We propose a novel in-context learning algorithm for building autonomous decision-making language agents. The language agent continuously attempts to solve the same task by self-correcting each time the task fails. Our selected language agent demonstrates the ability to solve tasks in a text-based game environment. Our results show that the gemma-2-9b-it language model, using our proposed method, can successfully complete two of six tasks that failed in the first attempt. This highlights the effectiveness of our approach in enhancing the problem-solving capabilities of a single language model through self-correction, paving the way for more advanced autonomous agents. The code is publicly available at https://github.com/YenCheHsiao/AutonomousLLMAgentwithAdaptingPlanning.
Metaphor Components Identification (MCI) contributes to enhancing machine understanding of metaphors, thereby advancing downstream natural language processing tasks. However, the complexity, diversity, and dependency on context and background knowledge pose significant challenges for MCI. Large language models (LLMs) offer new avenues for accurate comprehension of complex natural language texts due to their strong semantic analysis and extensive commonsense knowledge. In this research, a new LLM-based framework is proposed, named Linguistics-aware In-context Learning with Data Augmentation (LaiDA). Specifically, ChatGPT and supervised fine-tuning are utilized to tailor a high-quality dataset. LaiDA incorporates a simile dataset for pre-training. A graph attention network encoder generates linguistically rich feature representations to retrieve similar examples. Subsequently, LLM is fine-tuned with prompts that integrate linguistically similar examples. LaiDA ranked 2nd in Subtask 2 of NLPCC2024 Shared Task 9, demonstrating its effectiveness. Code and data are available at https://github.com/WXLJZ/LaiDA.
In-context learning (ICL) greatly improves the performance of large language models (LLMs) on various down-stream tasks, where the improvement highly depends on the quality of demonstrations. In this work, we introduce syntactic knowledge to select better in-context examples for machine translation (MT). We propose a new strategy, namely Syntax-augmented COverage-based In-context example selection (SCOI), leveraging the deep syntactic structure beyond conventional word matching. Specifically, we measure the set-level syntactic coverage by computing the coverage of polynomial terms with the help of a simplified tree-to-polynomial algorithm, and lexical coverage using word overlap. Furthermore, we devise an alternate selection approach to combine both coverage measures, taking advantage of syntactic and lexical information. We conduct experiments with two multi-lingual LLMs on six translation directions. Empirical results show that our proposed SCOI obtains the highest average COMET score among all learning-free methods, indicating that combining syntactic and lexical coverage successfully helps to select better in-context examples for MT. Our code is available at https://github.com/JamyDon/SCOI.
Despite the impressive performance on information-seeking tasks, large language models (LLMs) still struggle with hallucinations. Attributed LLMs, which augment generated text with in-line citations, have shown potential in mitigating hallucinations and improving verifiability. However, current approaches suffer from suboptimal citation quality due to their reliance on in-context learning. Furthermore, the practice of citing only coarse document identifiers makes it challenging for users to perform fine-grained verification. In this work, we introduce FRONT, a training framework designed to teach LLMs to generate Fine-Grained Grounded Citations. By grounding model outputs in fine-grained supporting quotes, these quotes guide the generation of grounded and consistent responses, not only improving citation quality but also facilitating fine-grained verification. Experiments on the ALCE benchmark demonstrate the efficacy of FRONT in generating superior grounded responses and highly supportive citations. With LLaMA-2-7B, the framework significantly outperforms all the baselines, achieving an average of 14.21% improvement in citation quality across all datasets, even surpassing ChatGPT.
