Kaijie Zhu

Yupeng Chang, Xu Wang, Jindong Wang et al. · cmu, pku

Large language models (LLMs) are gaining increasing popularity in both academia and industry, owing to their unprecedented performance in various applications. As LLMs continue to play a vital role in both research and daily use, their evaluation becomes increasingly critical, not only at the task level, but also at the society level for better understanding of their potential risks. Over the past years, significant efforts have been made to examine LLMs from various perspectives. This paper presents a comprehensive review of these evaluation methods for LLMs, focusing on three key dimensions: what to evaluate, where to evaluate, and how to evaluate. Firstly, we provide an overview from the perspective of evaluation tasks, encompassing general natural language processing tasks, reasoning, medical usage, ethics, educations, natural and social sciences, agent applications, and other areas. Secondly, we answer the `where' and `how' questions by diving into the evaluation methods and benchmarks, which serve as crucial components in assessing performance of LLMs. Then, we summarize the success and failure cases of LLMs in different tasks. Finally, we shed light on several future challenges that lie ahead in LLMs evaluation. Our aim is to offer invaluable insights to researchers in the realm of LLMs evaluation, thereby aiding the development of more proficient LLMs. Our key point is that evaluation should be treated as an essential discipline to better assist the development of LLMs. We consistently maintain the related open-source materials at: https://github.com/MLGroupJLU/LLM-eval-survey.

Kaijie Zhu, Jindong Wang, Jiaheng Zhou et al. · cmu, pku

The increasing reliance on Large Language Models (LLMs) across academia and industry necessitates a comprehensive understanding of their robustness to prompts. In response to this vital need, we introduce PromptRobust, a robustness benchmark designed to measure LLMs' resilience to adversarial prompts. This study uses a plethora of adversarial textual attacks targeting prompts across multiple levels: character, word, sentence, and semantic. The adversarial prompts, crafted to mimic plausible user errors like typos or synonyms, aim to evaluate how slight deviations can affect LLM outcomes while maintaining semantic integrity. These prompts are then employed in diverse tasks including sentiment analysis, natural language inference, reading comprehension, machine translation, and math problem-solving. Our study generates 4,788 adversarial prompts, meticulously evaluated over 8 tasks and 13 datasets. Our findings demonstrate that contemporary LLMs are not robust to adversarial prompts. Furthermore, we present a comprehensive analysis to understand the mystery behind prompt robustness and its transferability. We then offer insightful robustness analysis and pragmatic recommendations for prompt composition, beneficial to both researchers and everyday users.

Qinlin Zhao, Jindong Wang, Yixuan Zhang et al.

Large language models (LLMs) have been widely used as agents to complete different tasks, such as personal assistance or event planning. While most of the work has focused on cooperation and collaboration between agents, little work explores competition, another important mechanism that promotes the development of society and economy. In this paper, we seek to examine the competition dynamics in LLM-based agents. We first propose a general framework for studying the competition between agents. Then, we implement a practical competitive environment using GPT-4 to simulate a virtual town with two types of agents, restaurant agents and customer agents. Specifically, the restaurant agents compete with each other to attract more customers, where competition encourages them to transform, such as cultivating new operating strategies. Simulation experiments reveal several interesting findings at the micro and macro levels, which align well with existing market and sociological theories. We hope that the framework and environment can be a promising testbed to study competition that fosters understanding of society. Code is available at: https://github.com/microsoft/competeai.

Kaijie Zhu, Jindong Wang, Xixu Hu et al.

Deep neural networks are susceptible to adversarial examples, posing a significant security risk in critical applications. Adversarial Training (AT) is a well-established technique to enhance adversarial robustness, but it often comes at the cost of decreased generalization ability. This paper proposes Robustness Critical Fine-Tuning (RiFT), a novel approach to enhance generalization without compromising adversarial robustness. The core idea of RiFT is to exploit the redundant capacity for robustness by fine-tuning the adversarially trained model on its non-robust-critical module. To do so, we introduce module robust criticality (MRC), a measure that evaluates the significance of a given module to model robustness under worst-case weight perturbations. Using this measure, we identify the module with the lowest MRC value as the non-robust-critical module and fine-tune its weights to obtain fine-tuned weights. Subsequently, we linearly interpolate between the adversarially trained weights and fine-tuned weights to derive the optimal fine-tuned model weights. We demonstrate the efficacy of RiFT on ResNet18, ResNet34, and WideResNet34-10 models trained on CIFAR10, CIFAR100, and Tiny-ImageNet datasets. Our experiments show that \method can significantly improve both generalization and out-of-distribution robustness by around 1.5% while maintaining or even slightly enhancing adversarial robustness. Code is available at https://github.com/microsoft/robustlearn.

Cheng Li, Jindong Wang, Yixuan Zhang et al.

