The evolution of Generative AI, particularly Large Language Models (LLMs), has seen remarkable advancements since 2020 with the introduction of models like Chat-GPT and Bard. LLMs have revolutionized tasks, such as writing assistance, code generation, and customer support automation, by leveraging vast amounts of data to generate coherent and contextually relevant natural language (NL) responses [1, 2]. As a subset of Generative AI--systems designed to create new content--LLMs go beyond traditional AI techniques, which focus primarily on analyzing existing data. LLMs, in contrast, are capable of generating text, images, and music that mimic human creativity [3]. This capability is powered by advancements in neural network architectures, especially transformers, which enable LLMs to learn the nuances of human language and produce semantically accurate content [4].

We present rStar-Math to demonstrate that small language models (SLMs) can rival or even surpass the math reasoning capability of OpenAI o1, without distillation from superior models. rStar-Math achieves this by exercising "deep thinking" through Monte Carlo Tree Search (MCTS), where a math policy SLM performs test-time search guided by an SLM-based process reward model. rStar-Math introduces three innovations to tackle the challenges in training the two SLMs: (1) a novel code-augmented CoT data sythesis method, which performs extensive MCTS rollouts to generate step-by-step verified reasoning trajectories used to train the policy SLM; (2) a novel process reward model training method that avoids na\"ive step-level score annotation, yielding a more effective process preference model (PPM); (3) a self-evolution recipe in which the policy SLM and PPM are built from scratch and iteratively evolved to improve reasoning capabilities. Through 4 rounds of self-evolution with millions of synthesized solutions for 747k math problems, rStar-Math boosts SLMs' math reasoning to state-of-the-art levels. On the MATH benchmark, it improves Qwen2.5-Math-7B from 58.8% to 90.0% and Phi3-mini-3.8B from 41.4% to 86.4%, surpassing o1-preview by +4.5% and +0.9%. On the USA Math Olympiad (AIME), rStar-Math solves an average of 53.3% (8/15) of problems, ranking among the top 20% the brightest high school math students. Code and data will be available at https://github.com/microsoft/rStar.

Question answering over temporal knowledge graphs (TKGs) is crucial for understanding evolving facts and relationships, yet its development is hindered by limited datasets and difficulties in generating custom QA pairs. We propose a novel categorization framework based on timeline-context relationships, along with \textbf{TimelineKGQA}, a universal temporal QA generator applicable to any TKGs. The code is available at: \url{https://github.com/PascalSun/TimelineKGQA} as an open source Python package.

In remote sensing scene classification, leveraging the transfer methods with well-trained optical models is an efficient way to overcome label scarcity. However, cloud contamination leads to optical information loss and significant impacts on feature distribution, challenging the reliability and stability of transferred target models. Common solutions include cloud removal for optical data or directly using Synthetic aperture radar (SAR) data in the target domain. However, cloud removal requires substantial auxiliary data for support and pre-training, while directly using SAR disregards the unobstructed portions of optical data. This study presents a scene classification transfer method that synergistically combines multi-modality data, which aims to transfer the source domain model trained on cloudfree optical data to the target domain that includes both cloudy optical and SAR data at low cost. Specifically, the framework incorporates two parts: (1) the collaborative transfer strategy, based on knowledge distillation, enables the efficient prior knowledge transfer across heterogeneous data; (2) the information regulation mechanism (IRM) is proposed to address the modality imbalance issue during transfer. It employs auxiliary models to measure the contribution discrepancy of each modality, and automatically balances the information utilization of modalities during the target model learning process at the sample-level. The transfer experiments were conducted on simulated and real cloud datasets, demonstrating the superior performance of the proposed method compared to other solutions in cloud-covered scenarios. We also verified the importance and limitations of IRM, and further discussed and visualized the modality imbalance problem during the model transfer. Codes are available at https://github.com/wangyuze-csu/ESCCS

