This study investigates the internal representations of verb-particle combinations, called multi-word verbs, within transformer-based large language models (LLMs), specifically examining how these models capture lexical and syntactic properties at different neural network layers. Using the BERT architecture, we analyze the representations of its layers for two different verb-particle constructions: phrasal verbs like 'give up' and prepositional verbs like 'look at'. Our methodology includes training probing classifiers on the internal representations to classify these categories at both word and sentence levels. The results indicate that the model's middle layers achieve the highest classification accuracies. To further analyze the nature of these distinctions, we conduct a data separability test using the Generalized Discrimination Value (GDV). While GDV results show weak linear separability between the two verb types, probing classifiers still achieve high accuracy, suggesting that representations of these linguistic categories may be non-linearly separable. This aligns with previous research indicating that linguistic distinctions in neural networks are not always encoded in a linearly separable manner. These findings computationally support usage-based claims on the representation of verb-particle constructions and highlight the complex interaction between neural network architectures and linguistic structures.

Recent large language models (LLMs) support long contexts ranging from 128K to 1M tokens. A popular method for evaluating these capabilities is the needle-in-a-haystack (NIAH) test, which involves retrieving a "needle" (relevant information) from a "haystack" (long irrelevant context). Extensions of this approach include increasing distractors, fact chaining, and in-context reasoning. However, in these benchmarks, models can exploit existing literal matches between the needle and haystack to simplify the task. To address this, we introduce NoLiMa, a benchmark extending NIAH with a carefully designed needle set, where questions and needles have minimal lexical overlap, requiring models to infer latent associations to locate the needle within the haystack. We evaluate 12 popular LLMs that claim to support contexts of at least 128K tokens. While they perform well in short contexts (<1K), performance degrades significantly as context length increases. At 32K, for instance, 10 models drop below 50% of their strong short-length baselines. Even GPT-4o, one of the top-performing exceptions, experiences a reduction from an almost-perfect baseline of 99.3% to 69.7%. Our analysis suggests these declines stem from the increased difficulty the attention mechanism faces in longer contexts when literal matches are absent, making it harder to retrieve relevant information.

The integration of structured hierarchical embeddings into transformer-based architectures introduces a refined approach to lexical representation, ensuring that multi-scale semantic relationships are preserved without compromising computational efficiency. A projection mechanism that maps tokens onto a structured manifold provides improved lexical alignment, enhancing the adaptability of word representations across diverse linguistic tasks. The structured encoding framework ensures that hierarchical embeddings maintain coherence across varying abstraction levels, allowing for stable transitions between localized syntactic features and global semantic structures. Experimental evaluations indicate that hierarchical embeddings consistently outperform conventional token representations, improving accuracy in linguistic benchmarks while maintaining lower computational overhead. Comparative analysis across multiple domains highlights the ability of hierarchical embeddings to retain contextual consistency, particularly in specialized language applications where structured lexical alignment is essential. Statistical assessments further demonstrate that hierarchical embeddings exhibit enhanced robustness under perturbation conditions, ensuring that linguistic structures remain stable across adversarial text modifications. The integration of hierarchical projections with transformer attention mechanisms enables improved contextual adaptation, ensuring that token representations are dynamically adjusted based on varying linguistic distributions. The refined hierarchical organization of embeddings provides greater interpretability in lexical modeling, facilitating enhanced generalization capabilities across diverse text processing tasks.

Text Style Transfer (TST) is the task of transforming a text to reflect a particular style while preserving its original content. Evaluating TST outputs is a multidimensional challenge, requiring the assessment of style transfer accuracy, content preservation, and naturalness. Using human evaluation is ideal but costly, same as in other natural language processing (NLP) tasks, however, automatic metrics for TST have not received as much attention as metrics for, e.g., machine translation or summarization. In this paper, we examine both set of existing and novel metrics from broader NLP tasks for TST evaluation, focusing on two popular subtasks-sentiment transfer and detoxification-in a multilingual context comprising English, Hindi, and Bengali. By conducting meta-evaluation through correlation with human judgments, we demonstrate the effectiveness of these metrics when used individually and in ensembles. Additionally, we investigate the potential of Large Language Models (LLMs) as tools for TST evaluation. Our findings highlight that certain advanced NLP metrics and experimental-hybrid-techniques, provide better insights than existing TST metrics for delivering more accurate, consistent, and reproducible TST evaluations.

Large language model (LLM)-based agents have recently shown impressive progress in a variety of domains, including open-ended conversation and multi-step decision-making. However, applying these agents to social deduction games such as Werewolf, which requires both strategic decision-making and free-form language interaction, remains non-trivial. Traditional methods based on Counterfactual Regret Minimization (CFR) or reinforcement learning (RL) typically depend on a predefined action space, making them unsuitable for language games with unconstrained text action space. Meanwhile, pure LLM-based agents often suffer from intrinsic biases and require prohibitively large datasets for fine-tuning. We propose Latent Space Policy Optimization (LSPO), an iterative framework that addresses these challenges by first mapping free-form text to a discrete latent space, where methods like CFR and RL can learn strategic policy more effectively. We then translate the learned policy back into natural language dialogues, which are used to fine-tune an LLM via Direct Preference Optimization (DPO). By iteratively alternating between these stages, our LSPO agent progressively enhances both strategic reasoning and language communication. Experiment results on the Werewolf game show that our method improves the agent's performance in each iteration and outperforms existing Werewolf agents, underscoring its promise for free-form language decision-making.

The proliferation of digital interactions across diverse domains, such as healthcare, e-commerce, gaming, and finance, has resulted in the generation of vast volumes of event stream (ES) data. ES data comprises continuous sequences of timestamped events that encapsulate detailed contextual information relevant to each domain. While ES data holds significant potential for extracting actionable insights and enhancing decision-making, its effective utilization is hindered by challenges such as the scarcity of labeled data and the fragmented nature of existing research efforts. Self-Supervised Learning (SSL) has emerged as a promising paradigm to address these challenges by enabling the extraction of meaningful representations from unlabeled ES data. In this survey, we systematically review and synthesize SSL methodologies tailored for ES modeling across multiple domains, bridging the gaps between domain-specific approaches that have traditionally operated in isolation. We present a comprehensive taxonomy of SSL techniques, encompassing both predictive and contrastive paradigms, and analyze their applicability and effectiveness within different application contexts. Furthermore, we identify critical gaps in current research and propose a future research agenda aimed at developing scalable, domain-agnostic SSL frameworks for ES modeling. By unifying disparate research efforts and highlighting cross-domain synergies, this survey aims to accelerate innovation, improve reproducibility, and expand the applicability of SSL to diverse real-world ES challenges.

Intermediate Representations (IRs) are essential in compiler design and program analysis, yet their comprehension by Large Language Models (LLMs) remains underexplored. This paper presents a pioneering empirical study to investigate the capabilities of LLMs, including GPT-4, GPT-3, Gemma 2, LLaMA 3.1, and Code Llama, in understanding IRs. We analyze their performance across four tasks: Control Flow Graph (CFG) reconstruction, decompilation, code summarization, and execution reasoning. Our results indicate that while LLMs demonstrate competence in parsing IR syntax and recognizing high-level structures, they struggle with control flow reasoning, execution semantics, and loop handling. Specifically, they often misinterpret branching instructions, omit critical IR operations, and rely on heuristic-based reasoning, leading to errors in CFG reconstruction, IR decompilation, and execution reasoning. The study underscores the necessity for IR-specific enhancements in LLMs, recommending fine-tuning on structured IR datasets and integration of explicit control flow models to augment their comprehension and handling of IR-related tasks.

Mixture of Experts (MoE) architectures have significantly increased computational efficiency in both research and real-world applications of large-scale machine learning models. However, their scalability and efficiency under memory constraints remain relatively underexplored. In this work, we present joint scaling laws for dense and MoE models, incorporating key factors such as the number of active parameters, dataset size, and the number of experts. Our findings provide a principled framework for selecting the optimal MoE configuration under fixed memory and compute budgets. Surprisingly, we show that MoE models can be more memory-efficient than dense models, contradicting conventional wisdom. To derive and validate the theoretical predictions of our scaling laws, we conduct over 280 experiments with up to 2.7B active parameters and up to 5B total parameters. These results offer actionable insights for designing and deploying MoE models in practical large-scale training scenarios.

The rapid advancement of large language models (LLMs) has increased the need for guardrail models to ensure responsible use, particularly in detecting unsafe and illegal content. While substantial safety data exist in English, multilingual guardrail modeling remains underexplored due to the scarcity of open-source safety data in other languages. To address this gap, we propose a novel two-player Reinforcement Learning (RL) framework, where a generator and a guardrail model co-evolve adversarially to produce high-quality synthetic data for multilingual guardrail training. We theoretically formalize this interaction as a two-player game, proving convergence to a Nash equilibrium. Empirical evaluations show that our model \ours outperforms state-of-the-art models, achieving nearly 10% improvement over LlamaGuard3 (8B) on English benchmarks while being 4.5x faster at inference with a significantly smaller model (0.5B). We achieve substantial advancements in multilingual safety tasks, particularly in addressing the imbalance for lower-resource languages in a collected real dataset. Ablation studies emphasize the critical role of synthetic data generation in bridging the imbalance in open-source data between English and other languages. These findings establish a scalable and efficient approach to synthetic data generation, paving the way for improved multilingual guardrail models to enhance LLM safety. Code, model, and data will be open-sourced at https://github.com/yihedeng9/DuoGuard.

The development in Artificial Intelligence (AI) offers transformative potential for redefining student assessment methodologies. This paper aims to establish the idea of the advancement of Artificial Intelligence (AI) and its prospect in reshaping approaches to assessing students. It creates a system for the evaluation of students performance using Artificial intelligence, and particularly the Gemini API for the generation of questions, grading and report on the students performances. This is to facilitate easy use of the tools in creating, scheduling, and delivering assessments with minimal chances of cheating through options such as full screen and time limit. There are formats of questions in the system which comprises multiple choice, short answers and descriptive questions, developed by Gemini. The most conspicuous feature is the self-checking system whereby the user gets instant feedback for the correct score that each of the students would have scored instantly with explanations about wrong answers. Moreover, the platform has intelligent learning progressions where the user will be able to monitor his/her performances to be recommended a certain level of performance. It will allow students as well as educators to have real-time analytics and feedback on what they are good at and where they need to improve. Not only does it make the assessment easier, but it also improves the levels of accuracy in grading and effectively strengthens a data based learning process for students.