We study the problem of modeling spatiotemporal trajectories over long time horizons using expert demonstrations. For instance, in sports, agents often choose action sequences with long-term goals in mind, such as achieving a certain strategic position. Conventional policy learning approaches, such as those based on Markov decision processes, generally fail at learning cohesive long-term behavior in such high-dimensional state spaces, and are only effective when fairly myopic decisionmaking yields the desired behavior. The key difficulty is that conventional models are "single-scale" and only learn a single state-action policy. We instead propose a hierarchical policy class that automatically reasons about both long-term and shortterm goals, which we instantiate as a hierarchical neural network. We showcase our approach in a case study on learning to imitate demonstrated basketball trajectories, and show that it generates significantly more realistic trajectories compared to non-hierarchical baselines as judged by professional sports analysts.

Existing benchmarks for multi-hop question answering (QA) primarily evaluate models based on their ability to reason about entities and the relationships between them. However, there's a lack of insight into how these models perform in terms of both events and entities.

In recent times, Vision-Language Models (VLMs) have been trained under two predominant paradigms. Generative training has enabled Multimodal Large Language Models (MLLMs) to tackle various complex tasks, yet issues such as hallucinations and weak object discrimination persist. Discriminative training, exemplified by models like CLIP, excels in zero-shot image-text classification and retrieval, yet struggles with complex scenarios requiring fine-grained semantic differentiation.

Ive led the designs of the original iMac, the iPad, and the Apple Watch, among other era-defining products. Then, in 2019, he left Apple to start his own design firm called LoveFrom. The news of his move to OpenAI felt something like learning that LeBron James was joining the Miami Heat: Ive had become synonymous with Apple's success, perhaps second only to Jobs. Now, after a period of independence, he was choosing a new team. The announcement of the deal with OpenAI--for a reported 6.5 billion in OpenAI equity--came via a press release, featuring a rather cuddly portrait of Ive with OpenAI's C.E.O. and co-founder, Sam Altman (shot by the British fashion photographer Craig McDean) and a faux-casual videotaped interview session between the two at San Francisco's Cafe Zoetrope. In it, Altman describes "a family of devices that would let people use A.I. to create all sorts of wonderful things," enabled by "magic intelligence in the cloud."

Large language models (LLMs) have achieved impressive performance across various natural language benchmarks, prompting a continual need to curate more difficult datasets for larger LLMs, which is costly and time-consuming. In this paper, we propose to automate dataset updating and provide systematical analysis regarding its effectiveness in dealing with benchmark leakage issue, difficulty control, and stability. Thus, once current benchmark has been mastered or leaked, we can update it for timely and reliable evaluation. There are two updating strategies: 1) mimicking strategy to generate similar samples based on original data, preserving stylistic and contextual essence, and 2) extending strategy that further expands existing samples at varying cognitive levels by adapting Bloom's taxonomy of educational objectives. Extensive experiments on updated MMLU and BIG-Bench demonstrate the stability of the proposed strategies and find that the mimicking strategy can effectively alleviate issues of overestimation from benchmark leakage. In cases where the efficient mimicking strategy fails, our extending strategy still shows promising results. Additionally, by controlling the difficulty, we can better discern the models' performance and enable fine-grained analysis -- neither too difficult nor too easy an exam can fairly judge students' learning status. To the best of our knowledge, we are the first to automate updating benchmarks for reliable and timely evaluation.

How accurately can one recover the "histogram" (i.e.

Fox News Flash top sports headlines are here. Check out what's clicking on Foxnews.com. The WNBA and the Indiana Fever announced that the allegations of "hateful comments" directed toward Angel Reese on May 17 were "not substantiated." Reese and her Chicago Sky faced the Fever and Caitlin Clark, and at one point, the two had to be separated after a flagrant foul by Clark against Reese. The association announced the next day that it would launch an investigation into the alleged comments.