While logic puzzles have engaged individuals through problem-solving and critical thinking, the creation of new puzzle rules has largely relied on ad-hoc processes. Pencil puzzles, such as Slitherlink and Sudoku, represent a prominent subset of these games, celebrated for their intellectual challenges rooted in combinatorial logic and spatial reasoning. Despite extensive research into solving techniques and automated problem generation, a unified framework for systematic and scalable rule design has been lacking. Here, we introduce a mathematical framework for defining and systematizing pencil puzzle rules. This framework formalizes grid elements, their positional relationships, and iterative composition operations, allowing for the incremental construction of structures that form the basis of puzzle rules. Furthermore, we establish a formal method to describe constraints and domains for each structure, ensuring solvability and coherence. Applying this framework, we successfully formalized the rules of well-known Nikoli puzzles, including Slitherlink and Sudoku, demonstrating the formal representation of a significant portion (approximately one-fourth) of existing puzzles. These results validate the potential of the framework to systematize and innovate puzzle rule design, establishing a pathway to automated rule generation. By providing a mathematical foundation for puzzle rule creation, this framework opens avenues for computers, potentially enhanced by AI, to design novel puzzle rules tailored to player preferences, expanding the scope of puzzle diversity. Beyond its direct application to pencil puzzles, this work illustrates how mathematical frameworks can bridge recreational mathematics and algorithmic design, offering tools for broader exploration in logic-based systems, with potential applications in educational game design, personalized learning, and computational creativity.

Video Large Language Models (Video LLMs) have recently exhibited remarkable capabilities in general video understanding. However, they mainly focus on holistic comprehension and struggle with capturing fine-grained spatial and temporal details. Besides, the lack of high-quality object-level video instruction data and a comprehensive benchmark further hinders their advancements. To tackle these challenges, we introduce the VideoRefer Suite to empower Video LLM for finer-level spatial-temporal video understanding, i.e., enabling perception and reasoning on any objects throughout the video. Specially, we thoroughly develop VideoRefer Suite across three essential aspects: dataset, model, and benchmark. Firstly, we introduce a multi-agent data engine to meticulously curate a large-scale, high-quality object-level video instruction dataset, termed VideoRefer-700K. Next, we present the VideoRefer model, which equips a versatile spatial-temporal object encoder to capture precise regional and sequential representations. Finally, we meticulously create a VideoRefer-Bench to comprehensively assess the spatial-temporal understanding capability of a Video LLM, evaluating it across various aspects. Extensive experiments and analyses demonstrate that our VideoRefer model not only achieves promising performance on video referring benchmarks but also facilitates general video understanding capabilities.

Accurate and consistent evaluation is crucial for decision-making across numerous fields, yet it remains a challenging task due to inherent subjectivity, variability, and scale. Large Language Models (LLMs) have achieved remarkable success across diverse domains, leading to the emergence of "LLM-as-a-Judge," where LLMs are employed as evaluators for complex tasks. With their ability to process diverse data types and provide scalable, cost-effective, and consistent assessments, LLMs present a compelling alternative to traditional expert-driven evaluations. However, ensuring the reliability of LLM-as-a-Judge systems remains a significant challenge that requires careful design and standardization. This paper provides a comprehensive survey of LLM-as-a-Judge, addressing the core question: How can reliable LLM-as-a-Judge systems be built? We explore strategies to enhance reliability, including improving consistency, mitigating biases, and adapting to diverse assessment scenarios. Additionally, we propose methodologies for evaluating the reliability of LLM-as-a-Judge systems, supported by a novel benchmark designed for this purpose. To advance the development and real-world deployment of LLM-as-a-Judge systems, we also discussed practical applications, challenges, and future directions. This survey serves as a foundational reference for researchers and practitioners in this rapidly evolving field.

