Large language models (LLMs) have achieved impressive success on many benchmarks for mathematical reasoning. However, there is growing concern that some of this performance actually reflects dataset contamination, where data closely resembling benchmark questions leaks into the training data, instead of true reasoning ability. To investigate this claim rigorously, we commission Grade School Math 1000 (GSM1k). GSM1k is designed to mirror the style and complexity of the established GSM8k benchmark, the gold standard for measuring elementary mathematical reasoning. We ensure that the two benchmarks are comparable across important metrics such as human solve rates, number of steps in solution, answer magnitude, and more. When evaluating leading open- and closed-source LLMs on GSM1k, we observe accuracy drops of up to 13%, with several families of models (e.g., Phi and Mistral) showing evidence of systematic overfitting across almost all model sizes. At the same time, many models, especially those on the frontier, (e.g., Gemini/GPT/Claude) show minimal signs of overfitting. Further analysis suggests a positive relationship (Spearman's r^2=0.32) between a model's probability of generating an example from GSM8k and its performance gap between GSM8k and GSM1k, suggesting that many models may have partially memorized GSM8k.

There is a growing demand to deploy computation-intensive deep learning (DL) models on resource-constrained mobile devices for real-time intelligent applications. Equipped with a variety of processing units such as CPUs, GPUs, and NPUs, the mobile devices hold potential to accelerate DL inference via parallel execution across heterogeneous processors. Various efficient parallel methods have been explored to optimize computation distribution, achieve load balance, and minimize communication cost across processors. Yet their practical effectiveness in the dynamic and diverse real-world mobile environment is less explored. This paper presents a holistic empirical study to assess the capabilities and challenges associated with parallel DL inference on heterogeneous mobile processors. Through carefully designed experiments covering various DL models, mobile software/hardware environments, workload patterns, and resource availability, we identify limitations of existing techniques and highlight opportunities for cross-level optimization.

The diversity of knowledge encoded in large language models (LLMs) and their ability to apply this knowledge zero-shot in a range of settings makes them a promising candidate for use in decision-making. However, they are currently limited by their inability to reliably provide outputs which are explainable and contestable. In this paper, we attempt to reconcile these strengths and weaknesses by introducing a method for supplementing LLMs with argumentative reasoning. Concretely, we introduce argumentative LLMs, a method utilising LLMs to construct argumentation frameworks, which then serve as the basis for formal reasoning in decision-making. The interpretable nature of these argumentation frameworks and formal reasoning means that any decision made by the supplemented LLM may be naturally explained to, and contested by, humans. We demonstrate the effectiveness of argumentative LLMs experimentally in the decision-making task of claim verification. We obtain results that are competitive with, and in some cases surpass, comparable state-of-the-art techniques.

During the three month long eruption of Kilauea volcano, Hawaii in 2018, the pre-existing summit caldera collapsed in over 60 quasi-periodic failure events. The last 40 of these events, which generated Mw >5 very long period (VLP) earthquakes, had inter-event times between 0.8 - 2.2 days. These failure events offer a unique dataset for testing methods for predicting earthquake recurrence based on locally recorded GPS, tilt, and seismicity data. In this work, we train a deep learning graph neural network (GNN) to predict the time-to-failure of the caldera collapse events using only a fraction of the data recorded at the start of each cycle. We find that the GNN generalizes to unseen data and can predict the time-to-failure to within a few hours using only 0.5 days of data, substantially improving upon a null model based only on inter-event statistics. Predictions improve with increasing input data length, and are most accurate when using high-SNR tilt-meter data. Applying the trained GNN to synthetic data with different magma pressure decay times predicts failure at a nearly constant stress threshold, revealing that the GNN is sensing the underling physics of caldera collapse. These findings demonstrate the predictability of caldera collapse sequences under well monitored conditions, and highlight the potential of machine learning methods for forecasting real world catastrophic events with limited training data.

We investigate the potential of large language models (LLMs) to disentangle text variables--to remove the textual traces of an undesired forbidden variable in a task sometimes known as text distillation and closely related to the fairness in AI and causal inference literature. We employ a range of various LLM approaches in an attempt to disentangle text by identifying and removing information about a target variable while preserving other relevant signals. We show that in the strong test of removing sentiment, the statistical association between the processed text and sentiment is still detectable to machine learning classifiers post-LLM-disentanglement. Furthermore, we find that human annotators also struggle to disentangle sentiment while preserving other semantic content. This suggests there may be limited separability between concept variables in some text contexts, highlighting limitations of methods relying on text-level transformations and also raising questions about the robustness of disentanglement methods that achieve statistical independence in representation space.

