The performance of large language models (LLMs) on verifiable tasks is usually measured by pass@k, the probability of answering a question correctly at least once in k trials. At a fixed budget, a more suitable metric is coverage@cost, the average number of unique questions answered as a function of the total number of attempts. We connect the two metrics and show that the empirically-observed power-law behavior in pass@k leads to a sublinear growth of the coverage@cost (diminishing returns). To solve this problem, we propose Reset-and-Discard (ReD), a query method of LLMs that increases coverage@cost for any given budget, regardless of the pass@k form. Moreover, given a pass@k, we can quantitatively predict the savings in the total number of attempts using ReD. If pass@k is not available for the model, ReD can infer its power-law exponent. Experiments on three LLMs using HumanEval demonstrate that ReD substantially reduces the required attempts, tokens, and USD cost to reach a desired coverage, while also offering an efficient way to measure inference power-laws.

Target search problems are central to a wide range of fields, from biological foraging to the optimization algorithms. Recently, the ability to reset the search has been shown to significantly improve the searcher's efficiency. However, the optimal resetting strategy depends on the specific properties of the search problem and can often be challenging to determine. In this work, we propose a reinforcement learning (RL)-based framework to train agents capable of optimizing their search efficiency in environments by learning how to reset. First, we validate the approach in a well-established benchmark: the Brownian search with resetting. There, RL agents consistently recover strategies closely resembling the sharp resetting distribution, known to be optimal in this scenario. We then extend the framework by allowing agents to control not only when to reset, but also their spatial dynamics through turning actions. In this more complex setting, the agents discover strategies that adapt both resetting and turning to the properties of the environment, outperforming the proposed benchmarks. These results demonstrate how reinforcement learning can serve both as an optimization tool and a mechanism for uncovering new, interpretable strategies in stochastic search processes with resetting.

Machine learning models have become indispensable tools in applications across the physical sciences. Their training is often time-consuming, vastly exceeding the inference timescales. Several protocols have been developed to perturb the learning process and improve the training, such as shrink and perturb, warm restarts, and stochastic resetting. For classifiers, these perturbations have been shown to result in enhanced speedups or improved generalization. However, the design of such perturbations is usually done \textit{ad hoc} by intuition and trial and error. To rationally optimize training protocols, we frame them as first-passage processes and consider their response to perturbations. We show that if the unperturbed learning process reaches a quasi-steady state, the response at a single perturbation frequency can predict the behavior at a wide range of frequencies. We demonstrate that this is the case when training a CIFAR-10 classifier using the ResNet-18 model and use this approach to identify an optimal perturbation and frequency. Our work allows optimization of training protocols of machine learning models using a statistical mechanical approach.

Gloat, an internal marketplace for corporate talent, today announced that it raised $90 million in a Series D round led by Generation Investment Management, bringing the startup's total raised to $192 million. Generation, notably, is chaired by former U.S. Vice President Al Gore. In an email Q&A with TechCrunch, CEO Ben Reuveni said that the proceeds will be put toward expanding Gloat's presence, growing its team of over 250 employees and "strengthening" its R&D initiatives. The list of employee recruitment, acquisition and jobs boards products is practically endless -- see Workday, LinkedIn and SAP SuccessFactors to start. Reuveni doesn't deny that Gloat faces stiff competition, but he sums up what he believes to be the company's differentiators thusly: "Gloat is unique in that we started by powering a solution for internal mobility. Across the market, there are a number of recruiting … tools to source external candidates, but internal mobility poses unique and nuanced challenges. It requires a real understanding of transferable skills and titles that may not be obvious without the deep, organization- and industry-specific insight Gloat's technology was built to offer."

Tel Aviv is the city with the highest number of startups per capita in the world, according to the 2018 Global Startup Ecosystem report -- more than 6,000, of which 18 are unicorns. The city's tech cluster, dubbed Silicon Wadi, is home to more than 100 venture capital funds, plus hundreds of accelerators and co-working places. "Tel Aviv is transitioning from startup nation to scale-up nation," says Eyal Gura, co-founder of Zebra Medical Vision. Amit Gilon, an investor at Kaedan Capital VC fund, agrees – adding that Israel is not just about successful B2B companies anymore, such as Checkpoint, Nice and Amdocs, but also about "big B2C success stories like Playtika, Wix, Fiverr and others". Founded in 2015, Arbe has built a 4D ultra-high-resolution imaging radar for cars.

Artificial intelligence is changing the way we recruit across almost all categories. We know that the technology can improve success in acquiring candidates, but what we need to discover whether it can go further when predicting candidate potential. AI is fueled by data: The more information you give it, the more it can tell you. But, as with all technology, it's only as smart as the data it has. So the better the information you give it, the better the results.

If you're accustomed to hearing the terms artificial intelligence (AI) and machine learning mentioned in the same breath as flying cars or living on other planets, you're due for a wake-up call. Machine learning is here -- and it's expected to grow even more in the next few years. The industry, valued at $8 billion in 2016, is anticipated to reach $47 billion by 2020. While a lion's share of that spend will go to banking and retailers, with data analytics poised to do everything from manage supply chains to predict what we'll be shopping for next, a large portion will be spent by business looking for ways to enhance the HR function. Startups are engineering AI for a piece of the huge HR assessment market and they're already seeing results.