The impressive performance of large language models (LLMs) also brings inherent toxicity risks, prompting the need for effective detoxification to support responsible deployment. Prevailing methods generally follow an inflexible model-specific fashion, addressing only individual models or model families. Moreover, overlooking the underlying toxic risks involved in the input prefix can lead to toxic accumulation during autoregressive generation. Existing methods rely on external strong attribute interventions to address this issue, which further exacerbates contextual semantic inconsistencies and makes it difficult to balance toxicity efficacy and generation quality. To address these concerns, we propose a novel Model-Agnostic Adaptive Detoxification (MAAD) framework. To address accumulating toxicity, we present prefix heuristics that serve as contextual signals, guiding the base LLM toward safer generation. Along this line, we construct an antidote dataset to support a lightweight model, Detoxifier, which steers the base LLM to make in-scope and reliable detoxifying distribution adjustments while preserving fluency and contextual understanding. Designed as an easy-to-deploy module, Detoxifier requires a small amount of data and can be seamlessly applied to various base LLMs with one-off training. Since over-purifying often reduces diversity, we also propose a dynamic truncation method called CW-cutoff sampling to trade off language model quality and diversity. Extensive experiments demonstrate that MAAD strikes a better balance between detoxification effectiveness and generation quality, while also maintaining model utility.

Dianyun Wang, Qingsen Ma, Yuhu Shang, Zhifeng Lu, Zhenbo Xu, Lechen Ning, Huijia Wu, Zhaofeng He. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

The task of Knowledge-Based Question Generation (KBQG) involves generating natural language questions from structured knowledge sources, posing unique challenges in balancing linguistic diversity and semantic relevance. Existing models often focus on maximizing surface-level similarity to ground-truth questions, neglecting the need for diverse syntactic forms and leading to semantic drift during generation. To overcome these challenges, we propose Refine-Reinforced Diverse Question Generation (R2DQG), a two-phase framework leveraging a generation-then-refinement paradigm. The Generator first constructs a diverse set of expressive templates using dependency parse tree similarity, capturing a wide range of syntactic patterns and styles. These templates guide the creation of question drafts, ensuring both diversity and semantic relevance. In the second phase, a Corrector module refines the drafts to mitigate semantic drift and enhance overall coherence and quality. Experiments on public datasets show that R2DQG outperforms state-of-the-art models in generating diverse, contextually accurate questions. Moreover, synthetic datasets generated by R2DQG enhance downstream QA performance, underscoring the practical utility of our approach.

Knowledge tracing(KT) is one of the supporting technologies for adaptive learning. It obtains learners' knowledge level by analyzing their online historical answer records, thereby predicting their future answer performance. Since there are a large number of continuous or repeated exercises in KT datasets, and traditional graph structures can usually only represent one-to-one or one-to-many relationships. Some existing models tend to ignore the potential connections between knowledge concept, and thus fail to accurately obtain high-level representations of students' knowledge state. Hypergraphs can directly represent many-to-many relationships and have advantages in extracting high-level information and processing complex relationships. In this paper, we propose a KT method (HTKT) based on Hypergraph Transformer. This method is based on the Encoder-only architecture and introduces a hypergraph attention mechanism to obtain a high-order feature representation of students' knowledge state. Experiments on multiple public and general datasets show that our proposed HTKT method outperforms existing baseline methods.

Knowledge Tracing (KT) involves using deep neural networks (DNNs) to track students’ learning progress, but over-fitting can be an issue with small datasets. Adversarial examples have been introduced to improve generalization, but they may overlook individual student differences. To address this issue, we propose a new model called Customized Adversarial Training Knowledge Tracing (CATKT). The model generates unique adversarial perturbations for each sample based on the characteristics of knowledge tracking tasks, thereby better adapting to students’ learning traits and enhancing the effectiveness and accuracy of knowledge tracking tasks. Specifically, CATKT can dynamically adjust the level of perturbations according to the difficulty of the knowledge, adding non-uniform and effective perturbations to each interaction embedding, and replacing the original labels with adaptively smoothed labels to improve task accuracy. Experimental results show that CATKT outperforms previous knowledge tracking methods in terms of performance and provides new ideas and methods for teaching assessment and personalized learning in the education field.

To reduce the repetitive and complex work of instructors, exam paper generation (EPG) technique has become a salient topic in the intelligent education field, which targets at generating high-quality exam paper automatically according to instructor-specified assessment criteria. The current advances utilize the ability of heuristic algorithms to optimize several well-known objective constraints, such as difficulty degree, number of questions, etc., for producing optimal solutions. However, in real scenarios, considering other equally relevant objectives (e.g., distribution of exam scores, skill coverage) is extremely important. Besides, how to develop an automatic multi-objective solution that finds an optimal subset of questions from a huge search space of large- sized question datasets and thus composes a high-quality exam paper is urgent but non-trivial. To this end, we skillfully design a reinforcement learning guided Multi-Objective Exam Paper Generation framework, termed MOEPG, to simultaneously optimize three exam domain-specific objectives including difficulty degree, distribution of exam scores, and skill coverage. Specifically, to accurately measure the skill proficiency of the examinee group, we first employ deep knowledge tracing to model the interaction information between examinees and response logs. We then design the flexible Exam Q-Network, a function approximator, which automatically selects the appropriate question to update the exam paper composition process. Later, MOEPG divides the decision space into multiple subspaces to better guide the updated direction of the exam paper. Through extensive experiments on two real-world datasets, we demonstrate that MOEPG is feasible in addressing the multiple dilemmas of exam paper generation scenario1. 1https://github.com/researcher-tiger/MOEPG