We introduce LLMStinger, a novel approach that leverages Large Language Models (LLMs) to automatically generate adversarial suffixes for jailbreak attacks. Unlike traditional methods, which require complex prompt engineering or white-box access, LLMStinger uses a reinforcement learning (RL) loop to fine-tune an attacker LLM, generating new suffixes based on existing attacks for harmful questions from the HarmBench benchmark. Our method significantly outperforms existing red-teaming approaches (we compared against 15 of the latest methods), achieving a +57.2% improvement in Attack Success Rate (ASR) on LLaMA2-7B-chat and a +50.3% ASR increase on Claude 2, both models known for their extensive safety measures. Additionally, we achieved a 94.97% ASR on GPT-3.5 and 99.4% on Gemma-2B-it, demonstrating the robustness and adaptability of LLMStinger across open and closed-source models.
Construction remains one of the most hazardous sectors. Recent advancements in AI, particularly Large Language Models (LLMs), offer promising opportunities for enhancing workplace safety. However, responsible integration of LLMs requires systematic evaluation, as deploying them without understanding their capabilities and limitations risks generating inaccurate information, fostering misplaced confidence, and compromising worker safety. This study evaluates the performance of two widely used LLMs, GPT-3.5 and GPT-4o, across three standardized exams administered by the Board of Certified Safety Professionals (BCSP). Using 385 questions spanning seven safety knowledge areas, the study analyzes the models' accuracy, consistency, and reliability. Results show that both models consistently exceed the BCSP benchmark, with GPT-4o achieving an accuracy rate of 84.6% and GPT-3.5 reaching 73.8%. Both models demonstrate strengths in safety management systems and hazard identification and control, but exhibit weaknesses in science, mathematics, emergency response, and fire prevention. An error analysis identifies four primary limitations affecting LLM performance: lack of knowledge, reasoning flaws, memory issues, and calculation errors. Our study also highlights the impact of prompt engineering strategies, with variations in accuracy reaching 13.5% for GPT-3.5 and 7.9% for GPT-4o. However, no single prompt configuration proves universally effective. This research advances knowledge in three ways: by identifying areas where LLMs can support safety practices and where human oversight remains essential, by offering practical insights into improving LLM implementation through prompt engineering, and by providing evidence-based direction for future research and development. These contributions support the responsible integration of AI in construction safety management toward achieving zero injuries.
TIPO (Text to Image with text pre-sampling for Prompt Optimization) is an innovative framework designed to enhance text-to-image (T2I) generation by language model (LM) for automatic prompt engineering. By refining and extending user-provided prompts, TIPO bridges the gap between simple inputs and the detailed prompts required for high-quality image generation. Unlike previous approaches that rely on Large Language Models (LLMs) or reinforcement learning (RL), TIPO adjusts user input prompts with the distribution of a trained prompt dataset, eliminating the need for complex runtime cost via lightweight model. This pre-sampling approach enables efficient and scalable prompt optimization, grounded in the model's training distribution. Experimental results demonstrate TIPO's effectiveness in improving aesthetic scores, reducing image corruption, and better aligning generated images with dataset distributions. These findings highlight the critical role of prompt engineering in T2I systems and open avenues for broader applications of automatic prompt refinement.
Automatic prompt engineering aims to enhance the generation quality of large language models (LLMs). Recent works utilize feedbacks generated from erroneous cases to guide the prompt optimization. During inference, they may further retrieve several semantically-related exemplars and concatenate them to the optimized prompts to improve the performance. However, those works only utilize the feedback at the current step, ignoring historical and unseleccted feedbacks which are potentially beneficial. Moreover, the selection of exemplars only considers the general semantic relationship and may not be optimal in terms of task performance and matching with the optimized prompt. In this work, we propose an Exemplar-Guided Reflection with Memory mechanism (ERM) to realize more efficient and accurate prompt optimization. Specifically, we design an exemplar-guided reflection mechanism where the feedback generation is additionally guided by the generated exemplars. We further build two kinds of memory to fully utilize the historical feedback information and support more effective exemplar retrieval. Empirical evaluations show our method surpasses previous state-of-the-arts with less optimization steps, i.e., improving F1 score by 10.1 on LIAR dataset, and reducing half of the optimization steps on ProTeGi.
