Large Language Models (LLMs) often struggle with structural ambiguity in optimization problems, where a single problem admits multiple related but conflicting modeling paradigms, hindering effective solution generation. To address this, we propose Dual-Cluster Memory Agent (DCM-Agent) to enhance performance by leveraging historical solutions in a training-free manner. Central to this is Dual-Cluster Memory Construction. This agent assigns historical solutions to modeling and coding clusters, then distills each cluster’s content into three structured types: Approach, Checklist, and Pitfall. This process derives generalizable guidance knowledge. Furthermore, this agent introduces Memory-augmented Inference to dynamically navigate solution paths, detect and repair errors, and adaptively switch reasoning paths with structured knowledge. The experiments across seven optimization benchmarks demonstrate that DCM-Agent achieves an average performance improvement of 11%- 21%. Notably, our analysis reveals a “knowledge inheritance” phenomenon: memory constructed by larger models can guide smaller models toward superior performance, highlighting the framework’s scalability and efficiency.

Long-term conversational memory is a core capability for LLM-baseddialogue systems, yet existing benchmarks and evaluation protocolsprimarily focus on surface-level factual recall.In realistic interactions, appropriate responses often depend onimplicit constraints such as user state, goals, or values that are notexplicitly queried later.To evaluate this setting, we introduce LoCoMo-Plus, a benchmarkfor assessing cognitive memory under cue–trigger semantic disconnect,where models must retain and apply latent constraints across longconversational contexts.We further show that conventional string-matching metrics and explicittask-type prompting are misaligned with such scenarios, and propose aunified evaluation framework based on constraint consistency.Experiments across diverse backbone models, retrieval-based methods, andmemory systems demonstrate that cognitive memory remains challenging andreveals failures not captured by existing benchmarks.Our code and evaluation framework are publicly available at https://github.com/xjtuleeyf/Locomo-Plus.

Geometry problem solving (GPS) poses significant challenges for Multimodal Large Language Models (MLLMs) in diagram comprehension, knowledge application, long-step reasoning, and auxiliary line construction. However, current benchmarks lack fine-grained evaluation for long-step problems necessitating auxiliary construction. To address these limitations, we present GeoLaux, a fine-grained annotated dataset comprising 2186 calculation and proof problems. It features long-step reasoning (with an average solution length of 6.51 steps, maximum of 24 steps) and auxiliary line construction (required in 41.8% of problems). Building on the dataset, we conduct a comprehensive five-dimensional evaluation of 23 leading MLLMs. The evaluation yields three pivotal findings: First, models perform significantly worse on long-step problems compared to short-step ones, with 18 models exhibiting a performance drop of over 50%. Second, it is crucial to enhance models’ understanding, awareness, and proficiency in auxiliary line construction, which is vital for overall geometric reasoning. Third, limited answer hints effectively improve process correctness, whereas explicit answers lead models to neglect intermediate reasoning steps. These findings position GeoLaux both to benchmark MLLMs geometry reasoning abilities and to guide their improvement. Data and code are available at https://github.com/Candice-yu/GeoLaux

Visual Question Generation (VQG) research focuses predominantly on natural images while neglecting the diagram, which is a critical component in educational materials. To meet the needs of pedagogical assessment, we propose the Diagram-Driven Course Questions Generation (DDCQG) task and construct DiagramQG, a comprehensive dataset with 15,720 diagrams and 25,798 questions across 37 subjects and 371 courses. Our approach employs course and input text constraints to generate course-relevant questions about specific diagram elements. We reveal three challenges of DDCQG: domain-specific knowledge requirements across courses, long-tail distribution in course coverage, and high information density in diagrams. To address these, we propose the Hierarchical Knowledge Integration framework (HKI-DDCQG), which utilizes trainable CLIP for identifying relevant diagram patches, leverages frozen vision-language models for knowledge extraction, and generates questions with trainable T5. Experiments demonstrate that HKI-DDCQG outperforms existing models on DiagramQG while maintaining strong generalizability across natural image datasets, establishing a strong baseline for DDCQG.