Process mining offers powerful techniques for discovering, analyzing, and enhancing real-world business processes. In this context, Petri nets provide an expressive means of modeling process behavior. However, directly analyzing and comparing intricate Petri net presents challenges. This study introduces PetriNet2Vec, a novel unsupervised methodology based on Natural Language Processing concepts inspired by Doc2Vec and designed to facilitate the effective comparison, clustering, and classification of process models represented as embedding vectors. These embedding vectors allow us to quantify similarities and relationships between different process models. Our methodology was experimentally validated using the PDC Dataset, featuring 96 diverse Petri net models. We performed cluster analysis, created UMAP visualizations, and trained a decision tree to provide compelling evidence for the capability of PetriNet2Vec to discern meaningful patterns and relationships among process models and their constituent tasks. Through a series of experiments, we demonstrated that PetriNet2Vec was capable of learning the structure of Petri nets, as well as the main properties used to simulate the process models of our dataset. Furthermore, our results showcase the utility of the learned embeddings in two crucial downstream tasks within process mining enhancement: process classification and process retrieval.

The United States Department of Defense has acknowledged that a drone strike in Syria, initially said to have successfully targeted an al-Qaeda leader, actually killed a farmer. The Pentagon stated on Thursday that the drone strike on May 3, 2023, killed a 56-year-old shepherd named Lutfi Hasan Masto, whom they initially misidentified as a senior member of al-Qaeda. US Central Command, which oversees military activities in the Middle East, wrote that it "acknowledges and regrets the civilian harm that resulted from the airstrike". A year ago a US strike in Syria killed a'senior Al-Qaeda leader' The US military has now officially accepted it was a mistake, blaming'confirmation bias'https://t.co/WG9Mf0Rdq2 The killing of Masto is the latest incident to raise questions about the impact of US drone warfare on civilians, who often pay the price for botched strikes.

This paper presents a set of intersectional feminist principles for conducting equitable, ethical, and sustainable AI research. In Data Feminism (2020), we offered seven principles for examining and challenging unequal power in data science. Here, we present a rationale for why feminism remains deeply relevant for AI research, rearticulate the original principles of data feminism with respect to AI, and introduce two potential new principles related to environmental impact and consent. Together, these principles help to 1) account for the unequal, undemocratic, extractive, and exclusionary forces at work in AI research, development, and deployment; 2) identify and mitigate predictable harms in advance of unsafe, discriminatory, or otherwise oppressive systems being released into the world; and 3) inspire creative, joyful, and collective ways to work towards a more equitable, sustainable world in which all of us can thrive.

Traditionally, natural language processing (NLP) models often use a rich set of features created by linguistic expertise, such as semantic representations. However, in the era of large language models (LLMs), more and more tasks are turned into generic, end-to-end sequence generation problems. In this paper, we investigate the question: what is the role of semantic representations in the era of LLMs? Specifically, we investigate the effect of Abstract Meaning Representation (AMR) across five diverse NLP tasks. We propose an AMR-driven chain-of-thought prompting method, which we call AMRCoT, and find that it generally hurts performance more than it helps. To investigate what AMR may have to offer on these tasks, we conduct a series of analysis experiments. We find that it is difficult to predict which input examples AMR may help or hurt on, but errors tend to arise with multi-word expressions, named entities, and in the final inference step where the LLM must connect its reasoning over the AMR to its prediction. We recommend focusing on these areas for future work in semantic representations for LLMs. Our code: https://github.com/causalNLP/amr_llm.

This paper introduces UQA, a novel dataset for question answering and text comprehension in Urdu, a low-resource language with over 70 million native speakers. UQA is generated by translating the Stanford Question Answering Dataset (SQuAD2.0), a large-scale English QA dataset, using a technique called EATS (Enclose to Anchor, Translate, Seek), which preserves the answer spans in the translated context paragraphs. The paper describes the process of selecting and evaluating the best translation model among two candidates: Google Translator and Seamless M4T. The paper also benchmarks several state-of-the-art multilingual QA models on UQA, including mBERT, XLM-RoBERTa, and mT5, and reports promising results. For XLM-RoBERTa-XL, we have an F1 score of 85.99 and 74.56 EM. UQA is a valuable resource for developing and testing multilingual NLP systems for Urdu and for enhancing the cross-lingual transferability of existing models. Further, the paper demonstrates the effectiveness of EATS for creating high-quality datasets for other languages and domains. The UQA dataset and the code are publicly available at www.github.com/sameearif/UQA.