Recent work finds that retrieval-augmented generation with large language models is prone to be influenced by the order of retrieved documents in the context. However, the lack of in-depth analysis limits the use of this phenomenon for prompt engineering in practice. In this study, we posit that likelihoods serve as an effective gauge for language model performance. Through experiments on two question-answering datasets with a variety of state-of-the-art language models, we reveal correlations between answer accuracy and the likelihood of the question at both the corpus level and the instance level. In addition, we find that question likelihood can also indicate the position of the task-relevant information in the context. Based on these findings, we propose two methods that use question likelihood as a gauge for selecting and constructing prompts that lead to better performance. We demonstrate their effectiveness with experiments. In addition, our likelihood-based methods are efficient, as they only need to compute the likelihood of the input, requiring much fewer language model passes than heuristic prompt engineering methods that require generating responses. Our analysis deepens our understanding of how input prompts affect model performance and provides a promising direction for efficient prompt optimization.
ChatGPT and other large language models (LLMs) promise to revolutionize software development by automatically generating code from program specifications. We assess the performance of ChatGPT's GPT-3.5-turbo model on LeetCode, a popular platform with algorithmic coding challenges for technical interview practice, across three difficulty levels: easy, medium, and hard. We test three main hypotheses. First, ChatGPT solves fewer problems as difficulty rises (Hypothesis 1). Second, prompt engineering improves ChatGPT's performance, with greater gains on easier problems and diminishing returns on harder ones (Hypothesis 2). Third, ChatGPT performs better in popular languages like Python, Java, and C++ than in less common ones like Elixir, Erlang, and Racket (Hypothesis 3). To investigate these hypotheses, we conduct automated experiments using Python scripts to generate prompts that instruct ChatGPT to create Python solutions. These solutions are stored and manually submitted on LeetCode to check their correctness. For Hypothesis 1, results show the GPT-3.5-turbo model successfully solves 92% of easy, 79% of medium, and 51% of hard problems. For Hypothesis 2, prompt engineering yields improvements: 14-29% for Chain of Thought Prompting, 38-60% by providing failed test cases in a second feedback prompt, and 33-58% by switching to GPT-4. From a random subset of problems ChatGPT solved in Python, it also solved 78% in Java, 50% in C++, and none in Elixir, Erlang, or Racket. These findings generally validate all three hypotheses.
This paper explores a new black-box, zero-shot language model inversion problem and proposes an innovative framework for prompt reconstruction using only text outputs from a language model. Leveraging a large language model alongside an optimization algorithm, the proposed method effectively recovers prompts with minimal resources. Experimental results on several datasets derived from public sources indicate that the proposed approach achieves high-quality prompt recovery and generates prompts more similar to the originals than current state-of-the-art methods. Additionally, the use-case study demonstrates the method's strong potential for generating high-quality text data.
This paper introduces an innovative approach using Retrieval-Augmented Generation (RAG) pipelines with Large Language Models (LLMs) to enhance information retrieval and query response systems for university-related question answering. By systematically extracting data from the university official webpage and employing advanced prompt engineering techniques, we generate accurate, contextually relevant responses to user queries. We developed a comprehensive university benchmark, UniversityQuestionBench (UQB), to rigorously evaluate our system performance, based on common key metrics in the filed of RAG pipelines, assessing accuracy and reliability through various metrics and real-world scenarios. Our experimental results demonstrate significant improvements in the precision and relevance of generated responses, enhancing user experience and reducing the time required to obtain relevant answers. In summary, this paper presents a novel application of RAG pipelines and LLMs, supported by a meticulously prepared university benchmark, offering valuable insights into advanced AI techniques for academic data retrieval and setting the stage for future research in this domain.
We present an innovative approach for solving mathematical problems in Bengali, developed for the DL Sprint 3.0 BUET CSE Fest 2024 Competition. Our method uses advanced deep learning models, notably the Qwen 2.5 series, with improvements made through prompt engineering, model quantization, and Tool Integrated Reasoning (TIR) to handle complex calculations. Initially, we explored various model architectures, including fine-tuned Mistral and quantized Qwen models, refining them with translation techniques, Retrieval-Augmented Generation (RAG), and custom dataset curation. Manual hyperparameter tuning optimized parameters like temperature and top-p to enhance model adaptability and accuracy. Removal of RAG and parameter adjustments further improved robustness. Our approach highlights the potential of advanced NLP techniques in solving Bengali mathematical problems.