Emotional intelligence significantly impacts our daily behaviors and interactions. Although Large Language Models (LLMs) are increasingly viewed as a stride toward artificial general intelligence, exhibiting impressive performance in numerous tasks, it is still uncertain if LLMs can genuinely grasp psychological emotional stimuli. Understanding and responding to emotional cues gives humans a distinct advantage in problem-solving. In this paper, we take the first step towards exploring the ability of LLMs to understand emotional stimuli. To this end, we first conduct automatic experiments on 45 tasks using various LLMs, including Flan-T5-Large, Vicuna, Llama 2, BLOOM, ChatGPT, and GPT-4. Our tasks span deterministic and generative applications that represent comprehensive evaluation scenarios. Our automatic experiments show that LLMs have a grasp of emotional intelligence, and their performance can be improved with emotional prompts (which we call "EmotionPrompt" that combines the original prompt with emotional stimuli), e.g., 8.00% relative performance improvement in Instruction Induction and 115% in BIG-Bench. In addition to those deterministic tasks that can be automatically evaluated using existing metrics, we conducted a human study with 106 participants to assess the quality of generative tasks using both vanilla and emotional prompts. Our human study results demonstrate that EmotionPrompt significantly boosts the performance of generative tasks (10.9% average improvement in terms of performance, truthfulness, and responsibility metrics). We provide an in-depth discussion regarding why EmotionPrompt works for LLMs and the factors that may influence its performance. We posit that EmotionPrompt heralds a novel avenue for exploring interdisciplinary knowledge for human-LLMs interaction.

Kaijie Zhu, Xianjun Yang, Jindong Wang et al.

Recent research has explored that LLM agents are vulnerable to indirect prompt injection (IPI) attacks, where malicious tasks embedded in tool-retrieved information can redirect the agent to take unauthorized actions. Existing defenses against IPI have significant limitations: either require essential model training resources, lack effectiveness against sophisticated attacks, or harm the normal utilities. We present MELON (Masked re-Execution and TooL comparisON), a novel IPI defense. Our approach builds on the observation that under a successful attack, the agent's next action becomes less dependent on user tasks and more on malicious tasks. Following this, we design MELON to detect attacks by re-executing the agent's trajectory with a masked user prompt modified through a masking function. We identify an attack if the actions generated in the original and masked executions are similar. We also include three key designs to reduce the potential false positives and false negatives. Extensive evaluation on the IPI benchmark AgentDojo demonstrates that MELON outperforms SOTA defenses in both attack prevention and utility preservation. Moreover, we show that combining MELON with a SOTA prompt augmentation defense (denoted as MELON-Aug) further improves its performance. We also conduct a detailed ablation study to validate our key designs. Code is available at https://github.com/kaijiezhu11/MELON.

Wenyue Hua, Kaijie Zhu, Lingyao Li et al.

This study intends to systematically disentangle pure logic reasoning and text understanding by investigating the contrast across abstract and contextualized logical problems from a comprehensive set of domains. We explore whether LLMs demonstrate genuine reasoning capabilities across various domains when the underlying logical structure remains constant. We focus on two main questions (1) Can abstract logical problems alone accurately benchmark an LLM's reasoning ability in real-world scenarios, disentangled from contextual support in practical settings? (2) Does fine-tuning LLMs on abstract logic problem generalize to contextualized logic problems and vice versa? To investigate these questions, we focus on standard propositional logic, specifically propositional deductive and abductive logic reasoning. In particular, we construct instantiated datasets for deductive and abductive reasoning with 4 levels of difficulty, encompassing 12 distinct categories or domains based on the categorization of Wikipedia. Our experiments aim to provide insights into disentangling context in logical reasoning and the true reasoning capabilities of LLMs and their generalization potential. The code and dataset are available at: https://github.com/agiresearch/ContextHub.

Xiang Li, Wenyue Hua, Kaijie Zhu et al.

Large Vision-Language Models (LVLMs) have demonstrated impressive capabilities in multimodal understanding, yet their reasoning abilities remain underexplored. Existing benchmarks tend to focus on perception or text-based comprehension, offering limited insight into how well these models perform on structured, logic-driven tasks that require both visual and linguistic reasoning. To address this gap, we introduce NPHardEval4V, a multimodal benchmark suite grounded in four classical NP-hard problems: Knapsack, Set Cover, Traveling Salesperson, and Vertex Cover. Each task is presented through a combination of structured visual layouts and textual prompts, designed to assess the ability of LVLMs to perform combinatorial reasoning under visual-linguistic constraints. We evaluate a set of advanced open-source and closed-source vision-language models under a unified prompting and problem representation framework. This enables fair comparison across models and task types, while isolating key variables affecting performance. Our results show that while these models perform reasonably well on perception-based inputs, they struggle with global optimization, abstraction, and constraint satisfaction. No single model demonstrates consistent reasoning capability across all problem types, and common failure patterns reveal fundamental limitations in current architectures. By leveraging the structure and complexity of NP-hard problems, NPHardEval4V provides a scalable, interpretable, and challenging testbed for diagnosing reasoning behaviors in LVLMs. We hope this benchmark can support the community in building more robust, inference-capable multimodal systems. The benchmark dataset and code are available at https://github.com/lizhouf/NPHardEval4.

Cheng Li, Jindong Wang, Yixuan Zhang et al.

Emotion significantly impacts our daily behaviors and interactions. While recent generative AI models, such as large language models, have shown impressive performance in various tasks, it remains unclear whether they truly comprehend emotions. This paper aims to address this gap by incorporating psychological theories to gain a holistic understanding of emotions in generative AI models. Specifically, we propose three approaches: 1) EmotionPrompt to enhance AI model performance, 2) EmotionAttack to impair AI model performance, and 3) EmotionDecode to explain the effects of emotional stimuli, both benign and malignant. Through extensive experiments involving language and multi-modal models on semantic understanding, logical reasoning, and generation tasks, we demonstrate that both textual and visual EmotionPrompt can boost the performance of AI models while EmotionAttack can hinder it. Additionally, EmotionDecode reveals that AI models can comprehend emotional stimuli akin to the mechanism of dopamine in the human brain. Our work heralds a novel avenue for exploring psychology to enhance our understanding of generative AI models.