Large Language Models (LLMs) exhibit remarkable capabilities across a spectrum of tasks in financial services, including report generation, chatbots, sentiment analysis, regulatory compliance, investment advisory, financial knowledge retrieval, and summarization. However, their intrinsic complexity and lack of transparency pose significant challenges, especially in the highly regulated financial sector, where interpretability, fairness, and accountability are critical. As far as we are aware, this paper presents the first application in the finance domain of understanding and utilizing the inner workings of LLMs through mechanistic interpretability, addressing the pressing need for transparency and control in AI systems. Mechanistic interpretability is the most intuitive and transparent way to understand LLM behavior by reverse-engineering their internal workings. By dissecting the activations and circuits within these models, it provides insights into how specific features or components influence predictions - making it possible not only to observe but also to modify model behavior. In this paper, we explore the theoretical aspects of mechanistic interpretability and demonstrate its practical relevance through a range of financial use cases and experiments, including applications in trading strategies, sentiment analysis, bias, and hallucination detection. While not yet widely adopted, mechanistic interpretability is expected to become increasingly vital as adoption of LLMs increases. Advanced interpretability tools can ensure AI systems remain ethical, transparent, and aligned with evolving financial regulations. In this paper, we have put special emphasis on how these techniques can help unlock interpretability requirements for regulatory and compliance purposes - addressing both current needs and anticipating future expectations from financial regulators globally.

Large vision-language models have become widely adopted to advance in various domains. However, developing a trustworthy system with minimal interpretable characteristics of large-scale models presents a significant challenge. One of the most prevalent terms associated with the fallacy functions caused by these systems is hallucination, where the language model generates a response that does not correspond to the visual content. To mitigate this problem, several approaches have been developed, and one prominent direction is to ameliorate the decoding process. In this paper, we propose a new Bijective Maximum Likelihood Learning (BIMA) approach to hallucination mitigation using normalizing flow theories. The proposed BIMA method can efficiently mitigate the hallucination problem in prevailing vision-language models, resulting in significant improvements. Notably, BIMA achieves the average F1 score of 85.06% on POPE benchmark and remarkably reduce CHAIRS and CHAIRI by 7.6% and 2.6%, respectively. To the best of our knowledge, this is one of the first studies that contemplates the bijection means to reduce hallucination induced by large vision-language models.

Semantic Text Embedding is a fundamental NLP task that encodes textual content into vector representations, where proximity in the embedding space reflects semantic similarity. While existing embedding models excel at capturing general meaning, they often overlook ideological nuances, limiting their effectiveness in tasks that require an understanding of political bias. To address this gap, we introduce PRISM, the first framework designed to Produce inteRpretable polItical biaS eMbeddings. PRISM operates in two key stages: (1) Controversial Topic Bias Indicator Mining, which systematically extracts fine-grained political topics and their corresponding bias indicators from weakly labeled news data, and (2) Cross-Encoder Political Bias Embedding, which assigns structured bias scores to news articles based on their alignment with these indicators. This approach ensures that embeddings are explicitly tied to bias-revealing dimensions, enhancing both interpretability and predictive power. Through extensive experiments on two large-scale datasets, we demonstrate that PRISM outperforms state-of-the-art text embedding models in political bias classification while offering highly interpretable representations that facilitate diversified retrieval and ideological analysis. The source code is available at https://github.com/dukesun99/ACL-PRISM.

We introduce a novel framework for analyzing sorting algorithms in pairwise ranking prompting (PRP), re-centering the cost model around LLM inferences rather than traditional pairwise comparisons. While classical metrics based on comparison counts have traditionally been used to gauge efficiency, our analysis reveals that expensive LLM inferences overturn these predictions; accordingly, our framework encourages strategies such as batching and caching to mitigate inference costs. We show that algorithms optimal in the classical setting can lose efficiency when LLM inferences dominate the cost under certain optimizations.

This paper introduces a Dual Evaluation Framework to comprehensively assess the multilingual capabilities of LLMs. By decomposing the evaluation along the dimensions of linguistic medium and cultural context, this framework enables a nuanced analysis of LLMs' ability to process questions within both native and cross-cultural contexts cross-lingually. Extensive evaluations are conducted on a wide range of models, revealing a notable "CulturalLinguistic Synergy" phenomenon, where models exhibit better performance when questions are culturally aligned with the language. This phenomenon is further explored through interpretability probing, which shows that a higher proportion of specific neurons are activated in a language's cultural context. This activation proportion could serve as a potential indicator for evaluating multilingual performance during model training. Our findings challenge the prevailing notion that LLMs, primarily trained on English data, perform uniformly across languages and highlight the necessity of culturally and linguistically model evaluations. Our code can be found at https://yingjiahao14. github.io/Dual-Evaluation/.