Building a general-purpose agent is a long-standing vision in the field of artificial intelligence. Existing agents have made remarkable progress in many domains, yet they still struggle to complete long-horizon tasks in an open world. We attribute this to the lack of necessary world knowledge and multimodal experience that can guide agents through a variety of long-horizon tasks. In this paper, we propose a Hybrid Multimodal Memory module to address the above challenges. It 1) transforms knowledge into Hierarchical Directed Knowledge Graph that allows agents to explicitly represent and learn world knowledge, and 2) summarises historical information into Abstracted Multimodal Experience Pool that provide agents with rich references for in-context learning. On top of the Hybrid Multimodal Memory module, a multimodal agent, Optimus-1, is constructed with dedicated Knowledge-guided Planner and Experience-Driven Reflector, contributing to a better planning and reflection in the face of long-horizon tasks in Minecraft. Extensive experimental results show that Optimus-1 significantly outperforms all existing agents on challenging long-horizon task benchmarks, and exhibits near human-level performance on many tasks. In addition, we introduce various Multimodal Large Language Models (MLLMs) as the backbone of Optimus-1. Experimental results show that Optimus-1 exhibits strong generalization with the help of the Hybrid Multimodal Memory module, outperforming the GPT-4V baseline on many tasks.
Log parsing is a critical step that transforms unstructured log data into structured formats, facilitating subsequent log-based analysis. Traditional syntax-based log parsers are efficient and effective, but they often experience decreased accuracy when processing logs that deviate from the predefined rules. Recently, large language models (LLM) based log parsers have shown superior parsing accuracy. However, existing LLM-based parsers face three main challenges: 1)time-consuming and labor-intensive manual labeling for fine-tuning or in-context learning, 2)increased parsing costs due to the vast volume of log data and limited context size of LLMs, and 3)privacy risks from using commercial models like ChatGPT with sensitive log information. To overcome these limitations, this paper introduces LibreLog, an unsupervised log parsing approach that leverages open-source LLMs (i.e., Llama3-8B) to enhance privacy and reduce operational costs while achieving state-of-the-art parsing accuracy. LibreLog first groups logs with similar static text but varying dynamic variables using a fixed-depth grouping tree. It then parses logs within these groups using three components: i)similarity scoring-based retrieval augmented generation: selects diverse logs within each group based on Jaccard similarity, helping the LLM distinguish between static text and dynamic variables; ii)self-reflection: iteratively query LLMs to refine log templates to improve parsing accuracy; and iii) log template memory: stores parsed templates to reduce LLM queries for improved parsing efficiency. Our evaluation on LogHub-2.0 shows that LibreLog achieves 25% higher parsing accuracy and processes logs 2.7 times faster compared to state-of-the-art LLM-based parsers. In short, LibreLog addresses privacy and cost concerns of using commercial LLMs while achieving state-of-the-arts parsing efficiency and accuracy.
Knowledge models are fundamental to dialogue systems for enabling conversational interactions, which require handling domain-specific knowledge. Ensuring effective communication in information-providing conversations entails aligning user understanding with the knowledge available to the system. However, dialogue systems often face challenges arising from semantic inconsistencies in how information is expressed in natural language compared to how it is represented within the system's internal knowledge. To address this problem, we study the potential of large language models for conversational grounding, a mechanism to bridge information gaps by establishing shared knowledge between dialogue participants. Our approach involves annotating human conversations across five knowledge domains to create a new dialogue corpus called BridgeKG. Through a series of experiments on this dataset, we empirically evaluate the capabilities of large language models in classifying grounding acts and identifying grounded information items within a knowledge graph structure. Our findings offer insights into how these models use in-context learning for conversational grounding tasks and common prediction errors, which we illustrate with examples from challenging dialogues. We discuss how the models handle knowledge graphs as a semantic layer between unstructured dialogue utterances and structured information items.
This paper aims to extend the code generation capability of large language models (LLMs) to automatically manage comprehensive software requirements from given textual descriptions. Such requirements include both functional (i.e. achieving expected behavior for inputs) and non-functional (e.g., time/space performance, robustness, maintainability) requirements. However, textual descriptions can either express requirements verbosely or may even omit some of them. We introduce ARCHCODE, a novel framework that leverages in-context learning to organize requirements observed in descriptions and to extrapolate unexpressed requirements from them. ARCHCODE generates requirements from given descriptions, conditioning them to produce code snippets and test cases. Each test case is tailored to one of the requirements, allowing for the ranking of code snippets based on the compliance of their execution results with the requirements. Public benchmarks show that ARCHCODE enhances to satisfy functional requirements, significantly improving Pass@k scores. Furthermore, we introduce HumanEval-NFR, the first evaluation of LLMs' non-functional requirements in code generation, demonstrating ARCHCODE's superiority over baseline methods. The implementation of ARCHCODE and the HumanEval-NFR benchmark are both publicly accessible.
