Zheng Zhang

Weibo Gao, Qi Liu, Linan Yue et al.

Learners sharing similar implicit cognitive states often display comparable observable problem-solving performances. Leveraging collaborative connections among such similar learners proves valuable in comprehending human learning. Motivated by the success of collaborative modeling in various domains, such as recommender systems, we aim to investigate how collaborative signals among learners contribute to the diagnosis of human cognitive states (i.e., knowledge proficiency) in the context of intelligent education. The primary challenges lie in identifying implicit collaborative connections and disentangling the entangled cognitive factors of learners for improved explainability and controllability in learner Cognitive Diagnosis (CD). However, there has been no work on CD capable of simultaneously modeling collaborative and disentangled cognitive states. To address this gap, we present Coral, a Collaborative cognitive diagnosis model with disentangled representation learning. Specifically, Coral first introduces a disentangled state encoder to achieve the initial disentanglement of learners' states. Subsequently, a meticulously designed collaborative representation learning procedure captures collaborative signals. It dynamically constructs a collaborative graph of learners by iteratively searching for optimal neighbors in a context-aware manner. Using the constructed graph, collaborative information is extracted through node representation learning. Finally, a decoding process aligns the initial cognitive states and collaborative states, achieving co-disentanglement with practice performance reconstructions. Extensive experiments demonstrate the superior performance of Coral, showcasing significant improvements over state-of-the-art methods across several real-world datasets. Our code is available at https://github.com/bigdata-ustc/Coral.

Chen Li, Weiqi Wang, Jingcheng Hu et al.

Mathematical capabilities were previously believed to emerge in common language models only at a very large scale or require extensive math-related pre-training. This paper shows that the LLaMA-2 7B model with common pre-training already exhibits strong mathematical abilities, as evidenced by its impressive accuracy of 97.7% and 72.0% on the GSM8K and MATH benchmarks, respectively, when selecting the best response from 256 random generations. The primary issue with the current base model is the difficulty in consistently eliciting its inherent mathematical capabilities. Notably, the accuracy for the first answer drops to 49.5% and 7.9% on the GSM8K and MATH benchmarks, respectively. We find that simply scaling up the SFT data can significantly enhance the reliability of generating correct answers. However, the potential for extensive scaling is constrained by the scarcity of publicly available math questions. To overcome this limitation, we employ synthetic data, which proves to be nearly as effective as real data and shows no clear saturation when scaled up to approximately one million samples. This straightforward approach achieves an accuracy of 82.6% on GSM8K and 40.6% on MATH using LLaMA-2 7B models, surpassing previous models by 14.2% and 20.8%, respectively. We also provide insights into scaling behaviors across different reasoning complexities and error types.

Zheng Zhang

Large Language Models (LLMs) display striking surface fluency yet systematically fail at tasks requiring symbolic reasoning, arithmetic accuracy, and logical consistency. This paper offers a structural diagnosis of such failures, revealing a persistent gap between \textit{comprehension} and \textit{competence}. Through controlled experiments and architectural analysis, we demonstrate that LLMs often articulate correct principles without reliably applying them--a failure rooted not in knowledge access, but in computational execution. We term this phenomenon the computational \textit{split-brain syndrome}, where instruction and action pathways are geometrically and functionally dissociated. This core limitation recurs across domains, from mathematical operations to relational inferences, and explains why model behavior remains brittle even under idealized prompting. We argue that LLMs function as powerful pattern completion engines, but lack the architectural scaffolding for principled, compositional reasoning. Our findings delineate the boundary of current LLM capabilities and motivate future models with metacognitive control, principle lifting, and structurally grounded execution. This diagnosis also clarifies why mechanistic interpretability findings may reflect training-specific pattern coordination rather than universal computational principles, and why the geometric separation between instruction and execution pathways suggests limitations in neural introspection and mechanistic analysis.

Xinjie Sun, Qi Liu, Kai Zhang et al.

Cognitive diagnosis (CD) aims to reveal students' proficiency in specific knowledge concepts. With the increasing adoption of intelligent education applications, accurately assessing students' knowledge mastery has become an urgent challenge. Although existing cognitive diagnosis frameworks enhance diagnostic accuracy by analyzing students' explicit response records, they primarily focus on individual knowledge state, failing to adequately reflect the relative ability performance of students within hierarchies. To address this, we propose the Hierarchy Constraint-Aware Cognitive Diagnosis Framework (HCD), designed to more accurately represent student ability performance within real educational contexts. Specifically, the framework introduces a hierarchy mapping layer to identify students' levels. It then employs a hierarchy convolution-enhanced attention layer for in-depth analysis of knowledge concepts performance among students at the same level, uncovering nuanced differences. A hierarchy inter-sampling attention layer captures performance differences across hierarchies, offering a comprehensive understanding of the relationships among students' knowledge state. Finally, through personalized diagnostic enhancement, the framework integrates hierarchy constraint perception features with existing models, improving the representation of both individual and group characteristics. This approach enables precise inference of students' knowledge state. Research shows that this framework not only reasonably constrains changes in students' knowledge states to align with real educational settings, but also supports the scientific rigor and fairness of educational assessments, thereby advancing the field of cognitive diagnosis.

Jie Shao, Dequan Wang, Xiangyang Xue et al.

This paper proposes the problem of point-and-count as a test case to break the what-and-where deadlock. Different from the traditional detection problem, the goal is to discover key salient points as a way to localize and count the number of objects simultaneously. We propose two alternatives, one that counts first and then point, and another that works the other way around. Fundamentally, they pivot around whether we solve "what" or "where" first. We evaluate their performance on dataset that contains multiple instances of the same class, demonstrating the potentials and their synergies. The experiences derive a few important insights that explains why this is a much harder problem than classification, including strong data bias and the inability to deal with object scales robustly in state-of-art convolutional neural networks.