Learned representations are a central component in modern ML systems, serving a multitude of downstream tasks. When training such representations, it is often the case that computational and statistical constraints for each downstream task are unknown. In this context, rigid fixed-capacity representations can be either over or under-accommodating to the task at hand.

Learning binary representations of instances and classes is a classical problem with several high potential applications.

A new kind of large language model, developed by researchers at the Allen Institute for AI (Ai2), makes it possible to control how training data is used even after a model has been built. The new model, called FlexOlmo, could challenge the current industry paradigm of big artificial intelligence companies slurping up data from the web, books, and other sources--often with little regard for ownership--and then owning the resulting models entirely. Once data is baked into an AI model today, extracting it from that model is a bit like trying to recover the eggs from a finished cake. "Conventionally, your data is either in or out," says Ali Farhadi, CEO of Ai2, based in Seattle, Washington. "Once I train on that data, you lose control. And you have no way out, unless you force me to go through another multi-million-dollar round of training."

OpenAI claims that Operator outperforms similar rival tools, including Anthropic's Computer Use (a version of Claude 3.5 Sonnet that can carry out simple tasks on a computer) and Google DeepMind's Mariner (a web-browsing agent built on top of Gemini 2.0). The fact that three of the world's top AI firms have converged on the same vision of what agent-based models could be makes one thing clear. The battle for AI supremacy has a new frontier--and it's our computer screens. "Moving from generating text and images to doing things is the right direction," says Ali Farhadi, CEO of the Allen Institute for AI (AI2). "It unlocks business, solves new problems."

The most capable open source AI model with visual abilities yet could see more developers, researchers, and startups develop AI agents that can carry out useful chores on your computers for you. Released today by the Allen Institute for AI (Ai2), the Multimodal Open Language Model, or Molmo, can interpret images as well as converse through a chat interface. This means it can make sense of a computer screen, potentially helping an AI agent perform tasks such as browsing the web, navigating through file directories, and drafting documents. "With this release, many more people can deploy a multimodal model," says Ali Farhadi, CEO of Ai2, a research organization based in Seattle, Washington, and a computer scientist at the University of Washington. "It should be an enabler for next-generation apps."

This past Monday, about a dozen engineers and executives at data science and AI company Databricks gathered in conference rooms connected via Zoom to learn if they had succeeded in building a top artificial intelligence language model. The team had spent months, and about 10 million, training DBRX, a large language model similar in design to the one behind OpenAI's ChatGPT. But they wouldn't know how powerful their creation was until results came back from the final tests of its abilities. "We've surpassed everything," Jonathan Frankle, chief neural network architect at Databricks and leader of the team that built DBRX, eventually told the team, which responded with whoops, cheers, and applause emojis. Frankle usually steers clear of caffeine but was taking sips of iced latte after pulling an all-nighter to write up the results.

We consider an outsourcing problem where a software agent procures multiple services  from providers with uncertain reliabilities to complete a computational task before a  strict deadline. The service consumer’s goal is to design an outsourcing strategy (defining  which services to procure and when) so as to maximize a specific objective function. This  objective function can be different based on the consumer’s nature; a socially-focused consumer  often aims to maximize social welfare, while a self-interested consumer often aims  to maximize its own utility. However, in both cases, the objective function depends on  the providers’ execution costs, which are privately held by the self-interested providers and  hence may be misreported to influence the consumer’s decisions. For such settings, we  develop a unified approach to design truthful procurement auctions that can be used by  both socially-focused and, separately, self-interested consumers. This approach benefits  from our proposed weighted threshold payment scheme which pays the provably minimum  amount to make an auction with a monotone outsourcing strategy incentive compatible.  This payment scheme can handle contingent outsourcing plans, where additional procurement  happens gradually over time and only if the success probability of the already hired  providers drops below a time-dependent threshold. Using a weighted threshold payment  scheme, we design two procurement auctions that maximize, as well as two low-complexity  heuristic-based auctions that approximately maximize, the consumer’s expected utility and  expected social welfare, respectively. We demonstrate the effectiveness and strength of our  proposed auctions through both game-theoretical and empirical analysis. 

Characterizing the patterns of errors that a system makes helps researchers focus future development on increasing its accuracy and robustness. We propose a novel form of "meta learning" that automatically learns interpretable rules that characterize the types of errors that a system makes, and demonstrate these rules' ability to help understand and improve two NLP systems. Our approach works by collecting error cases on validation data, extracting meta-features describing these samples, and finally learning rules that characterize errors using these features. We apply our approach to VilBERT, for Visual Question Answering, and RoBERTa, for Common Sense Question Answering. Our system learns interpretable rules that provide insights into systemic errors these systems make on the given tasks. Using these insights, we are also able to "close the loop" and modestly improve performance of these systems.

For oil and gas companies looking at drilling wells in a new field, the issue becomes one of return vs. cost. The goal is simple enough: install the fewest number of wells that will draw them the most oil or gas from the underground reservoirs for the longest amount of time. The more wells installed, the higher the cost and the larger the impact on the environment. However, finding the right well placements quickly becomes a highly complex math problem. Too few wells sited in the wrong places leaves a lot of resources in the ground.

What's the best way to arrange wells in an oil or gas field? It's a simple enough question, but the answer can be very complex. Now a Cal Tech/JPL spinoff is developing a new approach that blends traditional HPC simulation with deep reinforcement learning running on GPUs to optimize energy extraction. The well placement game is a familiar one to oil and gas companies. For years, they have been using simulators running atop HPC systems to model underground reservoirs.