Graphical user interface (GUI) agents have recently emerged as an intriguing paradigm for human-computer interaction, capable of automatically executing user instructions to operate intelligent terminal devices. However, when encountering out-of-distribution (OOD) instructions that violate environmental constraints or exceed the current capabilities of agents, GUI agents may suffer task breakdowns or even pose security threats. Therefore, effective OOD detection for GUI agents is essential. Traditional OOD detection methods perform suboptimally in this domain due to the complex embedding space and evolving GUI environments. In this work, we observe that the in-distribution input semantic space of GUI agents exhibits a clustering pattern with respect to the distance from the centroid. Based on the finding, we propose GEM, a novel method based on fitting a Gaussian mixture model over input embedding distances extracted from the GUI Agent that reflect its capability boundary. Evaluated on eight datasets spanning smartphones, computers, and web browsers, our method achieves an average accuracy improvement of 23.70% over the best-performing baseline while only increasing training time by 4.9% and testing time by 6.5%. We also experimentally demonstrate that GEM can improve the step-wise success rate by 9.40% by requesting assistance from the cloud model when encountering OOD samples. Analysis verifies the generalization ability of our method through experiments on nine different backbones.

In modern software development workflows, the open-source software supply chain significantly contributes to efficient and convenient engineering practices. With increasing system complexity, it has become a common practice to use open-source software as third-party dependencies. However, due to the lack of maintenance for underlying dependencies and insufficient community auditing, ensuring the security of source code and the legitimacy of repository maintainers has become a challenge, particularly in the context of high-stealth backdoor attacks such as the XZ-Util incident. To address these problems, we propose a fine-grained project evaluation framework for backdoor risk assessment in open-source software. Our evaluation framework models highly stealthy backdoor attacks from the attacker’s perspective and defines targeted metrics for each attack stage. Moreover, to overcome the limitations of static analysis in assessing the reliability of repository maintenance activities, such as irregular committer privilege escalation and insufficient review participation, we employ large language models (LLMs) to perform semantic evaluation of code repositories while avoiding reliance on manually crafted patterns. The effectiveness of our framework is validated on 66 high-priority packages in the Debian ecosystem, and the experimental results reveal that the current open-source software supply chain is exposed to a series of security risks.

Qiushi Sun, Mukai Li, Zhoumianze Liu, Zhihui Xie, Fangzhi Xu, Zhangyue Yin, Kanzhi Cheng, Zehao Li, Zichen Ding, Qi Liu, Zhiyong Wu, Zhuosheng Zhang, Ben Kao, Lingpeng Kong. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

The rapid advancement of multimodal large language models has enabled agents to operate mobile devices by directly interacting with graphical user interfaces, opening new possibilities for mobile automation. However, real-world mobile tasks are often complex and allow for multiple valid solutions. This contradicts current mobile agent evaluation standards: offline static benchmarks can only validate a single predefined ''golden path'', while online dynamic testing is constrained by the complexity and non-reproducibility of real devices, making both approaches inadequate for comprehensively assessing agent capabilities. To bridge the gap between offline and online evaluation and enhance testing stability, this paper introduces a novel graph-structured benchmarking framework. By modeling the finite states observed during real-device interactions, it achieves static simulation of dynamic behaviors. Building on this, we develop ColorBench, a benchmark focused on complex long-horizon tasks. It supports evaluation of multiple valid solutions, subtask completion rate statistics, and atomic-level capability analysis. ColorBench contains 175 tasks (74 single-app, 101 cross-app) with an average length of over 13 steps. Each task includes at least two correct paths and several typical error paths, enabling quasi-dynamic interaction.

Recent Transformer-based language representation techniques have commonly adopted a straightforward approach to modeling textual context as a linear sequence of successive tokens. However, this sequential modeling strategy falls short in actively exploring intermediate structures present in natural languages and does not account for the rich interactive relationships between sentences. To overcome these limitations, we propose a discourse-aware framework that bridges the gap between sequential contextualization and the interactive nature of conversational reading comprehension. Concretely, we first divide the context into elementary discourse units (EDUs), ensuring that each unit contains precisely one condition. Then, we systematically explore three instantiations for modeling discourse features: sequential EDU encoding, discourse-aware masking, and discourse graph network. These techniques allow us to capture the nuanced interactions within the discourse. To assess the efficacy of our methodologies, we perform experiments on three conversational reading comprehension tasks: multi-turn response selection, conversational question answering, and conversational machine reading. Experimental results demonstrate the superiority of our proposed approach. Moreover, analysis reveals that the discourse-aware approach enables the model to effectively capture intricate relationships within the context and fosters reasoning interpretability. Additionally, our method exhibits efficacy across various backbone PLMs and diverse domains.

The remarkable capabilities of Large Language Models (LLMs) are fueled by massive internet-scale corpora. However, scraped data owners often do not consent to its use for training, raising significant legal and ethical concerns over copyright and privacy.Data auditingtechniques seek to verify whether a protected dataset was used in training a target LLM, typically framing the task as membership inference: estimating binary sample-level membership and aggregating to a dataset-level decision. In this paper, we identify a fundamental limitation of this crisp binary paradigm: in realistic training pipelines, datasets are rarely used in full. Instead, models are trained on mixtures of partial subsets drawn from multiple sources. Existing auditing techniques, built upon anall-or-noneassumption—declaring a dataset either entirely present or absent from training—collapse inpartial dataset usagescenarios. Their predictions fluctuate unpredictably with the member ratio, causing unstable performance and high false-negative rates. Inspired byfuzzy set theory, we relax the crisp notion of binary membership to a continuousfuzzy membershipin [0,1], quantifying each sample's degree of inclusion in the model's training set. We establish a theoretical bridge between sample-level fuzzy memberships and the dataset-level usage ratio, facilitating inference of the proportion of a protected dataset used during training. Aneural network fuzzifierfirst estimates sample-level fuzzy memberships from binary labels in a reference set, then refines them using dataset-level member ratios as higher-order supervision. Finally, adefuzzificationstage aggregates calibrated memberships to determine partial usage. Across LLMs of varying scales and multiple auditing datasets, ourFuzzy Auditorsubstantially outperforms state-of-the-art crisp binary techniques in detecting partial usage, estimating member proportions, and identifying individual member samples.

The malicious misuse and widespread dissemination of AI-generated images pose a significant threat to the authenticity of online information. Current detection methods often struggle to generalize to unseen generative models, and the rapid evolution of generative techniques continuously exacerbates this challenge. Without adaptability, detection models risk becoming ineffective in real-world applications. To address this critical issue, we propose a novel three-stage domain continual learning framework designed for continuous adaptation to evolving generative models. In the first stage, we employ a strategic parameter-efficient fine-tuning approach to develop a transferable offline detection model with strong generalization capabilities. Building upon this foundation, the second stage integrates unseen data streams into a continual learning process. To efficiently learn from limited samples of novel generated models and mitigate overfitting, we design a data augmentation chain with progressively increasing complexity. Furthermore, we leverage the Kronecker-Factored Approximate Curvature (K-FAC) method to approximate the Hessian and alleviate catastrophic forgetting. Finally, the third stage utilizes a linear interpolation strategy based on Linear Mode Connectivity, effectively capturing commonalities across diverse generative models and further enhancing overall performance. We establish a comprehensive benchmark of 27 generative models, including GANs, deepfakes, and diffusion models, chronologically structured up to August 2024 to simulate real-world scenarios. Extensive experiments demonstrate that our initial offline detectors surpass the leading baseline by +5.51% in terms of mean average precision. Our continual learning strategy achieves an average accuracy of 92.20%, outperforming state-of-the-art methods.