Vision-language-action (VLA) models represent a promising direction for developing general-purpose robotic systems, demonstrating the ability to combine visual understanding, language comprehension, and action generation. However, systematic evaluation of these models across diverse robotic tasks remains limited. In this work, we present a comprehensive evaluation framework and benchmark suite for assessing VLA models. We profile three state-of-the-art VLM and VLAs - GPT-4o, OpenVLA, and JAT - across 20 diverse datasets from the Open-X-Embodiment collection, evaluating their performance on various manipulation tasks. Our analysis reveals several key insights: 1. current VLA models show significant variation in performance across different tasks and robot platforms, with GPT-4o demonstrating the most consistent performance through sophisticated prompt engineering, 2. all models struggle with complex manipulation tasks requiring multi-step planning, and 3. model performance is notably sensitive to action space characteristics and environmental factors. We release our evaluation framework and findings to facilitate systematic assessment of future VLA models and identify critical areas for improvement in the development of general purpose robotic systems.
Current metadata creation for web archives is time consuming and costly due to reliance on human effort. This paper explores the use of gpt-4o for metadata generation within the Web Archive Singapore, focusing on scalability, efficiency, and cost effectiveness. We processed 112 Web ARChive (WARC) files using data reduction techniques, achieving a notable 99.9% reduction in metadata generation costs. By prompt engineering, we generated titles and abstracts, which were evaluated both intrinsically using Levenshtein Distance and BERTScore, and extrinsically with human cataloguers using McNemar's test. Results indicate that while our method offers significant cost savings and efficiency gains, human curated metadata maintains an edge in quality. The study identifies key challenges including content inaccuracies, hallucinations, and translation issues, suggesting that Large Language Models (LLMs) should serve as complements rather than replacements for human cataloguers. Future work will focus on refining prompts, improving content filtering, and addressing privacy concerns through experimentation with smaller models. This research advances the integration of LLMs in web archiving, offering valuable insights into their current capabilities and outlining directions for future enhancements. The code is available at https://github.com/masamune-prog/warc2summary for further development and use by institutions facing similar challenges.
Political scientists often grapple with data scarcity in text classification. Recently, fine-tuned BERT models and their variants have gained traction as effective solutions to address this issue. In this study, we investigate the potential of GPT-based models combined with prompt engineering as a viable alternative. We conduct a series of experiments across various classification tasks, differing in the number of classes and complexity, to evaluate the effectiveness of BERT-based versus GPT-based models in low-data scenarios. Our findings indicate that while zero-shot and few-shot learning with GPT models provide reasonable performance and are well-suited for early-stage research exploration, they generally fall short - or, at best, match - the performance of BERT fine-tuning, particularly as the training set reaches a substantial size (e.g., 1,000 samples). We conclude by comparing these approaches in terms of performance, ease of use, and cost, providing practical guidance for researchers facing data limitations. Our results are particularly relevant for those engaged in quantitative text analysis in low-resource settings or with limited labeled data.
Enterprise searches require users to have complex knowledge of queries, configurations, and metadata, rendering it difficult for them to access information as needed. Most go-to-market (GTM) platforms utilize advanced search, an interface that enables users to filter queries by various fields using categories or keywords, which, historically, however, has proven to be exceedingly cumbersome, as users are faced with seemingly hundreds of options, fields, and buttons. Consequently, querying with natural language has long been ideal, a notion further empowered by Large Language Models (LLMs). In this paper, we implement and evaluate a solution for the Zoominfo product for sellers, which prompts the LLM with natural language, producing search fields through entity extraction that are then converted into a search query. The intermediary search fields offer numerous advantages for each query, including the elimination of syntax errors, simpler ground truths, and an intuitive format for the LLM to interpret. We paired this pipeline with many advanced prompt engineering strategies, featuring an intricate system message, few-shot prompting, chain-of-thought (CoT) reasoning, and execution refinement. Furthermore, we manually created the ground truth for 500+ natural language queries, enabling the supervised fine-tuning of Llama-3-8B-Instruct and the introduction of sophisticated numerical metrics. Comprehensive experiments with closed, open source, and fine-tuned LLM models were conducted through exact, Jaccard, cosine, and semantic similarity on individual search entities to demonstrate the efficacy of our approach. Overall, the most accurate closed model had an average accuracy of 97% per query, with only one field performing under 90%, with comparable results observed from the fine-tuned models.
