Autonomy via agents using large language models (LLMs) for personalized, standardized tasks boosts human efficiency. Automating web tasks (like booking hotels within a budget) is increasingly sought after. Fulfilling practical needs, the web agent also serves as an important proof-of-concept example for various agent grounding scenarios, with its success promising advancements in many future applications. Prior research often handcrafts web agent strategies (e.g., prompting templates, multi-agent systems, search methods, etc.) and the corresponding in-context examples, which may not generalize well across all real-world scenarios. On the other hand, there has been limited study on the misalignment between a web agent's observation/action representation and the pre-training data of the LLM it's based on. This discrepancy is especially notable when LLMs are primarily trained for language completion rather than tasks involving embodied navigation actions and symbolic web elements. Our study enhances an LLM-based web agent by simply refining its observation and action space to better align with the LLM's capabilities. This approach enables our base agent to significantly outperform previous methods on a wide variety of web tasks. Specifically, on WebArena, a benchmark featuring general-purpose web interaction tasks, our agent AgentOccam surpasses the previous state-of-the-art and concurrent work by 9.8 (+29.4%) and 5.9 (+15.8%) absolute points respectively, and boosts the success rate by 26.6 points (+161%) over similar plain web agents with its observation and action space alignment. We achieve this without using in-context examples, new agent roles, online feedback or search strategies. AgentOccam's simple design highlights LLMs' impressive zero-shot performance on web tasks, and underlines the critical role of carefully tuning observation and action spaces for LLM-based agents.

JAX is a system for high-performance machine-learning research. It offers the familiarity of Python NumPy together with hardware acceleration. JAX enables the definition and composition of user-wielded function transformations useful for machine-learning programs. These transformations include automatic differentiation, automatic batching, end-to-end compilation (via XLA), parallelizing over multiple accelerators, and more. Composing these transformations is the key to JAX's power and simplicity.

Last week Voxel51 hosted the December 2022 Computer Vision Meetup. Our amazing speakers shared insightful presentations, the virtual room was packed, and the Q&A was vibrant! In this blog post we provide the recordings, recap presentation highlights and Q&A, as well as share the upcoming Meetup schedule so that you can join us at a future event. Hope to see you soon! In lieu of swag, we gave Meetup attendees the opportunity to help guide our monthly donation to charitable causes. The charity that received the highest number of votes was Children International.

But in the literature there lacks discussion on how to build the quantitative relationship between them. In this paper, we propose an agent-based SIR model: ASIR. ASIR can robustly reproduce the infection curve predicted by a given SIR model (the simplest CM.) Notably, one can deduce any parameter of ASIR from parameters of SIR without manual tuning. ASIR offers epidemiologists a method to transform a calibrated SIR model into an agent-based model that inherit SIR's performance without another round of calibration. The design ASIR is inspirational for building a general quantitative relationship between CM and AM.

Success of groups in Meetup is of utmost importance for members who organize them. However, measures of group success in Meetup is quite vague till now. In this paper, we take a step to quantify the success of Meetup groups. Driven by a comprehensive study of our Meetup dataset, we handpick a set of key properties which can potentially regulate a group's success. Finally, we develop a machine learning model leveraging on these features which can predict success of Meetup groups early with high accuracy.

We are looking to organize online x in-person meetups on HK island going forward. Thanks to our sponsor Darwinex to help us supporting the various costs. Abstract: We introduce a class of interpretable tree-based models (P-Trees) for analyzing panel data, with iterative and global (instead of recursive and local) splitting criteria to avoid overfitting and improve model performance. We apply P-Tree to generate a stochastic discount factor model and test assets for cross-sectional asset pricing. Unlike other tree algorithms, P-Trees accommodate imbalanced panels of asset returns and grow under the no-arbitrage condition.

I will show demo using Loan Approval using Azure Machine Learning Studio. This demo will also use the following technologies: * Azure Machine Learning Studio * Data Science The main goal of this session is to provide basic knowledge of Azure Machine Learning Studio with demo so attendee will leave with Azure Machine Learning Studio knowledge and how to apply this in different scenarios.

The Getting started with AI & Computer Vision - A guide of obstacles & insights from the field presentation by Kristina Mishra at the Microsoft Data and AI South Florida covers the challenges and conquests of a database professional taking a deep dive into the data world of AI and Computer Vision - Topics Covered: * Artificial Intelligence * Computer Vision - Abstract: . Are you interested in the world of AI and Computer Vision, but have no idea where to start? . What even is computer vision? Find out what happens when a database professional goes off-brand and raises a hand to take a deep dive into the data world of AI and computer vision. In this session, I share the common obstacles encountered from my first projects and some insights on how to make your first project successful (and less painful).

Ethics: A branch of philosophy that involves systematizing, defending, and recommending concepts of right and wrong behavior. It seeks to resolve questions of human morality by defining concepts such as good and evil, right and wrong, virtue and vice, and justice and crime. It also has 3 major areas of study which are meta-ethics, normative ethics, and applied ethics. Research suggests the problem-solving skills of a diverse group outperform those of groups comprised of the most talented individuals. It adds different perspectives from people with different identities and experiences that may have privileged access to relevant insights and understandings. It also helps form policies and develops research that better caters to the people's needs.

I've just come out of a long-term lockdown. Instead, they crave the friendships and social groups they have been starved of over the past year. That's the verdict of dating apps such as Tinder and Bumble, which are launching or acquiring new services focused entirely on making and maintaining friends. "There's a really interesting trend that has been taking place in the connection space, which is this desire to have platonic relationships," said Bumble founder and CEO Whitney Wolfe Herd. "People are seeking friendship in ways they would have only done offline before the pandemic."