Jailbreak attacks pose persistent threats to large language models (LLMs). Current safety alignment methods have attempted to address these issues, but they experience two significant limitations: insufficient safety alignment depth and unrobust internal defense mechanisms. These limitations make them vulnerable to adversarial attacks such as prefilling and refusal direction manipulation. We introduce DeepRefusal, a robust safety alignment framework that overcomes these issues. DeepRefusal forces the model to dynamically rebuild its refusal mechanisms from jailbreak states. This is achieved by probabilistically ablating the refusal direction across layers and token depths during fine-tuning. Our method not only defends against prefilling and refusal direction attacks but also demonstrates strong resilience against other unseen jailbreak strategies. Extensive evaluations on four open-source LLM families and six representative attacks show that DeepRefusal reduces attack success rates by approximately 95%, while maintaining model capabilities with minimal performance degradation.

Despite the abundance of prior social strategies possessed by humans, there remains a paucity of research dedicated to their transfer and integration into social agents. Our proposed SOTOPIA-Ω framework aims to address and bridge this gap, with a particular focus on enhancing the social capabilities of language agents. This framework dynamically injects a variety of social strategies into expert agents, thereby automating the construction of high-quality social dialogue training corpus. Additionally, we introduce the concept of Social Instruction Following (S-IF) and propose two new S-IF evaluation metrics that are complementary to social capability. We demonstrate that several 7B models trained on high-quality corpus not only significantly surpasses the expert agent (GPT-4) in achieving social goals but also enhances S-IF performance. Analysis and variant experiments validate the advantages of dynamic construction, which can especially break the agent’s prolonged deadlock.

“CCL2024-Eval任 务7为 中 小 学 生 作 文 流 畅 性 评 价 (Chinese Essay Fluency Evalua-tion,CEFE),该任务定义了三项重要且富有挑战性的问题,包括中小学作文病句类型识别、中小学作文病句改写、以及中小学作文流畅性评级。本队伍参加了评测任务7的三项子任务,分别获得了45.19、43.90和45.84的得分。本报告详细介绍本队伍在三个子任务上采用的技术方法,并对评测结果进行分析。”

“电信网络诈骗领域的案件分类作为文本分类的一项落地应用,其目的是为相关案件进行智能化的分析,有助于公安部门掌握诈骗案件的特点,针对性的预防、制止、侦查。本文以此问题为基础,从模型设计、训练过程、数据增强三个方面进行了研究,通过CLS动态加权平均、Multi-Sample Dropout、对抗训练FGM、回译等方法显著提升了模型对诈骗案件描述的分类性能。”

“命名实体识别旨在自动识别出文本中具有特定意义的实体(例如,人名、地名),古籍文献中的命名实体识别通过识别人名、书籍、官职等实体,为深度挖掘、组织古汉语人文知识提供重要支撑。现有的中文命名实体识别方法主要聚焦在现代文,但古籍中的实体识别具有更大的挑战,表现在实体的歧义性和边界模糊性两方面。由于古籍行文简练,单字表达加剧了实体的歧义性问题,句读及分词断句难度的提升使实体边界的识别更具挑战性。为有效处理上述问题,本文提出一种基于信息论及篇章信息的古籍命名实体识别方法。通过检索古籍文本的来源信息融入篇章先验知识,并在同一篇章的古籍文本上采取滑动窗口采样增强,以引入篇章背景信息,有效缓解实体歧义性问题。此外,在信息论视角下,约束实体的上下文信息及实体本身特征的编码,最大程度保留泛化特征,去除冗余信息,缓解实体边界模糊的问题,在词义复杂多样、句读困难的古文典籍中提升命名实体识别性能。最终,在token-wise和span-level感知的命名实体识别基础框架下,本文的方法取得了最优的评测性能。”