Jiawen Deng

Zhexin Zhang, Jiale Cheng, Hao Sun et al. · tsinghua

Large pretrained language models can easily produce toxic or biased content, which is prohibitive for practical use. In order to detect such toxic generations, existing methods rely on templates, real-world data extraction, crowdsourcing workers, or automatic generation to construct adversarial contexts that are likely to induce toxic generations. However, what type of context is more likely to induce unsafe responses is still under-explored. In this paper, we identify that context toxicity and context category (e.g., \textit{profanity}, \textit{insult}, \textit{drugs}, etc.) are two important factors to cause safety issues in response generation. Hence, we propose a method called \emph{reverse generation} to construct adversarial contexts conditioned on a given response, with the flexibility to control category, toxicity level, and inductivity of the generated contexts. Via reverse generation, we augment the existing BAD dataset and construct a new dataset BAD+ which contains more than 120K diverse and highly inductive contexts in 12 categories. We test three popular pretrained dialogue models (Blender, DialoGPT, and Plato2) and find that BAD+ can largely expose their safety problems. Furthermore, we show that BAD+ can greatly enhance the safety of generation and reveal the key factors of safety improvement. Our code and dataset is available at \url{https://github.com/thu-coai/Reverse_Generation}.

Sha Yuan, Hanyu Zhao, Shuai Zhao et al. · bytedance, pku

With the rapid development of deep learning, training Big Models (BMs) for multiple downstream tasks becomes a popular paradigm. Researchers have achieved various outcomes in the construction of BMs and the BM application in many fields. At present, there is a lack of research work that sorts out the overall progress of BMs and guides the follow-up research. In this paper, we cover not only the BM technologies themselves but also the prerequisites for BM training and applications with BMs, dividing the BM review into four parts: Resource, Models, Key Technologies and Application. We introduce 16 specific BM-related topics in those four parts, they are Data, Knowledge, Computing System, Parallel Training System, Language Model, Vision Model, Multi-modal Model, Theory&Interpretability, Commonsense Reasoning, Reliability&Security, Governance, Evaluation, Machine Translation, Text Generation, Dialogue and Protein Research. In each topic, we summarize clearly the current studies and propose some future research directions. At the end of this paper, we conclude the further development of BMs in a more general view.

Hao Sun, Zhexin Zhang, Jiawen Deng et al.

With the rapid popularity of large language models such as ChatGPT and GPT-4, a growing amount of attention is paid to their safety concerns. These models may generate insulting and discriminatory content, reflect incorrect social values, and may be used for malicious purposes such as fraud and dissemination of misleading information. Evaluating and enhancing their safety is particularly essential for the wide application of large language models (LLMs). To further promote the safe deployment of LLMs, we develop a Chinese LLM safety assessment benchmark. Our benchmark explores the comprehensive safety performance of LLMs from two perspectives: 8 kinds of typical safety scenarios and 6 types of more challenging instruction attacks. Our benchmark is based on a straightforward process in which it provides the test prompts and evaluates the safety of the generated responses from the evaluated model. In evaluation, we utilize the LLM's strong evaluation ability and develop it as a safety evaluator by prompting. On top of this benchmark, we conduct safety assessments and analyze 15 LLMs including the OpenAI GPT series and other well-known Chinese LLMs, where we observe some interesting findings. For example, we find that instruction attacks are more likely to expose safety issues of all LLMs. Moreover, to promote the development and deployment of safe, responsible, and ethical AI, we publicly release SafetyPrompts including 100k augmented prompts and responses by LLMs.

Yequan Wang, Jiawen Deng, Aixin Sun et al.

Recently, amounts of works utilize perplexity~(PPL) to evaluate the quality of the generated text. They suppose that if the value of PPL is smaller, the quality(i.e. fluency) of the text to be evaluated is better. However, we find that the PPL referee is unqualified and it cannot evaluate the generated text fairly for the following reasons: (i) The PPL of short text is larger than long text, which goes against common sense, (ii) The repeated text span could damage the performance of PPL, and (iii) The punctuation marks could affect the performance of PPL heavily. Experiments show that the PPL is unreliable for evaluating the quality of given text. Last, we discuss the key problems with evaluating text quality using language models.

Jiawen Deng, Jiale Cheng, Hao Sun et al.

As generative large model capabilities advance, safety concerns become more pronounced in their outputs. To ensure the sustainable growth of the AI ecosystem, it's imperative to undertake a holistic evaluation and refinement of associated safety risks. This survey presents a framework for safety research pertaining to large models, delineating the landscape of safety risks as well as safety evaluation and improvement methods. We begin by introducing safety issues of wide concern, then delve into safety evaluation methods for large models, encompassing preference-based testing, adversarial attack approaches, issues detection, and other advanced evaluation methods. Additionally, we explore the strategies for enhancing large model safety from training to deployment, highlighting cutting-edge safety approaches for each stage in building large models. Finally, we discuss the core challenges in advancing towards more responsible AI, including the interpretability of safety mechanisms, ongoing safety issues, and robustness against malicious attacks. Through this survey, we aim to provide clear technical guidance for safety researchers and encourage further study on the safety of large models.