We introduce "Method Actors" as a mental model for guiding LLM prompt engineering and prompt architecture. Under this mental model, LLMs should be thought of as actors; prompts as scripts and cues; and LLM responses as performances. We apply this mental model to the task of improving LLM performance at playing Connections, a New York Times word puzzle game that prior research identified as a challenging benchmark for evaluating LLM reasoning. Our experiments with GPT-4o show that a "Method Actors" approach can significantly improve LLM performance over both a vanilla and "Chain of Thoughts" approach. A vanilla approach solves 27% of Connections puzzles in our dataset and a "Chain of Thoughts" approach solves 41% of puzzles, whereas our strongest "Method Actor" approach solves 86% of puzzles. We also test OpenAI's newest model designed specifically for complex reasoning tasks, o1-preview. When asked to solve a puzzle all at once, o1-preview solves 79% of Connections puzzles in our dataset, and when allowed to build puzzle solutions one guess at a time over multiple API calls, o1-preview solves 100% of the puzzles. Incorporating a "Method Actor" prompt architecture increases the percentage of puzzles that o1-preview solves perfectly from 76% to 87%.
Run-time steering strategies like Medprompt are valuable for guiding large language models (LLMs) to top performance on challenging tasks. Medprompt demonstrates that a general LLM can be focused to deliver state-of-the-art performance on specialized domains like medicine by using a prompt to elicit a run-time strategy involving chain of thought reasoning and ensembling. OpenAI's o1-preview model represents a new paradigm, where a model is designed to do run-time reasoning before generating final responses. We seek to understand the behavior of o1-preview on a diverse set of medical challenge problem benchmarks. Following on the Medprompt study with GPT-4, we systematically evaluate the o1-preview model across various medical benchmarks. Notably, even without prompting techniques, o1-preview largely outperforms the GPT-4 series with Medprompt. We further systematically study the efficacy of classic prompt engineering strategies, as represented by Medprompt, within the new paradigm of reasoning models. We found that few-shot prompting hinders o1's performance, suggesting that in-context learning may no longer be an effective steering approach for reasoning-native models. While ensembling remains viable, it is resource-intensive and requires careful cost-performance optimization. Our cost and accuracy analysis across run-time strategies reveals a Pareto frontier, with GPT-4o representing a more affordable option and o1-preview achieving state-of-the-art performance at higher cost. Although o1-preview offers top performance, GPT-4o with steering strategies like Medprompt retains value in specific contexts. Moreover, we note that the o1-preview model has reached near-saturation on many existing medical benchmarks, underscoring the need for new, challenging benchmarks. We close with reflections on general directions for inference-time computation with LLMs.
Automated knowledge extraction from scientific literature can potentially accelerate materials discovery. We have investigated an approach for extracting synthesis protocols for reticular materials from scientific literature using large language models (LLMs). To that end, we introduce a Knowledge Extraction Pipeline (KEP) that automatizes LLM-assisted paragraph classification and information extraction. By applying prompt engineering with in-context learning (ICL) to a set of open-source LLMs, we demonstrate that LLMs can retrieve chemical information from PDF documents, without the need for fine-tuning or training and at a reduced risk of hallucination. By comparing the performance of five open-source families of LLMs in both paragraph classification and information extraction tasks, we observe excellent model performance even if only few example paragraphs are included in the ICL prompts. The results show the potential of the KEP approach for reducing human annotations and data curation efforts in automated scientific knowledge extraction.