Effective preference tuning is pivotal in aligning chatbot responses with human expectations, enhancing user satisfaction and engagement. Traditional approaches, notably Reinforcement Learning from Human Feedback (RLHF) as employed in advanced models like GPT-4, have demonstrated considerable success in this domain. However, RLHF methods are often computationally intensive and resource-demanding, limiting their scalability and accessibility for broader applications. To address these challenges, this study introduces LoRA-Lite Ensemble (LoRA-LiteE), an innovative framework that combines Supervised Fine-tuning (SFT) with Low-Rank Adaptation (LoRA) and Ensemble Learning techniques to effectively aggregate predictions of lightweight models, which aim to achieve a balance between the performance and computational cost. Utilizing the Chatbot Arena benchmark dataset, we conduct a comprehensive comparative analysis among our LoRA-LiteE model, corresponding base models at different scales, and GPT-4 trained with RLHF. Our empirical results demonstrate that the proposed LoRA-LiteE model achieves comparable performance to un-finetuned GPT-4 and outperforms the single larger-scale models under limited resource constraints. These findings highlight that our LoRA-LiteE provides a feasible and efficient methodology for human preference prediction in chatbot systems, enhancing scalability and accessibility, and thereby broadening the applicability of preference-tuned chatbots in resource-constrained environments.

Algorithmic bias is a pressing concern in educational data mining (EDM), as it risks amplifying inequities in learning outcomes. The Area Between ROC Curves (ABROCA) metric is frequently used to measure discrepancies in model performance across demographic groups to quantify overall model fairness. However, its skewed distribution--especially when class or group imbalances exist--makes significance testing challenging. This study investigates ABROCA's distributional properties and contributes robust methods for its significance testing. Specifically, we address (1) whether ABROCA follows any known distribution, (2) how to reliably test for algorithmic bias using ABROCA, and (3) the statistical power achievable with ABROCA-based bias assessments under typical EDM sample specifications. Simulation results confirm that ABROCA does not match standard distributions, including those suited to accommodate skewness. We propose nonparametric randomization tests for ABROCA and demonstrate that reliably detecting bias with ABROCA requires large sample sizes or substantial effect sizes, particularly in imbalanced settings. Findings suggest that ABROCA-based bias evaluation based on sample sizes common in EDM tends to be underpowered, undermining the reliability of conclusions about model fairness. By offering open-source code to simulate power and statistically test ABROCA, this paper aims to foster more reliable statistical testing in EDM research. It supports broader efforts toward replicability and equity in educational modeling.

Following the first part of our project, this paper comprehensively studies two types of extragradient-based methods: anchored extragradient and Nesterov's accelerated extragradient for solving [non]linear inclusions (and, in particular, equations), primarily under the Lipschitz continuity and the co-hypomonotonicity assumptions. We unify and generalize a class of anchored extragradient methods for monotone inclusions to a wider range of schemes encompassing existing algorithms as special cases. We establish $\mathcal{O}(1/k)$ last-iterate convergence rates on the residual norm of the underlying mapping for this general framework and then specialize it to obtain convergence guarantees for specific instances, where $k$ denotes the iteration counter. We extend our approach to a class of anchored Tseng's forward-backward-forward splitting methods to obtain a broader class of algorithms for solving co-hypomonotone inclusions. Again, we analyze $\mathcal{O}(1/k)$ last-iterate convergence rates for this general scheme and specialize it to obtain convergence results for existing and new variants. We generalize and unify Nesterov's accelerated extra-gradient method to a new class of algorithms that covers existing schemes as special instances while generating new variants. For these schemes, we can prove $\mathcal{O}(1/k)$ last-iterate convergence rates for the residual norm under co-hypomonotonicity, covering a class of nonmonotone problems. We propose another novel class of Nesterov's accelerated extragradient methods to solve inclusions. Interestingly, these algorithms achieve both $\mathcal{O}(1/k)$ and $o(1/k)$ last-iterate convergence rates, and also the convergence of iterate sequences under co-hypomonotonicity and Lipschitz continuity. Finally, we provide a set of numerical experiments encompassing different scenarios to validate our algorithms and theoretical guarantees.