Jincenzi Wu, Zhuang Chen, Jiawen Deng et al.

Theory of mind (ToM) refers to humans' ability to understand and infer the desires, beliefs, and intentions of others. The acquisition of ToM plays a key role in humans' social cognition and interpersonal relations. Though indispensable for social intelligence, ToM is still lacking for modern AI and NLP systems since they cannot access the human mental state and cognitive process beneath the training corpus. To empower AI systems with the ToM ability and narrow the gap between them and humans, in this paper, we propose COKE: the first cognitive knowledge graph for machine theory of mind. Specifically, COKE formalizes ToM as a collection of 45k+ manually verified cognitive chains that characterize human mental activities and subsequent behavioral/affective responses when facing specific social circumstances. In addition, we further generalize COKE using LLMs and build a powerful generation model COLM tailored for cognitive reasoning. Experimental results in both automatic and human evaluation demonstrate the high quality of COKE, the superior ToM ability of COLM, and its potential to significantly enhance social applications.

Eddie Guo, Mehul Gupta, Jiawen Deng et al.

Objective: To assess the performance of the OpenAI GPT API in accurately and efficiently identifying relevant titles and abstracts from real-world clinical review datasets and compare its performance against ground truth labelling by two independent human reviewers. Methods: We introduce a novel workflow using the OpenAI GPT API for screening titles and abstracts in clinical reviews. A Python script was created to make calls to the GPT API with the screening criteria in natural language and a corpus of title and abstract datasets that have been filtered by a minimum of two human reviewers. We compared the performance of our model against human-reviewed papers across six review papers, screening over 24,000 titles and abstracts. Results: Our results show an accuracy of 0.91, a sensitivity of excluded papers of 0.91, and a sensitivity of included papers of 0.76. On a randomly selected subset of papers, the GPT API demonstrated the ability to provide reasoning for its decisions and corrected its initial decision upon being asked to explain its reasoning for a subset of incorrect classifications. Conclusion: The GPT API has the potential to streamline the clinical review process, save valuable time and effort for researchers, and contribute to the overall quality of clinical reviews. By prioritizing the workflow and acting as an aid rather than a replacement for researchers and reviewers, the GPT API can enhance efficiency and lead to more accurate and reliable conclusions in medical research.

Jingyan Zhou, Jiawen Deng, Fei Mi et al.

The research of open-domain dialog systems has been greatly prospered by neural models trained on large-scale corpora, however, such corpora often introduce various safety problems (e.g., offensive languages, biases, and toxic behaviors) that significantly hinder the deployment of dialog systems in practice. Among all these unsafe issues, addressing social bias is more complex as its negative impact on marginalized populations is usually expressed implicitly, thus requiring normative reasoning and rigorous analysis. In this paper, we focus our investigation on social bias detection of dialog safety problems. We first propose a novel Dial-Bias Frame for analyzing the social bias in conversations pragmatically, which considers more comprehensive bias-related analyses rather than simple dichotomy annotations. Based on the proposed framework, we further introduce CDail-Bias Dataset that, to our knowledge, is the first well-annotated Chinese social bias dialog dataset. In addition, we establish several dialog bias detection benchmarks at different label granularities and input types (utterance-level and context-level). We show that the proposed in-depth analyses together with these benchmarks in our Dial-Bias Frame are necessary and essential to bias detection tasks and can benefit building safe dialog systems in practice.

Jiawen Deng, Jingyan Zhou, Hao Sun et al.

Offensive language detection is increasingly crucial for maintaining a civilized social media platform and deploying pre-trained language models. However, this task in Chinese is still under exploration due to the scarcity of reliable datasets. To this end, we propose a benchmark --COLD for Chinese offensive language analysis, including a Chinese Offensive Language Dataset --COLDATASET and a baseline detector --COLDETECTOR which is trained on the dataset. We show that the COLD benchmark contributes to Chinese offensive language detection which is challenging for existing resources. We then deploy the COLDETECTOR and conduct detailed analyses on popular Chinese pre-trained language models. We first analyze the offensiveness of existing generative models and show that these models inevitably expose varying degrees of offensive issues. Furthermore, we investigate the factors that influence the offensive generations, and we find that anti-bias contents and keywords referring to certain groups or revealing negative attitudes trigger offensive outputs easier.

Hao Sun, Guangxuan Xu, Jiawen Deng et al.

Dialogue safety problems severely limit the real-world deployment of neural conversational models and have attracted great research interests recently. However, dialogue safety problems remain under-defined and the corresponding dataset is scarce. We propose a taxonomy for dialogue safety specifically designed to capture unsafe behaviors in human-bot dialogue settings, with focuses on context-sensitive unsafety, which is under-explored in prior works. To spur research in this direction, we compile DiaSafety, a dataset with rich context-sensitive unsafe examples. Experiments show that existing safety guarding tools fail severely on our dataset. As a remedy, we train a dialogue safety classifier to provide a strong baseline for context-sensitive dialogue unsafety detection. With our classifier, we perform safety evaluations on popular conversational models and show that existing dialogue systems still exhibit concerning context-sensitive safety problems.