The under-representation of cloud formation is a long-standing bias associated with climate simulations. Parameterisation schemes are required to capture cloud processes within current climate models but have known biases. We overcome these biases by embedding a Multi-Output Gaussian Process (MOGP) trained on high resolution Unified Model simulations to represent the variability of temperature and specific humidity within a climate model. A trained MOGP model is coupled in-situ with a simplified Atmospheric General Circulation Model named SPEEDY. The temperature and specific humidity profiles of SPEEDY are perturbed at fixed intervals according to the variability predicted from the MOGP. Ten-year predictions are generated for both control and ML-hybrid models. The hybrid model reduces the global precipitation bias by 18\% and over the tropics by 22\%. To further understand the drivers of these improvements, physical quantities of interest are explored, such as the distribution of lifted index values and the alteration of the Hadley cell. The control and hybrid set-ups are also run in a plus 4K sea-surface temperature experiment to explore the effects of the approach on patterns relating to cloud cover and precipitation in a warmed climate setting.

Seasonal climate forecasts are socioeconomically important for managing the impacts of extreme weather events and for planning in sectors like agriculture and energy. Climate predictability on seasonal timescales is tied to boundary effects of the ocean on the atmosphere and coupled interactions in the ocean-atmosphere system. We present the Ocean-linked-atmosphere (Ola) model, a high-resolution (0.25{\deg}) Artificial Intelligence/ Machine Learning (AI/ML) coupled earth-system model which separately models the ocean and atmosphere dynamics using an autoregressive Spherical Fourier Neural Operator architecture, with a view towards enabling fast, accurate, large ensemble forecasts on the seasonal timescale. We find that Ola exhibits learned characteristics of ocean-atmosphere coupled dynamics including tropical oceanic waves with appropriate phase speeds, and an internally generated El Ni\~no/Southern Oscillation (ENSO) having realistic amplitude, geographic structure, and vertical structure within the ocean mixed layer. We present initial evidence of skill in forecasting the ENSO which compares favorably to the SPEAR model of the Geophysical Fluid Dynamics Laboratory.

Self-supervised speech models (S3Ms) have become an effective backbone for speech applications. Various analyses suggest that S3Ms encode linguistic properties. In this work, we seek a more fine-grained analysis of the word-level linguistic properties encoded in S3Ms. Specifically, we curate a novel dataset of near homophone (phonetically similar) and synonym (semantically similar) word pairs and measure the similarities between S3M word representation pairs. Our study reveals that S3M representations consistently and significantly exhibit more phonetic than semantic similarity. Further, we question whether widely used intent classification datasets such as Fluent Speech Commands and Snips Smartlights are adequate for measuring semantic abilities. Our simple baseline, using only the word identity, surpasses S3M-based models. This corroborates our findings and suggests that high scores on these datasets do not necessarily guarantee the presence of semantic content.

A deep learning (DL) model, based on a transformer architecture, is trained on a climate-model dataset and compared with a standard linear inverse model (LIM) in the tropical Pacific. We show that the DL model produces more accurate forecasts compared to the LIM when tested on a reanalysis dataset. We then assess the ability of an ensemble Kalman filter to reconstruct the monthly-averaged upper ocean from a noisy set of 24 sea-surface temperature observations designed to mimic existing coral proxy measurements, and compare results for the DL model and LIM. Due to signal damping in the DL model, we implement a novel inflation technique by adding noise from hindcast experiments. Results show that assimilating observations with the DL model yields better reconstructions than the LIM for observation averaging times ranging from one month to one year. The improved reconstruction is due to the enhanced predictive capabilities of the DL model, which map the memory of past observations to future assimilation times.

In this paper, we present the LingOly benchmark, a novel benchmark for advanced reasoning abilities in large language models. Using challenging Linguistic Olympiad puzzles, we evaluate (i) capabilities for in-context identification and generalisation of linguistic patterns in very low-resource or extinct languages, and (ii) abilities to follow complex task instructions. The LingOly benchmark covers more than 90 mostly low-resource languages, minimising issues of data contamination, and contains 1,133 problems across 6 formats and 5 levels of human difficulty. We assess performance with both direct accuracy and comparison to a no-context baseline to penalise memorisation. Scores from 11 state-of-the-art LLMs demonstrate the benchmark to be challenging, and models perform poorly on the higher difficulty problems. On harder problems, even the top model only achieved 38.7% accuracy, 24.7% improvement over the no-context baseline. Large closed models typically outperform open models, and in general, the higher resource the language, the better the scores. These results indicate, in absence of memorisation, true multi-step out-of-domain reasoning remains a challenge for current language models.