Despite being empowered with alignment mechanisms, large language models (LLMs) are increasingly vulnerable to emerging jailbreak attacks that can compromise their alignment mechanisms. This vulnerability poses significant risks to real-world applications. Existing work faces challenges in both training efficiency and generalization capabilities (i.e., Reinforcement Learning from Human Feedback and Red-Teaming). Developing effective strategies to enable LLMs to resist continuously evolving jailbreak attempts represents a significant challenge. To address this challenge, we propose a novel defensive paradigm called GuidelineLLM, which assists LLMs in recognizing queries that may have harmful content. Before LLMs respond to a query, GuidelineLLM first identifies potential risks associated with the query, summarizes these risks into guideline suggestions, and then feeds these guidelines to the responding LLMs. Importantly, our approach eliminates the necessity for additional safety fine-tuning of the LLMs themselves; only the GuidelineLLM requires fine-tuning. This characteristic enhances the general applicability of GuidelineLLM across various LLMs. Experimental results demonstrate that GuidelineLLM can significantly reduce the attack success rate (ASR) against LLM (an average reduction of 34.17% ASR) while maintaining the usefulness of LLM in handling benign queries.

Large language models (LLMs) have been widely deployed as autonomous agents capable of following user instructions and making decisions in real-world applications. Previous studies have made notable progress in benchmarking the instruction following capabilities of LLMs in general domains, with a primary focus on their inherent commonsense knowledge. Recently, LLMs have been increasingly deployed as domain-oriented agents, which rely on domain-oriented guidelines that may conflict with their commonsense knowledge. These guidelines exhibit two key characteristics: they consist of a wide range of domain-oriented rules and are subject to frequent updates. Despite these challenges, the absence of comprehensive benchmarks for evaluating the domain-oriented guideline following capabilities of LLMs presents a significant obstacle to their effective assessment and further development. In this paper, we introduce GuideBench, a comprehensive benchmark designed to evaluate guideline following performance of LLMs. GuideBench evaluates LLMs on three critical aspects: (i) adherence to diverse rules, (ii) robustness to rule updates, and (iii) alignment with human preferences. Experimental results on a range of LLMs indicate substantial opportunities for improving their ability to follow domain-oriented guidelines. Data and code are available at Anonymous.

Xinbei Ma, Yiting Wang, Yao Yao, Tongxin Yuan, Aston Zhang, Zhuosheng Zhang, Hai Zhao. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Backdoor attacks remain significant security threats to generative large language models (LLMs). Since generative LLMs output sequences of high-dimensional token logits instead of low-dimensional classification logits, most existing backdoor defense methods designed for discriminative models like BERT are ineffective for generative LLMs. Inspired by the observed differences in learning behavior between backdoor and clean mapping in the frequency space, we transform gradients of each training sample, directly influencing parameter updates, into the frequency space. Our findings reveal a distinct separation between the gradients of backdoor and clean samples in the frequency space. Based on this phenomenon, we propose Gradient Clustering in the Frequency Space for Backdoor Sample Filtering (GraCeFul), which leverages sample-wise gradients in the frequency space to effectively identify backdoor samples without requiring retraining LLMs. Experimental results show that GraCeFul outperforms baselines significantly. Notably, GraCeFul exhibits remarkable computational efficiency, achieving nearly 100% recall and F1 scores in identifying backdoor samples, reducing the average success rate of various backdoor attacks to 0% with negligible drops in clean accuracy across multiple free-style question answering datasets. Additionally, GraCeFul generalizes to Llama-2 and Vicuna. The codes are publicly available at https://github.com/ZrW00/GraceFul.

The success of modern deep learning hinges on vast training data, much of which is scraped from the web and may include copyrighted or private content—raising serious legal and ethical concerns when used without authorization. Dataset provenance seeks to identify whether a model has been trained on specific data collections, thus protecting copyright holders while preserving data utility. Existing techniques either watermark datasets to embed distinctive behaviors, or directly infer usage from discrepancies in model outputs between seen and unseen samples. These approaches exploit the fundamental problem of empirical risk minimization to overfit to seen features. Hence, provenance signals are considered inherently hard to erase, while the adversary’s perspective remains largely overlooked, limiting our ability to assess reliability in real-world scenarios. In this work, we present a unified framework that interprets both watermarking and inference-based provenance as manifestations of output divergence, modeling the interaction between auditor and adversary as a min-max game over such divergences. This perspective motivates DivMin, a simple yet effective learning strategy that minimizes the relevant divergence to suppress provenance cues. Experiments across diverse image datasets demonstrate that, starting from a pretrained vision-language model, DivMin retains over 93% of the full fine-tuning performance gain relative to a zero-shot baseline, while evading all six state-of-the-art auditing methods. Our findings establish divergence minimization as a direct and practical path to obfuscating provenance, offering a realistic simulation of potential adversary strategies to guide the development of more robust auditing techniques. Code and Appendix will be available at https://github.com/GradOpt/DivMin.

Low-rank adaptation (LoRA) has been prominently employed for parameter-efficient fine-tuning of large language models (LLMs). However, the limited expressive capacity of LoRA, stemming from the low-rank constraint, has been recognized as a bottleneck, particularly in rigorous tasks like code generation and mathematical reasoning. To address this limitation, we introduce Rank-Sharing Low-Rank Adaptation (RaSA), an innovative extension that enhances the expressive capacity of LoRA by leveraging partial rank sharing across layers. By forming a shared rank pool and applying layer-specific weighting, RaSA effectively increases the number of ranks without augmenting parameter overhead. Our theoretically grounded and empirically validated approach demonstrates that RaSA not only maintains the core advantages of LoRA but also significantly boosts performance in challenging code and math tasks. Code, data and scripts are available at: https://github.com/zwhe99/RaSA.

Chemical reasoning usually involves complex, multi-step processes that demand precise calculations, where even minor errors can lead to cascading failures. Furthermore, large language models (LLMs) encounter difficulties handling domain-specific formulas, executing reasoning steps accurately, and integrating code ef- effectively when tackling chemical reasoning tasks. To address these challenges, we present ChemAgent, a novel framework designed to improve the performance of LLMs through a dynamic, self-updating library. This library is developed by decomposing chemical tasks into sub-tasks and compiling these sub-tasks into a structured collection that can be referenced for future queries. Then, when presented with a new problem, ChemAgent retrieves and refines pertinent information from the library, which we call memory, facilitating effective task decomposition and the generation of solutions. Our method designs three types of memory and a library-enhanced reasoning component, enabling LLMs to improve over time through experience. Experimental results on four chemical reasoning datasets from SciBench demonstrate that ChemAgent achieves performance gains of up to 46% (GPT-4), significantly outperforming existing methods. Our findings suggest substantial potential for future applications, including tasks such as drug discovery and materials science. Our code can be found at https://github.com/gersteinlab/ChemAgent.

The remarkable performance of long reasoning models can be attributed to their ability to emulate human-like long-time thinking during inference. These models employ extended chain-of-thought (CoT) processes, exploring multiple strategies to enhance problem-solving capabilities. However, a critical question remains: How to intelligently and efficiently scale computational resources during testing. This paper presents the first comprehensive study on the prevalent issue of overthinking in these models, where long reasoning models generate redundant solutions that contribute minimally to accuracy and diversity, thereby wasting computational resources on simple problems with minimal benefit. We introduce novel efficiency metrics from both outcome and process perspectives to evaluate the rational use of computational resources by long reasoning models. Using a self-training paradigm, we propose strategies to mitigate overthinking, simplifying reasoning processes without compromising accuracy. Experimental results show that our approach successfully reduces computational overhead while preserving model performance across a range of testsets with varying difficulty levels, such as GSM8K, MATH500, GPQA, and AIME. Our code is open-source and available at https://github.com/galaxyChen/overthinking.

