Automatic Question Generation (QG) can make reading-comprehension practice and assessment scalable. For educational use, generated questions should diagnose a target comprehension skill and rely on the passage evidence an expert would use to write the question. Existing answer-aware and skill-conditioned QG methods improve controllability, but often leave these skill-to-question and evidence-to-question dependencies implicit. We present SagE-QG, a Skill-aligned and Evidence-Grounding Framework for Question Generation. SagE-QG exposes evidence-grounded trajectories, optimizes multi-aspect rewards for answerability, skill alignment, and human similarity, and applies Counterfactual Evidence Intervention in reinforcement learning. The intervention removes selected evidence and rewards the Interventional Grounding Gap between full-passage and masked-evidence generations. Experiments on three reading-comprehension QG datasets show that SagE-QG delivers stronger educational quality, especially in skill alignment and answerability, while achieving competitive reference-based results.

Current Large Language Models (LLMs) excel at solving complex mathematical problems, yet this proficiency does not inherently translate into effective tutoring. While advanced LLM tutors may leverage multi-agent frameworks or fine-tuning, most still lack a mechanism to systematically accumulate and reuse pedagogical experience over time, limiting their adaptability to diverse student needs during fluid, multi-turn interactions. To bridge this gap, we propose the Reflective Experience-Augmented Tutoring (REAT) framework, which couples experience distillation from historical dialogues with real-time adaptive retrieval. Driven by a multi-agent Observer-Critic-Mentor (OCM) distillation pipeline, REAT reviews past conversational trajectories and distills raw interactions into structured, problem-agnostic pedagogical experiences. During live tutoring, a state-aware retrieval module injects these curated experiences to provide adaptive scaffolding based on the student’s cognitive state. Experiments demonstrate that the proposed framework significantly outperforms both prompt-only and supervised fine-tuning (SFT) baselines, particularly in improving complex, low-scoring teaching scenarios. Crucially, the distilled experiences exhibit robust generalization across diverse model architectures and mathematical datasets.

Prior research indicates that although large language models (LLMs) can precisely articulate the theoretical probability distributions associated with optimal strategic choices, their actual decision-making systematically diverges from these prescriptions—a phenomenon we define as the cognition–behaviour gap in LLMs. For example, in a Rock–Paper–Scissors (RPS) game, LLMs correctly identify the strategy of Nash equilibrium as selecting each action (Rock, Paper, Scissors) with equal probability ¹⁄₃, but their observed choices systematically deviate from this uniform distribution. Through a comprehensive evaluation of 20 state-of-the-art LLMs, we identify two critical insights: (1) we demonstrate that intrinsic biases inherited from pre-training corpora alone are insufficient to explain the observed deviations; (2) we introduce a semantic-free paradigm that strips away intrinsic biases to isolate pure positional bias-LLMs exhibit distinct position preferences—for example, o1 favours the first option, DeepSeek-V3 peaks the middle and DeepSeek-R1 shows a bimodal bias toward first and last positions. Our findings advocate innovation to bridge the gap between strategic reasoning and decision-making in LLMs.

Generative Pre-trained Transformer 4 (GPT-4) demonstrates impressive chain-of-thought reasoning ability. Recent work on self-instruction tuning, such as Alpaca, has focused on enhancing the general proficiency of models. These instructions enable the model to achieve performance comparable to GPT-3.5 on general tasks like open-domain text generation and paraphrasing. However, they fall short of helping the model handle complex reasoning tasks. To bridge the gap, this paper presents LogiCoT, a new instruction-tuning dataset for Logical Chain-of-Thought reasoning with GPT-4. We elaborate on the process of harvesting instructions for prompting GPT-4 to generate chain-of-thought rationales. LogiCoT serves as an instruction set for teaching models of logical reasoning and elicits general reasoning skills.