Humans rarely learn one fact in isolation. Instead, learning a new fact induces knowledge of other facts about the world. For example, in learning a korat is a type of cat, you also infer it is a mammal and has claws, ensuring your model of the world is consistent. Knowledge editing aims to inject new facts into language models to improve their factuality, but current benchmarks fail to evaluate consistency, which is critical to ensure efficient, accurate, and generalizable edits. We manually create TAXI, a new benchmark dataset specifically created to evaluate consistency in categorical knowledge edits. TAXI contains 11,120 multiple-choice queries for 976 edits spanning 41 categories (e.g., Dogs), 164 subjects (e.g., Labrador), and 183 properties (e.g., is a mammal). We then use TAXI to evaluate popular editors' categorical consistency, measuring how often editing a subject's category appropriately edits its properties. We find that 1) the editors achieve marginal, yet non-random consistency, 2) their consistency far underperforms human baselines, and 3) consistency is more achievable when editing atypical subjects Our code and data are available at https://github.com/derekpowell/taxi.

Large language models (LLMs) have excelled in various natural language processing tasks, but challenges in interpretability and trustworthiness persist, limiting their use in high-stakes fields. Causal discovery offers a promising approach to improve transparency and reliability. However, current evaluations are often one-sided and lack assessments focused on interpretability performance. Additionally, these evaluations rely on synthetic data and lack comprehensive assessments of real-world datasets. These lead to promising methods potentially being overlooked. To address these issues, we propose a flexible evaluation framework with metrics for evaluating differences in causal structures and causal effects, which are crucial attributes that help improve the interpretability of LLMs. We introduce the Open Causal Discovery Benchmark (OCDB), based on real data, to promote fair comparisons and drive optimization of algorithms. Additionally, our new metrics account for undirected edges, enabling fair comparisons between Directed Acyclic Graphs (DAGs) and Completed Partially Directed Acyclic Graphs (CPDAGs). Experimental results show significant shortcomings in existing algorithms' generalization capabilities on real data, highlighting the potential for performance improvement and the importance of our framework in advancing causal discovery techniques.

The most popular Cross-Document Event Coreference Resolution (CDEC) datasets fail to convey the true difficulty of the task, due to the lack of lexical diversity between coreferring event triggers (words or phrases that refer to an event). Furthermore, there is a dearth of event datasets for figurative language, limiting a crucial avenue of research in event comprehension. We address these two issues by introducing ECB+META, a lexically rich variant of Event Coref Bank Plus (ECB+) for CDEC on symbolic and metaphoric language. We use ChatGPT as a tool for the metaphoric transformation of sentences in the documents of ECB+, then tag the original event triggers in the transformed sentences in a semi-automated manner. In this way, we avoid the re-annotation of expensive coreference links. We present results that show existing methods that work well on ECB+ struggle with ECB+META, thereby paving the way for CDEC research on a much more challenging dataset. Code/data: https://github.com/ahmeshaf/llms_coref

Retrieval augmented generation has revolutionized large language model (LLM) outputs by providing factual supports. Nevertheless, it struggles to capture all the necessary knowledge for complex reasoning questions. Existing retrieval methods typically divide reference documents into passages, treating them in isolation. These passages, however, are often interrelated, such as passages that are contiguous or share the same keywords. Therefore, recognizing the relatedness is crucial for enhancing the retrieval process. In this paper, we propose a novel retrieval method, called GNN-Ret, which leverages graph neural networks (GNNs) to enhance retrieval by considering the relatedness between passages. Specifically, we first construct a graph of passages by connecting passages that are structure-related and keyword-related. A graph neural network (GNN) is then leveraged to exploit the relationships between passages and improve the retrieval of supporting passages. Furthermore, we extend our method to handle multi-hop reasoning questions using a recurrent graph neural network (RGNN), named RGNN-Ret. At each step, RGNN-Ret integrates the graphs of passages from previous steps, thereby enhancing the retrieval of supporting passages. Extensive experiments on benchmark datasets demonstrate that GNN-Ret achieves higher accuracy for question answering with a single query of LLMs than strong baselines that require multiple queries, and RGNN-Ret further improves accuracy and achieves state-of-the-art performance, with up to 10.4% accuracy improvement on the 2WikiMQA dataset.

Advances towards more faithful and traceable answers of Large Language Models (LLMs) are crucial for various research and practical endeavors. One avenue in reaching this goal is basing the answers on reliable sources. However, this Evidence-Based QA has proven to work insufficiently with LLMs in terms of citing the correct sources (source quality) and truthfully representing the information within sources (answer attributability). In this work, we systematically investigate how to robustly fine-tune LLMs for better source quality and answer attributability. Specifically, we introduce a data generation pipeline with automated data quality filters, which can synthesize diversified high-quality training and testing data at scale. We further introduce four test sets to benchmark the robustness of fine-tuned specialist models. Extensive evaluation shows that fine-tuning on synthetic data improves performance on both in- and out-of-distribution. Furthermore, we show that data quality, which can be drastically improved by proposed quality filters, matters more than quantity in improving Evidence-Based QA.