Multi-Modal Large Language Models (MLLMs) have exhibited remarkable performance on various vision-language tasks such as Visual Question Answering (VQA). Despite accumulating evidence of privacy concerns associated with task-relevant content, it remains unclear whether MLLMs inadvertently memorize private content that is entirely irrelevant to the training tasks. In this paper, we investigate how randomly generated task-irrelevant private content can become spuriously correlated with downstream objectives due to partial mini-batch training dynamics, thus causing inadvertent memorization. Concretely, we randomly generate task-irrelevant watermarks into VQA fine-tuning images at varying probabilities and propose a novel probing framework to determine whether MLLMs have inadvertently encoded such content. Our experiments reveal that MLLMs exhibit notably different training behaviors in partial mini-batch settings with task-irrelevant watermarks embedded. Furthermore, through layer-wise probing, we demonstrate that MLLMs trigger distinct representational patterns when encountering previously seen task-irrelevant knowledge, even if this knowledge does not influence their output during prompting. Our code is available at https://github.com/illusionhi/ProbingPrivacy.

With the surge of large language models (LLMs), Large Vision-Language Models (VLMs)-which integrate vision encoders with LLMs for accurate visual grounding-have shown great potential in tasks like generalist agents and robotic control. However, VLMs are typically trained on massive web-scraped images, raising concerns over copyright infringement and privacy violations, and making data auditing increasingly urgent. Membership inference (MI), which determines whether a sample was used in training, has emerged as a key auditing technique, with promising results on open-source VLMs like LLaVA (AUC > 80%). In this work, we revisit these advances and uncover a critical issue: current MI benchmarks suffer from distribution shifts between member and non-member images, introducing shortcut cues that inflate MI performance. We further analyze the nature of these shifts and propose a principled metric based on optimal transport to quantify the distribution discrepancy. To evaluate MI in realistic settings, we construct new benchmarks with i.i.d. member and non-member images. Existing MI methods fail under these unbiased conditions, performing only marginally better than chance. Further, we explore the theoretical upper bound of MI by probing the Bayes Optimality within the VLM's embedding space and find the irreducible error rate remains high. Despite this pessimistic outlook, we analyze why MI for VLMs is particularly challenging and identify three practical scenarios-fine-tuning, access to ground-truth texts, and set-based inference-where auditing becomes feasible. Our study presents a systematic view of the limits and opportunities of MI for VLMs, providing guidance for future efforts in trustworthy data auditing. Code and data will be available at https://github.com/GradOpt/Revisiting-VLM-MIA\faGithub.

Long reasoning models (LRMs) such as OpenAI's o1 and DeepSeek's R1 have demonstrated remarkable abilities in complex reasoning tasks by scaling test-time compute and exhibiting human-like deep thinking. However, we identify a phenomenon we term underthinking, where LRMs frequently switch between different reasoning thoughts without sufficiently exploring promising paths to reach a correct solution. This behavior leads to inadequate depth of reasoning and decreased performance, particularly on challenging mathematical problems. To systematically analyze this issue, we conduct experiments on three challenging test sets and two representative open-source LRMs, revealing that frequent thought switching correlates with incorrect responses. We introduce a novel metric to quantify underthinking by measuring token efficiency in incorrect answers. To address underthinking, we propose a decoding strategy with thought switching penalty (Tip) that discourages premature transitions between thoughts, encouraging deeper exploration of each reasoning path. Experimental results demonstrate that our approach improves accuracy across challenging datasets without requiring model fine-tuning. Our findings contribute to understanding reasoning inefficiencies in LRMs and offer a practical solution to enhance their problem-solving capabilities. Our code is open-source and available at https://github.com/wangyuenlp/underthinking.

Test-time scaling enhances large language model performance by allocating additional compute resources during decoding. Best-of-$N$ (BoN) sampling serves as a common sampling-based scaling technique, broadening the search space in parallel to find better solutions from the model distribution. However, its cost–performance trade-off is still underexplored. Two main challenges limit the efficiency of BoN sampling: (1) Generating $N$ full samples consumes substantial GPU memory, reducing inference capacity under limited resources. (2) Reward models add extra memory and latency overhead, and training strong reward models introduces potential training data costs. Although some studies have explored efficiency improvements, none have addressed both challenges at once. To address this gap, we propose **Self-Truncation Best-of-$N$ (ST-BoN)**, a decoding method that avoids fully generating all $N$ samples and eliminates the need for reward models. It leverages early sampling consistency in the model’s internal states to identify the most promising path and truncate suboptimal ones. In terms of cost, ST-BoN reduces dynamic GPU memory usage by over 80% and inference latency by 50%. In terms of cost–performance trade-off, ST-BoN achieves the same performance as Full-BoN while saving computational cost by 70%–80%, and under the same cost, it can improve accuracy by 3–4 points.

Recent years have witnessed an increasing interest in training machines with reasoning ability, which deeply relies on accurately and clearly presented clue forms. The clues are usually modeled as entity-aware knowledge in existing studies. However, those entity-aware clues are primarily focused on commonsense, making them insufficient for tasks that require knowledge of temporary facts or events, particularly in logical reasoning for reading comprehension. To address this challenge, we are motivated to cover both commonsense and temporary knowledge clues hierarchically. Specifically, we propose a general formalism of knowledge units by extracting backbone constituents of the sentence, such as the subject-verb-object formed ``facts''. We then construct a supergraph on top of the fact units, allowing for the benefit of sentence-level (relations among fact groups) and entity-level interactions (concepts or actions inside a fact). Experimental results on logical reasoning benchmarks and dialogue modeling datasets show that our approach improves the baselines substantially, and it is general across backbone models. Code is available at https://github.com/ozyyshr/FocalReasoner.

Zhiwei He, Binglin Zhou, Hongkun Hao, Aiwei Liu, Xing Wang, Zhaopeng Tu, Zhuosheng Zhang, Rui Wang. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Tianjie Ju, Yijin Chen, Xinwei Yuan, Zhuosheng Zhang, Wei Du, Yubin Zheng, Gongshen Liu. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Despite the notable success of language models (LMs) in various natural language processing (NLP) tasks, the reliability of LMs is susceptible to backdoor attacks.Prior research attempts to mitigate backdoor learning while training the LMs on the poisoned dataset, yet struggles against complex backdoor attacks in real-world scenarios.In this paper, we investigate the learning mechanisms of backdoor LMs in the frequency space by Fourier analysis.Our findings indicate that the backdoor mapping presented on the poisoned datasets exhibits a more discernible inclination towards lower frequency compared to clean mapping, resulting in the faster convergence of backdoor mapping.To alleviate this dilemma, we propose Multi-Scale Low-Rank Adaptation (MuScleLoRA), which deploys multiple radial scalings in the frequency space with lowrank adaptation to the target model and further aligns the gradients when updating parameters.Through downscaling in the frequency space, MuScleLoRA encourages the model to prioritize the learning of relatively highfrequency clean mapping, consequently mitigating backdoor learning.Experimental results demonstrate that MuScleLoRA outperforms baselines significantly.Notably, MuScleLoRA reduces the average success rate of diverse backdoor attacks to below 15% across multiple datasets and generalizes to various backbone LMs, including BERT, RoBERTa, GPT2-XL, and Llama2.The codes are publicly available at https://github.com/ZrW00/MuScleLoRA.