Language Identification (LI) is crucial for various natural language processing tasks, serving as a foundational step in applications such as sentiment analysis, machine translation, and information retrieval. In multilingual societies like India, particularly among the youth engaging on social media, text often exhibits code-mixing, blending local languages with English at different linguistic levels. This phenomenon presents formidable challenges for LI systems, especially when languages intermingle within single words. Dravidian languages, prevalent in southern India, possess rich morphological structures yet suffer from under-representation in digital platforms, leading to the adoption of Roman or hybrid scripts for communication. This paper introduces a prompt based method for a shared task aimed at addressing word-level LI challenges in Dravidian languages. In this work, we leveraged GPT-3.5 Turbo to understand whether the large language models is able to correctly classify words into correct categories. Our findings show that the Kannada model consistently outperformed the Tamil model across most metrics, indicating a higher accuracy and reliability in identifying and categorizing Kannada language instances. In contrast, the Tamil model showed moderate performance, particularly needing improvement in precision and recall.
Large Language Models (LLMs) demonstrate outstanding performance in their reservoir of knowledge and understanding capabilities, but they have also been shown to be prone to illegal or unethical reactions when subjected to jailbreak attacks. To ensure their responsible deployment in critical applications, it is crucial to understand the safety capabilities and vulnerabilities of LLMs. Previous works mainly focus on jailbreak in single-round dialogue, overlooking the potential jailbreak risks in multi-round dialogues, which are a vital way humans interact with and extract information from LLMs. Some studies have increasingly concentrated on the risks associated with jailbreak in multi-round dialogues. These efforts typically involve the use of manually crafted templates or prompt engineering techniques. However, due to the inherent complexity of multi-round dialogues, their jailbreak performance is limited. To solve this problem, we propose a novel multi-round dialogue jailbreaking agent, emphasizing the importance of stealthiness in identifying and mitigating potential threats to human values posed by LLMs. We propose a risk decomposition strategy that distributes risks across multiple rounds of queries and utilizes psychological strategies to enhance attack strength. Extensive experiments show that our proposed method surpasses other attack methods and achieves state-of-the-art attack success rate. We will make the corresponding code and dataset available for future research. The code will be released soon.
Automatic feature recognition (AFR) is essential for transforming design knowledge into actionable manufacturing information. Traditional AFR methods, which rely on predefined geometric rules and large datasets, are often time-consuming and lack generalizability across various manufacturing features. To address these challenges, this study investigates vision-language models (VLMs) for automating the recognition of a wide range of manufacturing features in CAD designs without the need for extensive training datasets or predefined rules. Instead, prompt engineering techniques, such as multi-view query images, few-shot learning, sequential reasoning, and chain-of-thought, are applied to enable recognition. The approach is evaluated on a newly developed CAD dataset containing designs of varying complexity relevant to machining, additive manufacturing, sheet metal forming, molding, and casting. Five VLMs, including three closed-source models (GPT-4o, Claude-3.5-Sonnet, and Claude-3.0-Opus) and two open-source models (LLava and MiniCPM), are evaluated on this dataset with ground truth features labelled by experts. Key metrics include feature quantity accuracy, feature name matching accuracy, hallucination rate, and mean absolute error (MAE). Results show that Claude-3.5-Sonnet achieves the highest feature quantity accuracy (74%) and name-matching accuracy (75%) with the lowest MAE (3.2), while GPT-4o records the lowest hallucination rate (8%). In contrast, open-source models have higher hallucination rates (>30%) and lower accuracies (<40%). This study demonstrates the potential of VLMs to automate feature recognition in CAD designs within diverse manufacturing scenarios.
Course evaluation is a critical component in higher education pedagogy. It not only serves to identify limitations in existing course designs and provide a basis for curricular innovation, but also to offer quantitative insights for university administrative decision-making. Traditional evaluation methods, primarily comprising student surveys, instructor self-assessments, and expert reviews, often encounter challenges, including inherent subjectivity, feedback delays, inefficiencies, and limitations in addressing innovative teaching approaches. Recent advancements in large language models (LLMs) within artificial intelligence (AI) present promising new avenues for enhancing course evaluation processes. This study explores the application of LLMs in automated course evaluation from multiple perspectives and conducts rigorous experiments across 100 courses at a major university in China. The findings indicate that: (1) LLMs can be an effective tool for course evaluation; (2) their effectiveness is contingent upon appropriate fine-tuning and prompt engineering; and (3) LLM-generated evaluation results demonstrate a notable level of rationality and interpretability.