fight that will likely reach the Supreme Court. The rollback could also create uncertainty for various industries, including oil and car companies.
In Texas, the Federal Aviation Administration closed airspace up to 18,000 feet above the El Paso International Airport on Tuesday night, according to Ars Technica. The closure, initially expected to last 10 days, was lifted less than 10 hours later. US Secretary of Transportation Sean Duffy explained the closure as a response to a "cartel drone incursion."
SpaceX removed the Crew Access Arm from Launch Complex 39A at NASA's Kennedy Space Center in Florida on Wednesday, according to Ars Technica. The site, originally built for Saturn V rockets and later used for the Space Shuttle program, is now leased to SpaceX for its Falcon 9 and Falcon Heavy rockets.
Researchers at MIT, the Improbable AI Lab, and ETH Zurich have developed a new technique called self-distillation fine-tuning (SDFT) that allows large language models to learn new skills without forgetting their past capabilities, according to VentureBeat. This method allows models to learn directly from demonstrations and experiments, potentially eliminating the need for separate models for every skill. Experiments show that SDFT consistently outperforms traditional supervised fine-tuning.
Discussion
AI Experts & Community
Be the first to comment