It is an open question what semantic representations transformer-based language models can encode and whether they have access to more abstract aspects of semantic meaning. Here, we propose a diagnostic dataset to investigate how well language models understand the degree semantics of adjectives. In the dataset, referred as the Adjective Scale Probe (ASP), we semi-automatically generate 8 tests of Natural Language Inference (NLI) questions to test 8 key capabilities of adjective interpretation. We apply the ASP dataset to evaluate the performance of 3 language models, i.e., BERT, DeBERTa, and T0. It is found that language models perform below the majority baseline for most tests of the ASP, even when the models have been fine-tuned to achieve high performance on the large-scale MNLI dataset. But after we fine-tune the pre-trained models on a subset of the ASP, DeBERTa can achieve high performance on the untrained adjectives and untrained tests, suggesting that DeBERTa may have captured degree semantic information of adjectives through pre-training but it needs specific training data to learn how to apply such information to the current tasks. In sum, the ASP provides an easy-to-use method to test fine-grained formal semantic properties of adjectives, and reveals language models' abilities to access formal semantic information.

Adjective Scale Probe: Can Language Models Encode Formal Semantics Information?

This presentation examines Temple-Stay, a short-term retreat program held for lay persons at Buddhist monasteries, in the context of the prevailing social malaise and happiness discourse in South Korea. Temple-Stay was initially designed by the Chogye Order, the predominant sect of Korean Buddhism, and the Ministry of Culture and Tourism, to engage foreign visitors during the 2002 World Cup. However, it proved to be unexpectedly popular with local residents as well. In a society, where the percentage of the population seeking professional attention for mental health issues remains relatively low, many Koreans have been drawn to this government-subsidized retreat as a safe, affordable, and low-profile getaway. In fact, Temple-Stay has recently been advertised as "a journey for my happiness" and as a scientifically proven stress reducer. The instrumental use of Buddhist teachings and practices has been criticized by some scholars for misrepresenting the true nature of Buddhism or for appropriating it for other ulterior motivations. How should we make sense of the Buddhist establishment's own parceling of the monastic experience as a healing commodity? Based on ethnographic research and an examination of the history, marketing, and program content of Temple-Stay, this presentation shows that its popularity in the twenty-first century has been due to effective marketing of the program as a healing experience and to the perceived need for such a commodity. It then discusses both the risks and opportunities that the marketization of the program presents to the Buddhist establishment and lay participants.

Temple-Stay in South Korea: A Wellness Journey amid a Happiness Crisis

The importance of diversity for excellence in science

Engaging the Guatemala Scientific Diaspora for Development: Network Building and Shared Learning

DNA extraction from Common Bean (Phaseolus vulgaris) leaves Zach Alvarez1, Rasheed Habib2, and Venu (Kal) Kalavacharla2 1Long Reach High School 2Delaware State University, Department of Agriculture and Natural Resources, Dover, DE The extraction of DNA from different plants can be a difficult process. This is due to the existence of different plant secondary metabolites. There are many methods of extracting DNA from plants. However, in this study one specific method was done. The CTAB method has various steps that are done to extract pure DNA. The plant that DNA was extracted from is common bean (Phaseolus Vulgaris). This plant was selected due to its importance as a source of protein and carbohydrate for millions of people around the world. Also, it normally yields high quality DNA due to how easy the tissues are to grind. The plant tissue used in the experiment was obtained from the school farm. It was collected and flash frozen to reduce DNA degradation. The sample was then stored in the -80° until i

DNA extraction from Common Bean (Phaseolus vulgaris) leaves

Node representation learning for directed graphs is critically important to facilitate many graph mining tasks. To capture the directed edges between nodes, existing methods mostly learn two embedding vectors for each node, source vector and target vector. However, these methods learn the source and target vectors separately. For the node with very low indegree or outdegree, the corresponding target vector or source vector cannot be effectively learned. In this paper, we propose a novel Directed Graph embedding framework based on Generative Adversarial Network, called DGGAN. The main idea is to use adversarial mechanisms to deploy a discriminator and two generators that jointly learn each nodeâ€™s source and target vectors. For a given node, the two generators are trained to generate its fake target and source neighbor nodes from the same underlying distribution, and the discriminator aims to distinguish whether a neighbor node is real or fake. The two generators are formulated into a unified framework and could mutually reinforce each other to learn more robust source and target vectors. Extensive experiments show that DGGAN consistently and significantly outperforms existing state-of-the-art methods across multiple graph mining tasks on directed graphs.

Underline digital video library

Adjective Scale Probe: Can Language Models Encode Formal Semantics Information?

Temple-Stay in South Korea: A Wellness Journey amid a Happiness Crisis

The importance of diversity for excellence in science

Engaging the Guatemala Scientific Diaspora for Development: Network Building and Shared Learning

DNA extraction from Common Bean (Phaseolus vulgaris) leaves

Adversarial Directed Graph Embedding

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES