Large language models (LLMs) excel at reasoning tasks requiring long thought sequences for planning, reflection, and refinement. However, their substantial model size and high computational demands are impractical for widespread deployment. Yet, small language models (SLMs) often struggle to learn long-form CoT reasoning due to their limited capacity, a phenomenon we refer to as the "SLMs Learnability Gap". To address this, we introduce \textbf{Mi}d-\textbf{Co}T \textbf{T}eacher \textbf{A}ssistant Distillation (MiCoTAl), a framework for improving long CoT distillation for SLMs. MiCoTA employs intermediate-sized models as teacher assistants and utilizes intermediate-length CoT sequences to bridge both the capacity and reasoning length gaps. Our experiments on downstream tasks demonstrate that although SLMs distilled from large teachers can perform poorly, by applying MiCoTA, they achieve significant improvements in reasoning performance. Specifically, Qwen2.5-7B-Instruct and Qwen2.5-3B-Instruct achieve an improvement of 3.47 and 3.93 respectively on average score on AIME2024, AMC, Olympiad, MATH-500 and GSM8K benchmarks. To better understand the mechanism behind MiCoTA, we perform a quantitative experiment demonstrating that our method produces data more closely aligned with base SLM distributions. Our insights pave the way for future research into long-CoT data distillation for SLMs.

There is justifiable interest in leveraging conversational AI (CAI) for health across the majority world, but to be effective, CAI must respond appropriately within cultur ally and linguistically diverse context s . Therefore, we need ways to address the fact that current LLMs exclude many lived experience s globally . Various advances are underway which focus on top - down approaches and increas ing training data . In this paper, we aim to complement these with a bottom - up locally - grounded approach based on qualitative data collected during participatory workshops in Latin America. Our goal is to construct a rich and human - centred understanding o f: a) potential areas of cultural misalignment in digital health; b) regional perspectives on chatbots for health and c) strategies for creating culturally - appropriate CAI; with a focus on the understudied Latin American context . Our findings show that academic boundaries on notions of cultur e lose meaning at the ground level and technologies will need to engage with a broad er framework; one that encapsulates the way economics, politics, geogr aphy and local logistics are entangled in cultural experience. To this end, we introduce a framework for ' Pluriversal Conversational AI for H ealth ' which allows for the possibility that more relationality and tolerance, rather than just more data, may be called for .

Task-oriented dialogue (ToD) systems are designed to help users achieve specific goals through natural language interaction. While recent advances in large language models (LLMs) have significantly improved linguistic fluency and contextual understanding, building effective and emotionally intelligent ToD systems remains a complex challenge. Effective ToD systems must optimise for task success, emotional understanding and responsiveness, and precise information conveyance, all within inherently noisy and ambiguous conversational environments. In this work, we investigate architectural, representational, optimisational as well as emotional considerations of ToD systems. We set up systems covering these design considerations with a challenging evaluation environment composed of a natural-language user simulator coupled with an imperfect natural language understanding module. We propose \textbf{LUSTER}, an \textbf{L}LM-based \textbf{U}nified \textbf{S}ystem for \textbf{T}ask-oriented dialogue with \textbf{E}nd-to-end \textbf{R}einforcement learning with both short-term (user sentiment) and long-term (task success) rewards. Our findings demonstrate that combining LLM capability with structured reward modelling leads to more resilient and emotionally responsive ToD systems, offering a practical path forward for next-generation conversational agents.

To reveal when a large language model (LLM) is uncertain about a response, uncertainty quantification commonly produces percentage numbers along with the output. But is this all we can do? We argue that in the output space of LLMs, the space of strings, exist strings expressive enough to summarize the distribution over output strings the LLM deems possible. We lay a foundation for this new avenue of uncertainty explication and present SelfReflect, a theoretically-motivated metric to assess how faithfully a string summarizes an LLM's internal answer distribution. We show that SelfReflect is able to discriminate even subtle differences of candidate summary strings and that it aligns with human judgement, outperforming alternative metrics such as LLM judges and embedding comparisons. With SelfReflect, we investigate a number of self-summarization methods and find that even state-of-the-art reasoning models struggle to explicate their internal uncertainty. But we find that faithful summarizations can be generated by sampling and summarizing. To support the development of this universal form of LLM uncertainties, we publish our metric at https://github.com/apple/ml-selfreflect

We present a new dataset for chart question answering (CQA) constructed from visualization notebooks. The dataset features real-world, multi-view charts paired with natural language questions grounded in analytical narratives. Unlike prior benchmarks, our data reflects ecologically valid reasoning workflows. Benchmarking state-of-the-art multimodal large language models reveals a significant performance gap, with GPT-4.1 achieving an accuracy of 69.3%, underscoring the challenges posed by this more authentic CQA setting.

We develop a theoretical framework that explains how discrete symbolic structures can emerge naturally from continuous neural network training dynamics. By lifting neural parameters to a measure space and modeling training as Wasserstein gradient flow, we show that under geometric constraints, such as group invariance, the parameter measure $μ_t$ undergoes two concurrent phenomena: (1) a decoupling of the gradient flow into independent optimization trajectories over some potential functions, and (2) a progressive contraction on the degree of freedom. These potentials encode algebraic constraints relevant to the task and act as ring homomorphisms under a commutative semi-ring structure on the measure space. As training progresses, the network transitions from a high-dimensional exploration to compositional representations that comply with algebraic operations and exhibit a lower degree of freedom. We further establish data scaling laws for realizing symbolic tasks, linking representational capacity to the group invariance that facilitates symbolic solutions. This framework charts a principled foundation for understanding and designing neurosymbolic systems that integrate continuous learning with discrete algebraic reasoning.

The remarkable advancements of vision and language foundation models in multimodal understanding, reasoning, and generation has sparked growing efforts to extend such intelligence to the physical world, fueling the flourishing of vision-language-action (VLA) models. Despite seemingly diverse approaches, we observe that current VLA models can be unified under a single framework: vision and language inputs are processed by a series of VLA modules, producing a chain of \textit{action tokens} that progressively encode more grounded and actionable information, ultimately generating executable actions. We further determine that the primary design choice distinguishing VLA models lies in how action tokens are formulated, which can be categorized into language description, code, affordance, trajectory, goal state, latent representation, raw action, and reasoning. However, there remains a lack of comprehensive understanding regarding action tokens, significantly impeding effective VLA development and obscuring future directions. Therefore, this survey aims to categorize and interpret existing VLA research through the lens of action tokenization, distill the strengths and limitations of each token type, and identify areas for improvement. Through this systematic review and analysis, we offer a synthesized outlook on the broader evolution of VLA models, highlight underexplored yet promising directions, and contribute guidance for future research, hoping to bring the field closer to general-purpose intelligence.

Existing language learning tools, even those powered by Large Language Models (LLMs), often lack support for polyglot learners to build linguistic connections across vocabularies in multiple languages, provide limited customization for individual learning paces or needs, and suffer from detrimental cognitive offloading. To address these limitations, we design Do-It-Yourself Multilingual Knowledge Graph (DIY-MKG), an open-source system that supports polyglot language learning. DIY-MKG allows the user to build personalized vocabulary knowledge graphs, which are constructed by selective expansion with related words suggested by an LLM. The system further enhances learning through rich annotation capabilities and an adaptive review module that leverages LLMs for dynamic, personalized quiz generation. In addition, DIY-MKG allows users to flag incorrect quiz questions, simultaneously increasing user engagement and providing a feedback loop for prompt refinement. Our evaluation of LLM-based components in DIY-MKG shows that vocabulary expansion is reliable and fair across multiple languages, and that the generated quizzes are highly accurate, validating the robustness of DIY-MKG.

Traditional simulator-based training for maritime professionals is critical for ensuring safety at sea but often depends on subjective trainer assessments of technical skills, behavioral focus, communication, and body language, posing challenges such as subjectivity, difficulty in measuring key features, and cognitive limitations. Addressing these issues, this study develops an AI-driven framework to enhance maritime training by objectively assessing trainee performance through visual focus tracking, speech recognition, and stress detection, improving readiness for high-risk scenarios. The system integrates AI techniques, including visual focus determination using eye tracking, pupil dilation analysis, and computer vision; communication analysis through a maritime-specific speech-to-text model and natural language processing; communication correctness using large language models; and mental stress detection via vocal pitch. Models were evaluated on data from simulated maritime scenarios with seafarers exposed to controlled high-stress events. The AI algorithms achieved high accuracy, with ~92% for visual detection, ~91% for maritime speech recognition, and ~90% for stress detection, surpassing existing benchmarks. The system provides insights into visual attention, adherence to communication checklists, and stress levels under demanding conditions. This study demonstrates how AI can transform maritime training by delivering objective performance analytics, enabling personalized feedback, and improving preparedness for real-world operational challenges.

Reinforcement Learning (RL) has demonstrated significant potential in certain real-world industrial applications, yet its broader deployment remains limited by inherent challenges such as sample inefficiency and unstable learning dynamics. This study investigates the utilization of Genetic Algorithms (GAs) as a mechanism for improving RL performance in an industrially inspired sorting environment. We propose a novel approach in which GA-generated expert demonstrations are used to enhance policy learning. These demonstrations are incorporated into a Deep Q-Network (DQN) replay buffer for experience-based learning and utilized as warm-start trajectories for Proximal Policy Optimization (PPO) agents to accelerate training convergence. Our experiments compare standard RL training with rule-based heuristics, brute-force optimization, and demonstration data, revealing that GA-derived demonstrations significantly improve RL performance. Notably, PPO agents initialized with GA-generated data achieved superior cumulative rewards, highlighting the potential of hybrid learning paradigms, where heuristic search methods complement data-driven RL. The utilized framework is publicly available and enables further research into adaptive RL strategies for real-world applications.