The ability of generative large language models (LLMs) to perform in-context learning has given rise to a large body of research into how best to prompt models for various natural language processing tasks. In this paper, we focus on machine translation (MT), a task that has been shown to benefit from in-context translation examples. However no systematic studies have been published on how best to select examples, and mixed results have been reported on the usefulness of similarity-based selection over random selection. We provide a study covering multiple LLMs and multiple in-context example retrieval strategies, comparing multilingual sentence embeddings. We cover several language directions, representing different levels of language resourcedness (English into French, German, Swahili and Wolof). Contrarily to previously published results, we find that sentence embedding similarity can improve MT, especially for low-resource language directions, and discuss the balance between selection pool diversity and quality. We also highlight potential problems with the evaluation of LLM-based MT and suggest a more appropriate evaluation protocol, adapting the COMET metric to the evaluation of LLMs. Code and outputs are freely available at https://github.com/ArmelRandy/ICL-MT.
Text-to-image (T2I) generative models have revolutionized content creation by transforming textual descriptions into high-quality images. However, these models are vulnerable to jailbreaking attacks, where carefully crafted prompts bypass safety mechanisms to produce unsafe content. While researchers have developed various jailbreak attacks to expose this risk, these methods face significant limitations, including impractical access requirements, easily detectable unnatural prompts, restricted search spaces, and high query demands on the target system. In this paper, we propose JailFuzzer, a novel fuzzing framework driven by large language model (LLM) agents, designed to efficiently generate natural and semantically meaningful jailbreak prompts in a black-box setting. Specifically, JailFuzzer employs fuzz-testing principles with three components: a seed pool for initial and jailbreak prompts, a guided mutation engine for generating meaningful variations, and an oracle function to evaluate jailbreak success. Furthermore, we construct the guided mutation engine and oracle function by LLM-based agents, which further ensures efficiency and adaptability in black-box settings. Extensive experiments demonstrate that JailFuzzer has significant advantages in jailbreaking T2I models. It generates natural and semantically coherent prompts, reducing the likelihood of detection by traditional defenses. Additionally, it achieves a high success rate in jailbreak attacks with minimal query overhead, outperforming existing methods across all key metrics. This study underscores the need for stronger safety mechanisms in generative models and provides a foundation for future research on defending against sophisticated jailbreaking attacks. JailFuzzer is open-source and available at this repository: https://github.com/YingkaiD/JailFuzzer.
Text-aware recommender systems incorporate rich textual features, such as titles and descriptions, to generate item recommendations for users. The use of textual features helps mitigate cold-start problems, and thus, such recommender systems have attracted increased attention. However, we argue that the dependency on item descriptions makes the recommender system vulnerable to manipulation by adversarial sellers on e-commerce platforms. In this paper, we explore the possibility of such manipulation by proposing a new text rewriting framework to attack text-aware recommender systems. We show that the rewriting attack can be exploited by sellers to unfairly uprank their products, even though the adversarially rewritten descriptions are perceived as realistic by human evaluators. Methodologically, we investigate two different variations to carry out text rewriting attacks: (1) two-phase fine-tuning for greater attack performance, and (2) in-context learning for higher text rewriting quality. Experiments spanning 3 different datasets and 4 existing approaches demonstrate that recommender systems exhibit vulnerability against the proposed text rewriting attack. Our work adds to the existing literature around the robustness of recommender systems, while highlighting a new dimension of vulnerability in the age of large-scale automated text generation.