Large language models have manifested remarkable capabilities by leveraging chain-of-thought (CoT) reasoning techniques to solve intricate questions through step-by-step reasoning chains. Despite its success, the efficacy of such reasoning is inherently contingent upon the quality of CoT. However, flawless CoT reasoning cannot be guaranteed due to the presence of indecomposable questions and the potential for erroneous reasoning chains, particularly in the case of small-scale language models. To tackle this challenge, we propose a novel approach called the selective filtering reasoner (SelF-Reasoner) that assesses the entailment relationship between the question and the candidate reasoning chain. We proceed with CoT reasoning when the reasoning chain demonstrates confidence; otherwise, we opt to predict the answer directly. SelF-Reasoner improves the fine-tuned T5 baseline consistently over the ScienceQA, ECQA, and LastLetter tasks. Code is available at Anonymous.

Large language models (LLMs) empowered by chain-of-thought (CoT) prompting have yielded remarkable prowess in reasoning tasks. Nevertheless, current methods predominantly lean on handcrafted or task-specific demonstrations, lack reliable knowledge basis and thus struggle for trustworthy responses in an automated pattern. While recent works endeavor to improve upon one certain aspect, they ignore the importance and necessity of establishing an integrated and interpretable reasoning system. To address these drawbacks and provide a universal solution, we propose AuRoRA: a one-for-all platform for augmented reasoning and refining based on CoT prompting that excels in adaptability, reliability, integrity, and interpretability. The system exhibits superior performances across six reasoning tasks and offers real-time visual analysis, which has pivotal academic and application value in the era of LLMs. The AuRoRA platform is available at https://huggingface.co/spaces/Anni123/AuRoRA.

Large language models (LLMs) have played a pivotal role in building communicative AI, yet they encounter the challenge of efficient updates.Model editing enables the manipulation of specific knowledge memories and the behavior of language generation without retraining.However, the robustness of model editing remains an open question.This work seeks to understand the strengths and limitations of editing methods, facilitating practical applications of communicative AI.We focus on three key research questions.RQ1: Can edited LLMs behave consistently resembling communicative AI in realistic situations?RQ2: To what extent does the rephrasing of prompts lead LLMs to deviate from the edited knowledge memory?RQ3: Which knowledge features are correlated with the performance and robustness of editing?Our empirical studies uncover a substantial disparity between existing editing methods and the practical application of LLMs.On rephrased prompts that are flexible but common in realistic applications, the performance of editing experiences a significant decline.Further analysis shows that more popular knowledge is memorized better, easier to recall, and more challenging to edit effectively.

Despite the rapid progress of large language models (LLMs), their task performance remains sensitive to prompt design.Recent studies have explored leveraging the LLM itself as an optimizer to identify optimal prompts that maximize task accuracy.However, when evaluating prompts, such approaches heavily rely on elusive manually annotated gold labels to calculate task accuracy for each candidate prompt, which hinders its generality.To overcome the limitation, this work proposes GLaPE, a gold label-agnostic prompt evaluation method to alleviate dependence on gold labels.GLaPE is composed of two critical aspects: selfconsistency evaluation of a single prompt and mutual-consistency refinement across multiple prompts.Experimental results on 8 widely-recognized reasoning tasks demonstrate that GLaPE can produce more effective prompts, achieving performance comparable to those derived from manually annotated gold labels.Analysis shows that GLaPE provides reliable evaluations aligned with accuracy, even in the absence of gold labels.Code is publicly available at https://github. com/thunderous77/GLaPE. (a) Accuracy EvaluationQuestion: Oscar has 24 lollipops and eats 2 on his way to school.He passes 14 out to his friends.He buys twice as many lollipops on his way home as he gave to his friends.He eats 3 more that night and 2 more in the morning.How many lollipops does Oscar have?Gold Label (Answer): 31 Prompt1: By carefully analyzing all aspects of the situation, the optimal solution becomes crystal clear.

Large Language Models (LLMs) excel in diverse areas, yet struggle with complex scientific reasoning, especially in the field of chemistry. Different from the simple chemistry tasks (e.g., molecule classification) addressed in previous studies, complex chemistry problems require not only vast knowledge and precise calculation, but also compositional reasoning about rich dynamic interactions of different concepts (e.g., temperature changes). Our study shows that even advanced LLMs, like GPT-4, can fail easily in different ways. Interestingly, the errors often stem not from a lack of domain knowledge within the LLMs, but rather from the absence of an effective reasoning *structure* that guides the LLMs to elicit the right knowledge, incorporate the knowledge in step-by-step reasoning, and iteratively refine results for further improved quality. On this basis, we introduce StructChem, a simple yet effective prompting strategy that offers the desired guidance and substantially boosts the LLMs' chemical reasoning capability. Testing across four chemistry areas---quantum chemistry, mechanics, physical chemistry, and kinetics---StructChem substantially enhances GPT-4's performance, with up to 30% peak improvement. Our analysis also underscores the unique difficulties of precise grounded reasoning in science with LLMs, highlighting a need for more research in this area.

Artificial intelligence (AI) encompasses knowledge acquisition and real-world grounding across various modalities, including language, visual, auditory, and sensory data. Multimodal large language models (MLLMs) have thus recently garnered growing interest in both academia and industry, showing an unprecedented trend to achieve human-level AI. This tutorial aims to deliver a comprehensive review of cutting-edge research in MLLMs, focusing on three key areas: MLLM architecture design, instructional learning, and multimodal reasoning of MLLMs. We will explore technical advancements, synthesize key challenges, and discuss potential avenues for future research. All the resources and materials will be made available online. https://mllm2024.github.io/ACM-MM2024

Zhiwei He, Xing Wang, Wenxiang Jiao, Zhuosheng Zhang, Rui Wang, Shuming Shi, Zhaopeng Tu. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

Junlong Li, Jinyuan Wang, Zhuosheng Zhang, Hai Zhao. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

Real-world data deviating from the independent and identically distributed (\textit{i.i.d.}) assumption of in-distribution training data poses security threats to deep networks, thus advancing out-of-distribution (OOD) detection algorithms. Detection methods in generative language models (GLMs) mainly focus on uncertainty estimation and embedding distance measurement, with the latter proven to be most effective in traditional linguistic tasks like summarization and translation. However, another complex generative scenario mathematical reasoning poses significant challenges to embedding-based methods due to its high-density feature of output spaces, but this feature causes larger discrepancies in the embedding shift trajectory between different samples in latent spaces. Hence, we propose a trajectory-based method TV score, which uses trajectory volatility for OOD detection in mathematical reasoning. Experiments show that our method outperforms all traditional algorithms on GLMs under mathematical reasoning scenarios and can be extended to more applications with high-density features in output spaces, such as multiple-choice questions.

Abstract Large language models (LLMs) have demonstrated impressive capabilities in general scenarios, exhibiting a level of aptitude that approaches, in some aspects even surpasses, human-level intelligence. Among their numerous skills, the translation abilities of LLMs have received considerable attention. Compared to typical machine translation that focuses solely on source-to-target mapping, LLM-based translation can potentially mimic the human translation process, which might take preparatory steps to ensure high-quality translation. This work explores this possibility by proposing the MAPS framework, which stands for Multi-Aspect Prompting and Selection. Specifically, we enable LLMs first to analyze the given source sentence and induce three aspects of translation-related knowledge (keywords, topics, and relevant demonstrations) to guide the final translation process. Moreover, we employ a selection mechanism based on quality estimation to filter out noisy and unhelpful knowledge. Both automatic (3 LLMs × 11 directions × 2 automatic metrics) and human evaluation (preference study and MQM) demonstrate the effectiveness of MAPS. Further analysis shows that by mimicking the human translation process, MAPS reduces various translation errors such as hallucination, ambiguity, mistranslation, awkward style, untranslated text, and omission. Source code is available at https://github.com/zwhe99/MAPS-mt.