Large language models demonstrate remarkable capabilities across various domains, especially mathematics and logic reasoning. However, current evaluations overlook physics-based reasoning - a complex task requiring physics theorems and constraints. We present PhysReason, a 1,200-problem benchmark comprising knowledge-based (25%) and reasoning-based (75%) problems, where the latter are divided into three difficulty levels (easy, medium, hard). Notably, problems require an average of 8.1 solution steps, with hard requiring 15.6, reflecting the complexity of physics-based reasoning. We propose the Physics Solution Auto Scoring Framework, incorporating efficient answer-level and comprehensive step-level evaluations. Top-performing models like Deepseek-R1, Gemini-2.0-Flash-Thinking, and o3-mini-high achieve less than 60% on answer-level evaluation, with performance dropping from knowledge questions (75.11%) to hard problems (31.95%). Through step-level evaluation, we identified four key bottlenecks: Physics Theorem Application, Physics Process Understanding, Calculation, and Physics Condition Analysis. These findings position PhysReason as a novel and comprehensive benchmark for evaluating physics-based reasoning capabilities in large language models.

Current clustering-based open relation extraction (OpenRE) methods usually apply clustering algorithms on top of pre-trained language models. However, this practice has three drawbacks. First, embeddings from language models are high-dimensional and anisotropic, so using simple metrics to calculate distances between these embeddings may not accurately reflect the relational similarity. Second, there exists a gap between the pre-trained language models and downstream clustering for their different objective forms. Third, clustering with embeddings deviates from the primary aim of relation extraction, as it does not directly obtain relations. In this work, we propose a new idea for OpenRE in the era of LLMs, that is, extracting relational phrases and directly exploiting the knowledge in LLMs to assess the semantic similarity between phrases without relying on any additional metrics. Based on this idea, we developed a framework, oreLLM, that makes two LLMs work collaboratively to achieve clustering and address the above issues. Experimental results on different datasets show that oreLLM outperforms current baselines by 1.4%∼ 3.13% in terms of clustering accuracy.

Relation clustering is a general approach for open relation extraction (OpenRE). Current methods have two major problems. One is that their good performance relies on large amounts of labeled and pre-defined relational instances for pre-training, which are costly to acquire in reality. The other is that they only focus on learning a high-dimensional metric space to measure the similarity of novel relations and ignore the specific relational representations of clusters. In this work, we propose a new prompt-based framework named MatchPrompt, which can realize OpenRE with efficient knowledge transfer from only a few pre-defined relational instances as well as mine the specific meanings for cluster interpretability. To our best knowledge, we are the first to introduce a prompt-based framework for unlabeled clustering. Experimental results on different datasets show that MatchPrompt achieves the new SOTA results for OpenRE.

Relation prediction in knowledge graphs (KGs) aims at predicting missing relations in incomplete triples, whereas the dominant embedding paradigm has a restriction on handling unseen entities during testing. In the real-world scenario, the inductive setting is more common because entities in the training process are finite. Previous methods capture an inductive ability by implicit logic in KGs. However, it would be challenging to preciously acquire entity-independent relational semantics of compositional logic rules and to deal with the deficient supervision of logic caused by the scarcity of relational semantics. To this end, we propose a novel graph convolutional network (GCN)-based model LogCo with logical reasoning by contrastive representations. LogCo firstly extracts enclosing subgraphs and relational paths between two entities to supply the entity-independence. Then a contrastive strategy for relational path instances and the subgraph is proposed for the issue of deficient supervision. The contrastive representations are learned for a joint training regime. Finally, prediction results and logic rules for reasoning are attained. Comprehensive experiments on twelve inductive datasets show that LogCo achieves outstanding performance comparing with state-of-the-art inductive relation prediction baselines.