In-context learning (ICL) capabilities are foundational to the success of large language models (LLMs). Recently, context compression has attracted growing interest since it can largely reduce reasoning complexities and computation costs of LLMs. In this paper, we introduce a novel Query-gUIded aTtention cOmpression (QUITO) method, which leverages attention of the question over the contexts to filter useless information. Specifically, we take a trigger token to calculate the attention distribution of the context in response to the question. Based on the distribution, we propose three different filtering methods to satisfy the budget constraints of the context length. We evaluate the QUITO using two widely-used datasets, namely, NaturalQuestions and ASQA. Experimental results demonstrate that QUITO significantly outperforms established baselines across various datasets and downstream LLMs, underscoring its effectiveness. Our code is available at https://github.com/Wenshansilvia/attention_compressor.
Simple function classes have emerged as toy problems to better understand in-context-learning in transformer-based architectures used for large language models. But previously proposed simple function classes like linear regression or multi-layer-perceptrons lack the structure required to explore things like prompting and alignment within models capable of in-context-learning. We propose univariate polynomial regression as a function class that is just rich enough to study prompting and alignment, while allowing us to visualize and understand what is going on clearly.
Anomaly detection in computational workflows is critical for ensuring system reliability and security. However, traditional rule-based methods struggle to detect novel anomalies. This paper leverages large language models (LLMs) for workflow anomaly detection by exploiting their ability to learn complex data patterns. Two approaches are investigated: 1) supervised fine-tuning (SFT), where pre-trained LLMs are fine-tuned on labeled data for sentence classification to identify anomalies, and 2) in-context learning (ICL) where prompts containing task descriptions and examples guide LLMs in few-shot anomaly detection without fine-tuning. The paper evaluates the performance, efficiency, generalization of SFT models, and explores zero-shot and few-shot ICL prompts and interpretability enhancement via chain-of-thought prompting. Experiments across multiple workflow datasets demonstrate the promising potential of LLMs for effective anomaly detection in complex executions.
Large language models (LLMs) exhibit remarkable in-context learning (ICL) capabilities. However, the underlying working mechanism of ICL remains poorly understood. Recent research presents two conflicting views on ICL: One emphasizes the impact of similar examples in the demonstrations, stressing the need for label correctness and more shots. The other attributes it to LLMs' inherent ability of task recognition, deeming label correctness and shot numbers of demonstrations as not crucial. In this work, we provide a Two-Dimensional Coordinate System that unifies both views into a systematic framework. The framework explains the behavior of ICL through two orthogonal variables: whether similar examples are presented in the demonstrations (perception) and whether LLMs can recognize the task (cognition). We propose the peak inverse rank metric to detect the task recognition ability of LLMs and study LLMs' reactions to different definitions of similarity. Based on these, we conduct extensive experiments to elucidate how ICL functions across each quadrant on multiple representative classification tasks. Finally, we extend our analyses to generation tasks, showing that our coordinate system can also be used to interpret ICL for generation tasks effectively.
Game Description Language (GDL) provides a standardized way to express diverse games in a machine-readable format, enabling automated game simulation, and evaluation. While previous research has explored game description generation using search-based methods, generating GDL descriptions from natural language remains a challenging task. This paper presents a novel framework that leverages Large Language Models (LLMs) to generate grammatically accurate game descriptions from natural language. Our approach consists of two stages: first, we gradually generate a minimal grammar based on GDL specifications; second, we iteratively improve the game description through grammar-guided generation. Our framework employs a specialized parser that identifies valid subsequences and candidate symbols from LLM responses, enabling gradual refinement of the output to ensure grammatical correctness. Experimental results demonstrate that our iterative improvement approach significantly outperforms baseline methods that directly use LLM outputs. Our code is available at https://github.com/tsunehiko/ggdg
Building upon the strength of modern large language models (LLMs), generative error correction (GEC) has emerged as a promising paradigm that can elevate the performance of modern automatic speech recognition (ASR) systems. One representative approach is to leverage in-context learning to prompt LLMs so that a better hypothesis can be generated by the LLMs based on a carefully-designed prompt and an $N$-best list of hypotheses produced by ASR systems. However, it is yet unknown whether the existing prompts are the most effective ones for the task of post-ASR error correction. In this context, this paper first explores alternative prompts to identify an initial set of effective prompts, and then proposes to employ an evolutionary prompt optimization algorithm to refine the initial prompts. Evaluations results on the CHiME-4 subset of the Task $1$ of the SLT $2024$ GenSEC challenge show the effectiveness and potential of the proposed algorithms.