Training machines to understand natural language and interact with humans is one of the major goals of artificial intelligence. Recent years have witnessed an evolution from matching networks to pretrained language models (PrLMs). In contrast to the plain-text modeling as the focus of the PrLMs, dialog texts involve multiple speakers and reflect special characteristics, such as topic transitions and structure dependencies, between distant utterances. However, the related PrLM models commonly represent dialogs sequentially by processing the pairwise dialog history as a whole. Thus, the hierarchical information on either utterance interrelation or speaker roles coupled in such representations is not well addressed. In this work, we propose compositional learning for holistic interaction across the utterances beyond the sequential contextualization from PrLMs, in order to capture the utterance-aware and speaker-aware representations entailed in a dialog history. We decouple the contextualized word representations by masking mechanisms in transformer-based PrLM, making each word only focus on the words in the current utterance, other utterances, and two speaker roles (i.e., utterances of the sender and utterances of the receiver), respectively. In addition, we employ domain-adaptive training strategies to help the model adapt to the dialog domains. Experimental results show that our method substantially boosts the strong PrLM baselines in four public benchmark datasets, achieving new state-of-the-art performance over previous methods.

Discriminative pre-trained language models (PrLMs) learn to predict original texts from intentionally corrupted ones. Taking the former text as positive and the latter as negative samples, the PrLM can be trained effectively for contextualized representation. However, the training of such a type of PrLMs highly relies on the quality of the automatically constructed samples. Existing PrLMs simply treat all corrupted texts as equal negative without any examination, which actually lets the resulting model inevitably suffer from the false negative issue where training is carried out on pseudo-negative data and leads to less efficiency and less robustness in the resulting PrLMs. In this work, on the basis of defining the false negative issue in discriminative PrLMs that has been ignored for a long time, we design enhanced pre-training methods to counteract false negative predictions and encourage pre-training language models on true negatives by correcting the harmful gradient updates subject to false negative predictions. Experimental results on GLUE and SQuAD benchmarks show that our counter-false-negative pre-training methods indeed bring about better performance together with stronger robustness.

Automatic summarization generates concise summaries that contain key ideas of source documents.As the most mainstream datasets for the news sub-domain, CNN/DailyMail and BBC XSum have been widely used for performance benchmarking.However, the reference summaries of those datasets turn out to be noisy, mainly in terms of factual hallucination and information redundancy.To address this challenge, we first annotate new expertwriting Element-aware test sets following the "Lasswell Communication Model" proposed by Lasswell (1948), allowing reference summaries to focus on more fine-grained news elements objectively and comprehensively.Utilizing the new test sets, we observe the surprising zero-shot summary ability of LLMs, which addresses the issue of the inconsistent results between human preference and automatic evaluation metrics of LLMs' zero-shot summaries in prior work.Further, we propose a Summary Chain-of-Thought (SumCoT) technique to elicit LLMs to generate summaries step by step, which helps them integrate more finegrained details of source documents into the final summaries that correlate with the human writing mindset.Experimental results show our method outperforms state-of-the-art fine-tuned PLMs and zero-shot LLMs by +4.33/+4.77 in ROUGE-L on the two datasets, respectively.Dataset and code are publicly available at https://github.com/Alsace08/SumCoT.

Masked Language Modeling (MLM) has been widely used as the denoising objective in pretraining language models (PrLMs).Existing PrLMs commonly adopt a Random-Token Masking strategy where a fixed masking ratio is applied and different contents are masked by an equal probability throughout the entire training.However, the model may receive a complicated impact from pre-training status, which changes accordingly as training time goes on.In this paper, we show that such time-invariant MLM settings on masking ratio and masked content are unlikely to deliver an optimal outcome, which motivates us to explore the influence of time-variant MLM settings.We propose two scheduled masking approaches that adaptively tune the masking ratio and masked content in different training stages, which improves the pre-training efficiency and effectiveness verified on the downstream tasks.Our work is a pioneer study on time-variant masking strategy on ratio and content and gives a better understanding of how masking ratio and masked content influence the MLM pretraining 1 .

Spurred by advancements in scale, large language models (LLMs) have demonstrated the ability to perform a variety of natural language processing (NLP) tasks zero-shot-i.e., without adaptation on downstream data.Recently, the debut of ChatGPT 1 has drawn a great deal of attention from the natural language processing (NLP) community due to the fact that it can generate high-quality responses to human input and self-correct previous mistakes based on subsequent conversations.However, it is not yet known whether ChatGPT can serve as a generalist model that can perform many NLP tasks zero-shot.In this work, we empirically analyze the zero-shot learning ability of ChatGPT by evaluating it on 20 popular NLP datasets covering 7 representative task categories.With extensive empirical studies, we demonstrate both the effectiveness and limitations of the current version of ChatGPT.We find that ChatGPT performs well on many tasks favoring reasoning capabilities (e.g., arithmetic reasoning) while it still faces challenges when solving specific tasks such as sequence tagging.We additionally provide in-depth analysis through qualitative case studies.

Representation learning is the foundation of natural language processing (NLP). This work presents new methods to employ visual information as assistant signals to general NLP tasks. For each sentence, we first retrieve a flexible number of images either from a light topic-image lookup table extracted over the existing sentence-image pairs or a shared cross-modal embedding space that is pre-trained on out-of-shelf text-image pairs. Then, the text and images are encoded by a Transformer encoder and convolutional neural network, respectively. The two sequences of representations are further fused by an attention layer for the interaction of the two modalities. In this study, the retrieval process is controllable and flexible. The universal visual representation overcomes the lack of large-scale bilingual sentence-image pairs. Our method can be easily applied to text-only tasks without manually annotated multimodal parallel corpora. We apply the proposed method to a wide range of natural language generation and understanding tasks, including neural machine translation, natural language inference, and semantic similarity. Experimental results show that our method is generally effective for different tasks and languages. Analysis indicates that the visual signals enrich textual representations of content words, provide fine-grained grounding information about the relationship between concepts and events, and potentially conduce to disambiguation.

Multi-party multi-turn dialogue comprehension brings unprecedented challenges in handling complicated scenarios, as the co-occurrence of multiple speakers causes complexity and inconsistency. As a result of the multiple participation, the shift of speaker roles and crisscrossed discourse relations among utterances hinder reading comprehension. Motivated by this, we further integrate the enhancements of speaker-related features for dialogue comprehension performance. This work proposes a novel model with enhancement from both sides of speaker roles and speaker-aware relations. At the token level, we apply a speaker mask for attention, while at the discourse level, we utilize heterogeneous graph networks for comprehensive speaker-aware discourse clues. Experimental results show that ourEnhancedSpeaker-Aware method (ESA) helps achieve state-of-the-art performance on the Molweni dataset, as well as significant improvements on the FriendsQA dataset. We find that our method makes steady improvements on stronger backbones. Analysis shows that our model enhances the connections between utterances and their own speakers and captures the speaker-aware discourse relations. Discussions on data features and error cases are presented, and a visualized case is displayed. The findings reveal the importance of speaker-aware signals in dialogue comprehension.

Machine reading comprehension is a heavilystudied research and test field for evaluating new pre-trained language models (PrLMs) and fine-tuning strategies, and recent studies have enriched the pre-trained language models with syntactic, semantic and other linguistic information to improve the performance of the models.In this paper, we imitate the human reading process in connecting the anaphoric expressions and explicitly leverage the coreference information of the entities to enhance the word embeddings from the pretrained language model, in order to highlight the coreference mentions of the entities that must be identified for coreference-intensive question answering in QUOREF, a relatively new dataset that is specifically designed to evaluate the coreference-related performance of a model.We use two strategies to finetune a pre-trained language model, namely, placing an additional encoder layer after a pre-trained language model to focus on the coreference mentions or constructing a relational graph convolutional network to model the coreference relations.We demonstrate that the explicit incorporation of coreference information in the fine-tuning stage performs better than the incorporation of the coreference information in pre-training a language model.