Objective: Heart failure (HF) patients present with diverse phenotypes affecting treatment and prognosis. This study evaluates models for phenotyping HF patients based on left ventricular ejection fraction (LVEF) classes, using structured and unstructured data, assessing performance and interpretability. Materials and Methods: The study analyzes all HF hospitalizations at both Amsterdam UMC hospitals (AMC and VUmc) from 2015 to 2023 (33,105 hospitalizations, 16,334 patients). Data from AMC were used for model training, and from VUmc for external validation. The dataset was unlabelled and included tabular clinical measurements and discharge letters. Silver labels for LVEF classes were generated by combining diagnosis codes, echocardiography results, and textual mentions. Gold labels were manually annotated for 300 patients for testing. Multiple Transformer-based (black-box) and Aug-Linear (white-box) models were trained and compared with baselines on structured and unstructured data. To evaluate interpretability, two clinicians annotated 20 discharge letters by highlighting information they considered relevant for LVEF classification. These were compared to SHAP and LIME explanations from black-box models and the inherent explanations of Aug-Linear models. Results: BERT-based and Aug-Linear models, using discharge letters alone, achieved the highest classification results (AUC=0.84 for BERT, 0.81 for Aug-Linear on external validation), outperforming baselines. Aug-Linear explanations aligned more closely with clinicians' explanations than post-hoc explanations on black-box models. Conclusions: Discharge letters emerged as the most informative source for phenotyping HF patients. Aug-Linear models matched black-box performance while providing clinician-aligned interpretability, supporting their use in transparent clinical decision-making.

Endowing dialogue agents with persona information has proven to significantly improve the consistency and diversity of their generations. While much focus has been placed on aligning dialogues with provided personas, the adaptation to the interlocutor's profile remains largely underexplored. In this work, we investigate three key aspects: (1) a model's ability to align responses with both the provided persona and the interlocutor's; (2) its robustness when dealing with familiar versus unfamiliar interlocutors and topics, and (3) the impact of additional fine-tuning on specific persona-based dialogues. We evaluate dialogues generated with diverse speaker pairings and topics, framing the evaluation as an author identification task and employing both LLM-as-a-judge and human evaluations. By systematically masking or disclosing information about the interlocutor, we assess its impact on dialogue generation. Results show that access to the interlocutor's persona improves the recognition of the target speaker, while masking it does the opposite. Although models generalise well across topics, they struggle with unfamiliar interlocutors. Finally, we found that in zero-shot settings, LLMs often copy biographical details, facilitating identification but trivialising the task.

Next location prediction plays a critical role in understanding human mobility patterns. However, existing approaches face two core limitations: (1) they fall short in capturing the complex, multi-functional semantics of real-world locations; and (2) they lack the capacity to model heterogeneous behavioral dynamics across diverse user groups. To tackle these challenges, we introduce NextLocMoE, a novel framework built upon large language models (LLMs) and structured around a dual-level Mixture-of-Experts (MoE) design. Our architecture comprises two specialized modules: a Location Semantics MoE that operates at the embedding level to encode rich functional semantics of locations, and a Personalized MoE embedded within the Transformer backbone to dynamically adapt to individual user mobility patterns. In addition, we incorporate a history-aware routing mechanism that leverages long-term trajectory data to enhance expert selection and ensure prediction stability. Empirical evaluations across several real-world urban datasets show that NextLocMoE achieves superior performance in terms of predictive accuracy, cross-domain generalization, and interpretability

Semantic textual similarity (STS) is a critical task in natural language processing (NLP), enabling applications in retrieval, clustering, and understanding semantic relationships between texts. However, research in this area for the Arabic language remains limited due to the lack of high-quality datasets and pre-trained models. This scarcity of resources has restricted the accurate evaluation and advance of semantic similarity in Arabic text. This paper introduces General Arabic Text Embedding (GATE) models that achieve state-of-the-art performance on the Semantic Textual Similarity task within the MTEB benchmark. GATE leverages Matryoshka Representation Learning and a hybrid loss training approach with Arabic triplet datasets for Natural Language Inference, which are essential for enhancing model performance in tasks that demand fine-grained semantic understanding. GATE outperforms larger models, including OpenAI, with a 20-25% performance improvement on STS benchmarks, effectively capturing the unique semantic nuances of Arabic.

Summarizing long-form narratives--such as books, movies, and TV scripts--requires capturing intricate plotlines, character interactions, and thematic coherence, a task that remains challenging for existing LLMs. We introduce NexusSum, a multi-agent LLM framework for narrative summarization that processes long-form text through a structured, sequential pipeline--without requiring fine-tuning. Our approach introduces two key innovations: (1) Dialogue-to-Description Transformation: A narrative-specific preprocessing method that standardizes character dialogue and descriptive text into a unified format, improving coherence. (2) Hierarchical Multi-LLM Summarization: A structured summarization pipeline that optimizes chunk processing and controls output length for accurate, high-quality summaries. Our method establishes a new state-of-the-art in narrative summarization, achieving up to a 30.0% improvement in BERTScore (F1) across books, movies, and TV scripts. These results demonstrate the effectiveness of multi-agent LLMs in handling long-form content, offering a scalable approach for structured summarization in diverse storytelling domains.