We introduce Lifelong ICL, a problem setting that challenges long-context language models (LMs) to learn a sequence of language tasks through in-context learning (ICL). We further introduce Task Haystack, an evaluation suite dedicated to assessing and diagnosing how long-context LMs utilizes contexts in Lifelong ICL. When given a task instruction and test inputs, long-context LMs are expected to leverage the relevant demonstrations in the Lifelong ICL prompt, avoid distraction and interference from other tasks, and achieve test accuracies that are not significantly worse than those of the Single-task ICL baseline. Task Haystack draws inspiration from the widely-adopted "needle-in-a-haystack" (NIAH) evaluation, but presents distinct new challenges. It requires models (1) to utilize the contexts at a deeper level, rather than resorting to simple copying and pasting; (2) to navigate through long streams of evolving topics and tasks, proxying the complexities and dynamism of contexts in real-world scenarios. Additionally, Task Haystack inherits the controllability of NIAH, providing model developers with tools and visualizations to identify model vulnerabilities effectively. We benchmark 14 long-context LMs using Task Haystack, finding that frontier models like GPT-4o still struggle with the setting, failing on 15% of cases on average. Most open-weight models further lack behind by a large margin, with failure rates reaching up to 61%. In our controlled analysis, we identify factors such as distraction and recency bias as contributors to these failure cases. Further, performance declines when task instructions are paraphrased at test time or when ICL demonstrations are repeated excessively, raising concerns about the robustness, instruction understanding, and true context utilization of long-context LMs.
Deploying language models (LMs) necessitates outputs to be both high-quality and compliant with safety guidelines. Although Inference-Time Guardrails (ITG) offer solutions that shift model output distributions towards compliance, we find that current methods struggle in balancing safety with helpfulness. ITG Methods that safely address non-compliant queries exhibit lower helpfulness while those that prioritize helpfulness compromise on safety. We refer to this trade-off as the guardrail tax, analogous to the alignment tax. To address this, we propose PrimeGuard, a novel ITG method that utilizes structured control flow. PrimeGuard routes requests to different self-instantiations of the LM with varying instructions, leveraging its inherent instruction-following capabilities and in-context learning. Our tuning-free approach dynamically compiles system-designer guidelines for each query. We construct and release safe-eval, a diverse red-team safety benchmark. Extensive evaluations demonstrate that PrimeGuard, without fine-tuning, overcomes the guardrail tax by (1) significantly increasing resistance to iterative jailbreak attacks and (2) achieving state-of-the-art results in safety guardrailing while (3) matching helpfulness scores of alignment-tuned models. Extensive evaluations demonstrate that PrimeGuard, without fine-tuning, outperforms all competing baselines and overcomes the guardrail tax by improving the fraction of safe responses from 61% to 97% and increasing average helpfulness scores from 4.17 to 4.29 on the largest models, while reducing attack success rate from 100% to 8%. PrimeGuard implementation is available at https://github.com/dynamofl/PrimeGuard and safe-eval dataset is available at https://huggingface.co/datasets/dynamoai/safe_eval.
Vision-Language Models (VLMs) have shown remarkable capabilities in a large number of downstream tasks. Nonetheless, compositional image understanding remains a rather difficult task due to the object bias present in training data. In this work, we investigate the reasons for such a lack of capability by performing an extensive bench-marking of compositional understanding in VLMs. We compare contrastive models with generative ones and analyze their differences in architecture, pre-training data, and training tasks and losses. Furthermore, we leverage In-Context Learning (ICL) as a way to improve the ability of VLMs to perform more complex reasoning and understanding given an image. Our extensive experiments demonstrate that our proposed approach outperforms baseline models across multiple compositional understanding datasets.