We introduce PRELUDE, a benchmark for evaluating long-context understanding through the task of determining whether a character's prequel story is consistent with the canonical narrative of the original book. Our task poses a stronger demand for global comprehension and deep reasoning than existing benchmarks -- as the prequels are not part of the original story, assessing their plausibility typically requires searching and integrating information that is only indirectly related. Empirically, 88% of instances require evidence from multiple parts of the narrative. Experimental results highlight the challenge of our task: in-context learning, RAG and in-domain training with state-of-the-art LLMs, and commercial DeepResearch services, lag behind humans by >15%. A further human study reveals that models often produce correct answers with flawed reasoning, leading to an over 30% gap in reasoning accuracy compared to humans. These findings underscore the substantial room for improvement in long-context understanding and reasoning.

Knowledge syntheses (literature reviews) are essential to health professions education (HPE), consolidating findings to advance theory and practice. However, they are labor-intensive, especially during data extraction. Artificial Intelligence (AI)-assisted extraction promises efficiency but raises concerns about accuracy, making it critical to distinguish AI 'hallucinations' (fabricated content) from legitimate interpretive differences. We developed an extraction platform using large language models (LLMs) to automate data extraction and compared AI to human responses across 187 publications and 17 extraction questions from a published scoping review. AI-human, human-human, and AI-AI consistencies were measured using interrater reliability (categorical) and thematic similarity ratings (open-ended). Errors were identified by comparing extracted responses to source publications. AI was highly consistent with humans for concrete, explicitly stated questions (e.g., title, aims) and lower for questions requiring subjective interpretation or absent in text (e.g., Kirkpatrick's outcomes, study rationale). Human-human consistency was not higher than AI-human and showed the same question-dependent variability. Discordant AI-human responses (769/3179 = 24.2%) were mostly due to interpretive differences (18.3%); AI inaccuracies were rare (1.51%), while humans were nearly three times more likely to state inaccuracies (4.37%). Findings suggest AI variability depends more on interpretability than hallucination. Repeating AI extraction can identify interpretive complexity or ambiguity, refining processes before human review. AI can be a transparent, trustworthy partner in knowledge synthesis, though caution is needed to preserve critical human insights.

Recent R1-Zero-like research further demonstrates that reasoning extension has given large language models (LLMs) unprecedented reasoning capabilities, and Reinforcement Learning is the core technology to elicit its complex reasoning. However, conducting RL experiments directly on hyperscale models involves high computational costs and resource demands, posing significant risks. We propose the Compass-Thinker-7B model, which aims to explore the potential of Reinforcement Learning with less computational resources and costs, and provides insights for further research into RL recipes for larger models. Compass-Thinker-7B is trained from an open source model through a specially designed Reinforcement Learning Pipeline. We curate a dataset of 30k verifiable mathematics problems for the Reinforcement Learning Pipeline. By configuring data and training settings with different difficulty distributions for different stages, the potential of the model is gradually released and the training efficiency is improved. Extensive evaluations show that Compass-Thinker-7B possesses exceptional reasoning potential, and achieves superior performance on mathematics compared to the same-sized RL model. Especially in the challenging AIME2024 evaluation, Compass-Thinker-7B achieves 40% accuracy.

In English education tutoring, teacher feedback is essential for guiding students. Recently, AI-based tutoring systems have emerged to assist teachers; however, these systems require high-quality and large-scale teacher feedback data, which is both time-consuming and costly to generate manually. In this study, we propose FEAT, a cost-effective framework for generating teacher feedback, and have constructed three complementary datasets: (1) DIRECT-Manual (DM), where both humans and large language models (LLMs) collaboratively generate high-quality teacher feedback, albeit at a higher cost; (2) DIRECT-Generated (DG), an LLM-only generated, cost-effective dataset with lower quality;, and (3) DIRECT-Augmented (DA), primarily based on DG with a small portion of DM added to enhance quality while maintaining cost-efficiency. Experimental results showed that incorporating a small portion of DM (5-10%) into DG leads to superior performance compared to using 100% DM alone.

Large language models (LLMs) have created new opportunities to assist teachers and support student learning. While researchers have explored various prompt engineering approaches in educational contexts, the degree to which these approaches generalize across domains--such as science, computing, and engineering--remains underexplored. In this paper, we introduce Chain-of-Thought Prompting + Active Learning (CoTAL), an LLM-based approach to formative assessment scoring that (1) leverages Evidence-Centered Design (ECD) to align assessments and rubrics with curriculum goals, (2) applies human-in-the-loop prompt engineering to automate response scoring, and (3) incorporates chain-of-thought (CoT) prompting and teacher and student feedback to iteratively refine questions, rubrics, and LLM prompts. Our findings demonstrate that CoTAL improves GPT-4's scoring performance across domains, achieving gains of up to 38.9% over a non-prompt-engineered baseline (i.e., without labeled examples, chain-of-thought prompting, or iterative refinement). Teachers and students judge CoTAL to be effective at scoring and explaining responses, and their feedback produces valuable insights that enhance grading accuracy and explanation quality.

Matching markets are often organized in a multi-stage and decentralized manner. Moreover, participants in real-world matching markets often have uncertain preferences.

However, they often suffer from costly training and inference due to a tremendous amount of computations through neural networks.

The continual learning problem involves training models with limited capacity to perform well on a set of an unknown number of sequentially arriving tasks.