Training dense passage representations via contrastive learning has been shown effective for Open-Domain Passage Retrieval (ODPR).Existing studies focus on further optimizing by improving negative sampling strategy or extra pretraining.However, these studies keep unknown in capturing passage with internal representation conflicts from improper modeling granularity.Specifically, under our observation that a passage can be organized by multiple semantically different sentences, modeling such a passage as a unified dense vector is not optimal.This work thus presents a refined model on the basis of a smaller granularity, contextual sentences, to alleviate the concerned conflicts.In detail, we introduce an in-passage negative sampling strategy to encourage a diverse generation of sentence representations within the same passage.Experiments on three benchmark datasets verify the efficacy of our method, especially on datasets where conflicts are severe.Extensive experiments further present good transferability of our method across datasets.

Tangled multi-party dialogue contexts lead to challenges for dialogue reading comprehension, where multiple dialogue threads flow simultaneously within a common dialogue record, increasing difficulties in understanding the dialogue history for both human and machine.Previous studies mainly focus on utterance encoding methods with carefully designed features but pay inadequate attention to characteristic features of the structure of dialogues.We specially take structure factors into account and design a novel model for dialogue disentangling.Based on the fact that dialogues are constructed on successive participation and interactions between speakers, we model structural information of dialogues in two aspects: 1)speaker property that indicates whom a message is from, and 2) reference dependency that shows whom a message may refer to.The proposed method achieves new state-of-the-art on the Ubuntu IRC benchmark dataset and contributes to dialogue-related comprehension.

Machine reading comprehension (MRC) poses new challenges to logical reasoning, which aims to understand the implicit logical relations entailed in the given contexts and perform inference over them. Due to the complexity of logic, logical connections exist at different granularity levels. However, most existing methods of logical reasoning individually focus on either entity-aware or discourse-based information but ignore the hierarchical relations that may even have mutual effects. This paper proposes a holistic graph network (HGN) that deals with context at both discourse-level and word-level as the basis for logical reasoning to provide a more fine-grained relation extraction. Specifically, node-level and type-level relations, which can be interpreted as bridges in the reasoning process, are modeled by a hierarchical interaction mechanism to improve the interpretation of MRC systems. Experimental results on logical reasoning QA datasets (ReClor and LogiQA) and natural language inference datasets (SNLI and ANLI) show the effectiveness and generalization of our method, and in-depth analysis verifies its capability to understand complex logical relations.

Discriminative pre-trained language models (PrLMs) can be generalized as denoising auto-encoders that work with two procedures, ennoising and denoising.First, an ennoising process corrupts texts with arbitrary noising functions to construct training instances.Then, a denoising language model is trained to restore the corrupted tokens.Existing studies have made progress by optimizing independent strategies of either ennoising or denosing.They treat training instances equally throughout the training process, with little attention on the individual contribution of those instances.To model explicit signals of instance contribution, this work proposes to estimate the complexity of restoring the original sentences from corrupted ones in language model pretraining.The estimations involve the corruption degree in the ennoising data construction process and the prediction confidence in the denoising counterpart.Experimental results on natural language understanding and reading comprehension benchmarks show that our approach improves pre-training efficiency, effectiveness, and robustness.

Multi-turn dialogue modeling as a challenging branch of natural language understanding (NLU), aims to build representations for machines to understand human dialogues, which provides a solid foundation for multiple downstream tasks.Recent studies of dialogue modeling commonly employ pre-trained language models (PrLMs) to encode the dialogue history as successive tokens, which is insufficient in capturing the temporal characteristics of dialogues.Therefore, we propose Bidirectional Information Decoupling Network (BiDeN) as a universal dialogue encoder, which explicitly incorporates both the past and future contexts and can be generalized to a wide range of dialogue-related tasks.Experimental results on datasets of different downstream tasks demonstrate the universality and effectiveness of our BiDeN.The official implementation of BiDeN is available at https://github.com/ EricLee8/BiDeN.

A common thread of retrieval-augmented methods in the existing literature focuses on retrieving encyclopedic knowledge, such as Wikipedia, which facilitates well-defined entity and relation spaces that can be modeled.However, applying such methods to commonsense reasoning tasks faces two unique challenges, i.e., the lack of a general large-scale corpus for retrieval and a corresponding effective commonsense retriever.In this paper, we systematically investigate how to leverage commonsense knowledge retrieval to improve commonsense reasoning tasks.We proposed a unified framework of Retrieval-Augmented Commonsense reasoning (called RACO), including a newly constructed commonsense corpus with over 20 million documents and novel strategies for training a commonsense retriever.We conducted experiments on four different commonsense reasoning tasks.Extensive evaluation results showed that our proposed RACO can significantly outperform other knowledgeenhanced method counterparts, achieving new SoTA performance on the CommonGen 1 and CREAK 2 leaderboards.Our code is available at https://github.com/wyu97/RACo.

Text encoding is one of the most important steps in Natural Language Processing (NLP). It has been done well by the self-attention mechanism in the current state-of-the-art Transformer encoder, which has brought about significant improvements in the performance of many NLP tasks. Though the Transformer encoder may effectively capture general information in its resulting representations, the backbone information, meaning the gist of the input text, is not specifically focused on. In this paper, we propose explicit and implicit text compression approaches to enhance the Transformer encoding and evaluate models using this approach on several typical downstream tasks that rely on the encoding heavily. Our explicit text compression approaches use dedicated models to compress text, while our implicit text compression approach simply adds an additional module to the main model to handle text compression. We propose three ways of integration, namely backbone source-side fusion, target-side fusion, and both-side fusion, to integrate the backbone information into Transformer-based models for various downstream tasks. Our evaluation on benchmark datasets shows that the proposed explicit and implicit text compression approaches improve results in comparison to strong baselines. We therefore conclude, when comparing the encodings to the baseline models, text compression helps the encoders to learn better language representations.

Understanding human language is one of the key themes of artificial intelligence. For language representation, the capacity of effectively modeling the linguistic knowledge from the detail-riddled and lengthy texts and getting ride of the noises is essential to improve its performance. Traditional attentive models attend to all words without explicit constraint, which results in inaccurate concentration on some dispensable words. In this work, we propose using syntax to guide the text modeling by incorporating explicit syntactic constraints into attention mechanisms for better linguistically motivated word representations. In detail, for self-attention network (SAN) sponsored Transformer-based encoder, we introduce syntactic dependency of interest (SDOI) design into the SAN to form an SDOI-SAN with syntax-guided self-attention. Syntax-guided network (SG-Net) is then composed of this extra SDOI-SAN and the SAN from the original Transformer encoder through a dual contextual architecture for better linguistics inspired representation. The proposed SG-Net is applied to typical Transformer encoders. Extensive experiments on popular benchmark tasks, including machine reading comprehension, natural language inference, and neural machine translation show the effectiveness of the proposed SG-Net design.

Although pre-trained language models (PrLMs) have achieved significant success, recent studies demonstrate that PrLMs are vulnerable to adversarial attacks. By generating adversarial examples with slight perturbations on different levels (sentence / word / character), adversarial attacks can fool PrLMs to generate incorrect predictions, which questions the robustness of PrLMs. However, we find that most existing textual adversarial examples are unnatural, which can be easily distinguished by both human and machine. Based on a general anomaly detector, we propose a novel metric (Degree of Anomaly) as a constraint to enable current adversarial attack approaches to generate more natural and imperceptible adversarial examples. Under this new constraint, the success rate of existing attacks drastically decreases, which reveals that the robustness of PrLMs is not as fragile as they claimed. In addition, we find that four types of randomization can invalidate a large portion of textual adversarial examples. Based on anomaly detector and randomization, we design a universal defense framework, which is among the first to perform textual adversarial defense without knowing the specific attack. Empirical results show that our universal defense framework achieves comparable or even higher after-attack accuracy with other specific defenses, while preserving higher original accuracy at the same time. Our work discloses the essence of textual adversarial attacks, and indicates that (i) further works of adversarial attacks should focus more on how to overcome the detection and resist the randomization, otherwise their adversarial examples would be easily detected and invalidated; and (ii) compared with the unnatural and perceptible adversarial examples, it is those undetectable adversarial examples that pose real risks for PrLMs and require more attention for future robustness-enhancing strategies.

Recent pre-trained language models (PrLMs) offer a new performant method of contextualized word representations by leveraging the sequence-level context for modeling. Although the PrLMs generally provide more effective contextualized word representations than non-contextualized models, they are still subject to a sequence of text contexts without diverse hints from multimodality. This paper thus proposes a visual representation method to explicitly enhance conventional word embedding with multiple-aspect senses from visual guidance. In detail, we build a small-scale word-image dictionary from a multimodal seed dataset where each word corresponds to diverse related images. Experiments on 12 natural language understanding and machine translation tasks further verify the effectiveness and the generalization capability of the proposed approach. Analysis shows that our method with visual guidance pays more attention to content words, improves the representation diversity, and is potentially beneficial for enhancing the accuracy of disambiguation.

This paper presents a novel method to generate answers for non-extraction machine reading comprehension (MRC) tasks whose answers cannot be simply extracted as one span from the given passages. Using a pointer network-style extractive decoder for such type of MRC may result in unsatisfactory performance when the ground-truth answers are given by human annotators or highly re-paraphrased from parts of the passages. On the other hand, using a generative decoder cannot well guarantee the resulted answers with well-formed syntax and semantics when encountering long sentences. Therefore, to alleviate the obvious drawbacks of both sides, we propose an answer making-up method from extracted multi-spans that are learned by our model as highly confident$n$-gram candidates in the given passage. That is, the returned answers are composed of discontinuous multi-spans but not just one consecutive span in the given passages anymore. The proposed method is simple but effective: empirical experiments on MS MARCO show that the proposed method has a better performance on accurately generating long answers and substantially outperforms two typical competitive one-span and Seq2Seq baseline decoders.

Multi-choice Machine Reading Comprehension (MRC) as a challenge requires models to select the most appropriate answer from a set of candidates with a given passage and question. Most of the existing researches focus on the modeling of specific tasks or complex networks, without explicitly referring to relevant and credible external knowledge sources, which are supposed to greatly make up for the deficiency of the given passage. Thus we propose a novel reference-based knowledge enhancement model calledReferenceKnowledgeableNetwork (RekNet), which simulates human reading strategies to refine critical information from the passage and quote explicit knowledge in necessity. In detail,RekNetrefines fine-grained critical information and defines it asReference Span, then quotes explicit knowledge quadruples by the co-occurrence information ofReference Spanand candidates. The proposedRekNetis evaluated on three multi-choice MRC benchmarks: RACE, DREAM and Cosmos QA, obtaining consistent and remarkable performance improvement with observable statistical significance level over strong baselines. Our code is available athttps://github.com/Yilin1111/RekNet.

Multi-choice Machine Reading Comprehension (MRC) requires models to decide the correct answer from a set of answer options when given a passage and a question. Thus, in addition to a powerful Pre-trained Language Model (PrLM) as an encoder, multi-choice MRC especially relies on a matching network design that is supposed to effectively capture the relationships among the triplet of passage, question, and answers. While the newer and more powerful PrLMs have shown their strengths even without the support from a matching network, we propose a newDUalMulti-head Co-Attention (DUMA) model. It is inspired by the human transposition thinking process solving the multi-choice MRC problem by considering each other’s focus from the standpoint of passage and question. The proposed DUMA has been shown to be effective and is capable of generally promoting PrLMs. Our proposed method is evaluated on two benchmark multi-choice MRC tasks, DREAM, and RACE. Our results show that in terms of powerful PrLMs, DUMA can further boost the models to obtain higher performance.

In this paper, we report our discovery on named entity distribution in a general word embedding space, which helps an open definition on multilingual named entity definition rather than previous closed and constraint definition on named entities through a named entity dictionary, which is usually derived from human labor and replies on schedule update. Our initial visualization of monolingual word embeddings indicates named entities tend to gather together despite of named entity types and language difference, which enable us to model all named entities using a specific geometric structure inside embedding space, namely, the named entity hypersphere. For monolingual cases, the proposed named entity model gives an open description of diverse named entity types and different languages. For cross-lingual cases, mapping the proposed named entity model provides a novel way to build a named entity dataset for resource-poor languages. At last, the proposed named entity model may be shown as a handy clue to enhance state-of-the-art named entity recognition systems generally.

Machine reading comprehension (MRC) is an AI challenge that requires machines to determine the correct answers to questions based on a given passage. MRC systems must not only answer questions when necessary but also tactfully abstain from answering when no answer is available according to the given passage. When unanswerable questions are involved in the MRC task, an essential verification module called verifier is especially required in addition to the encoder, though the latest practice on MRC modeling still mostly benefits from adopting well pre-trained language models as the encoder block by only focusing on the "reading". This paper devotes itself to exploring better verifier design for the MRC task with unanswerable questions. Inspired by how humans solve reading comprehension questions, we proposed a retrospective reader (Retro-Reader) that integrates two stages of reading and verification strategies: 1) sketchy reading that briefly investigates the overall interactions of passage and question, and yields an initial judgment; 2) intensive reading that verifies the answer and gives the final prediction. The proposed reader is evaluated on two benchmark MRC challenge datasets SQuAD2.0 and NewsQA, achieving new state-of-the-art results. Significance tests show that our model is significantly better than strong baselines.

A multi-turn dialogue is composed of multiple utterances from two or more different speaker roles. Thus utterance- and speaker-aware clues are supposed to be well captured in models. However, in the existing retrieval-based multi-turn dialogue modeling, the pre-trained language models (PrLMs) as encoder represent the dialogues coarsely by taking the pairwise dialogue history and candidate response as a whole, the hierarchical information on either utterance interrelation or speaker roles coupled in such representations is not well addressed. In this work, we propose a novel model to fill such a gap by modeling the effective utterance-aware and speaker-aware representations entailed in a dialogue history. In detail, we decouple the contextualized word representations by masking mechanisms in Transformer-based PrLM, making each word only focus on the words in current utterance, other utterances, two speaker roles (i.e., utterances of sender and utterances of receiver), respectively. Experimental results show that our method boosts the strong ELECTRA baseline substantially in four public benchmark datasets, and achieves various new state-of-the-art performance over previous methods. A series of ablation studies are conducted to demonstrate the effectiveness of our method.

In the retrieval-based multi-turn dialogue modeling, it remains a challenge to select the most appropriate response according to extracting salient features in context utterances. As a conversation goes on, topic shift at discourse-level naturally happens through the continuous multi-turn dialogue context. However, all known retrieval-based systems are satisfied with exploiting local topic words for context utterance representation but fail to capture such essential global topic-aware clues at discourse-level. Instead of taking topic-agnostic n-gram utterance as processing unit for matching purpose in existing systems, this paper presents a novel topic-aware solution for multi-turn dialogue modeling, which segments and extracts topic-aware utterances in an unsupervised way, so that the resulted model is capable of capturing salient topic shift at discourse-level in need and thus effectively track topic flow during multi-turn conversation. Our topic-aware modeling is implemented by a newly proposed unsupervised topic-aware segmentation algorithm and Topic-Aware Dual-attention Matching (TADAM) Network, which matches each topic segment with the response in a dual cross-attention way. Experimental results on three public datasets show TADAM can outperform the state-of-the-art method, especially by 3.3% on E-commerce dataset that has an obvious topic shift.

Zhuosheng Zhang, Hai Zhao. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Conversational machine reading (CMR) requires machines to communicate with humans through multi-turn interactions between two salient dialogue states of decision making and question generation processes.In open CMR settings, as the more realistic scenario, the retrieved background knowledge would be noisy, which results in severe challenges in the information transmission.Existing studies commonly train independent or pipeline systems for the two subtasks.However, those methods are trivial by using hard-label decisions to activate question generation, which eventually hinders the model performance.In this work, we propose an effective gating strategy by smoothing the two dialogue states in only one decoder and bridge decision making and question generation to provide a richer dialogue state reference.Experiments on the OR-ShARC dataset show the effectiveness of our method, which achieves new state-of-the-art results.

Multi-turn dialogue reading comprehension aims to teach machines to read dialogue contexts and solve tasks such as response selection and answering questions. The major challenges involve noisy history contexts and especial prerequisites of commonsense knowledge that is unseen in the given material. Existing works mainly focus on context and response matching approaches. This work thus makes the first attempt to tackle the above two challenges by extracting substantially important turns as pivot utterances and utilizing external knowledge to enhance the representation of context. We propose a pivot-oriented deep selection model (PoDS) on top of the Transformer-based language models for dialogue comprehension. In detail, our model first picks out the pivot utterances from the conversation history according to the semantic matching with the candidate response or question, if any. Besides, knowledge items related to the dialogue context are extracted from a knowledge graph as external knowledge. Then, the pivot utterances and the external knowledge are combined together with a well-designed mechanism for refining predictions. Experimental results on four dialogue comprehension benchmark tasks show that our proposed model achieves great improvements on baselines. A series of empirical comparisons are conducted to show how our selection strategies and the extra knowledge injection influence the results.

For machine reading comprehension, the capacity of effectively modeling the linguistic knowledge from the detail-riddled and lengthy passages and getting ride of the noises is essential to improve its performance. Traditional attentive models attend to all words without explicit constraint, which results in inaccurate concentration on some dispensable words. In this work, we propose using syntax to guide the text modeling by incorporating explicit syntactic constraints into attention mechanism for better linguistically motivated word representations. In detail, for self-attention network (SAN) sponsored Transformer-based encoder, we introduce syntactic dependency of interest (SDOI) design into the SAN to form an SDOI-SAN with syntax-guided self-attention. Syntax-guided network (SG-Net) is then composed of this extra SDOI-SAN and the SAN from the original Transformer encoder through a dual contextual architecture for better linguistics inspired representation. To verify its effectiveness, the proposed SG-Net is applied to typical pre-trained language model BERT which is right based on a Transformer encoder. Extensive experiments on popular benchmarks including SQuAD 2.0 and RACE show that the proposed SG-Net design helps achieve substantial performance improvement over strong baselines.

The latest work on language representations carefully integrates contextualized features into language model training, which enables a series of success especially in various machine reading comprehension and natural language inference tasks. However, the existing language representation models including ELMo, GPT and BERT only exploit plain context-sensitive features such as character or word embeddings. They rarely consider incorporating structured semantic information which can provide rich semantics for language representation. To promote natural language understanding, we propose to incorporate explicit contextual semantics from pre-trained semantic role labeling, and introduce an improved language representation model, Semantics-aware BERT (SemBERT), which is capable of explicitly absorbing contextual semantics over a BERT backbone. SemBERT keeps the convenient usability of its BERT precursor in a light fine-tuning way without substantial task-specific modifications. Compared with BERT, semantics-aware BERT is as simple in concept but more powerful. It obtains new state-of-the-art or substantially improves results on ten reading comprehension and language inference tasks.

State-of-the-art Transformer-based neural machine translation (NMT) systems still follow a standard encoder-decoder framework, in which source sentence representation can be well done by an encoder with self-attention mechanism. Though Transformer-based encoder may effectively capture general information in its resulting source sentence representation, the backbone information, which stands for the gist of a sentence, is not specifically focused on. In this paper, we propose an explicit sentence compression method to enhance the source sentence representation for NMT. In practice, an explicit sentence compression goal used to learn the backbone information in a sentence. We propose three ways, including backbone source-side fusion, target-side fusion, and both-side fusion, to integrate the compressed sentence into NMT. Our empirical tests on the WMT English-to-French and English-to-German translation tasks show that the proposed sentence compression method significantly improves the translation performances over strong baselines.

Multi-choice reading comprehension is a challenging task to select an answer from a set of candidate options when given passage and question. Previous approaches usually only calculate question-aware passage representation and ignore passage-aware question representation when modeling the relationship between passage and question, which cannot effectively capture the relationship between passage and question. In this work, we propose dual co-matching network (DCMN) which models the relationship among passage, question and answer options bidirectionally. Besides, inspired by how humans solve multi-choice questions, we integrate two reading strategies into our model: (i) passage sentence selection that finds the most salient supporting sentences to answer the question, (ii) answer option interaction that encodes the comparison information between answer options. DCMN equipped with the two strategies (DCMN+) obtains state-of-the-art results on five multi-choice reading comprehension datasets from different domains: RACE, SemEval-2018 Task 11, ROCStories, COIN, MCTest.

Memory-based learning can be characterized as a lazy learning method in machine learning terminology because it delays the processing of input by storing the input until needed. Linguistic structure parsing, which has been in a performance improvement bottleneck since the latest series of works was presented, determines the syntactic or semantic structure of a sentence. In this article, we construct a memory component and use it to augment a linguistic structure parser which allows the parser to directly extract patterns from the known training treebank to form memory. The experimental results show that existing state-of-the-art parsers reach new heights of performance on the main benchmarks for dependency parsing and semantic role labeling with this memory network.

Semantic role labeling (SRL) aims to discover the predicateargument structure of a sentence. End-to-end SRL without syntactic input has received great attention. However, most of them focus on either span-based or dependency-based semantic representation form and only show specific model optimization respectively. Meanwhile, handling these two SRL tasks uniformly was less successful. This paper presents an end-to-end model for both dependency and span SRL with a unified argument representation to deal with two different types of argument annotations in a uniform fashion. Furthermore, we jointly predict all predicates and arguments, especially including long-term ignored predicate identification subtask. Our single model achieves new state-of-the-art results on both span (CoNLL 2005, 2012) and dependency (CoNLL 2008, 2009) SRL benchmarks.

Pinyin-to-character (P2C) conversion is the core component of pinyin-based Chinese input method engine (IME).However, the conversion is seriously compromised by the ambiguities of Chinese characters corresponding to pinyin as well as the predefined fixed vocabularies.To alleviate such inconveniences, we propose a neural P2C conversion model augmented by an online updated vocabulary with a sampling mechanism to support open vocabulary learning during IME working.Our experiments show that the proposed method outperforms commercial IMEs and state-of-theart traditional models on standard corpus and true inputting history dataset in terms of multiple metrics and thus the online updated vocabulary indeed helps our IME effectively follows user inputting behavior.

Representation learning is the foundation of machine reading comprehension and inference. In state-of-the-art models, character-level representations have been broadly adopted to alleviate the problem of effectively representing rare or complex words. However, character itself is not a natural minimal linguistic unit for representation or word embedding composing due to ignoring the linguistic coherence of consecutive characters inside word. This paper presents a general subword-augmented embedding framework for learning and composing computationally derived subword-level representations. We survey a series of unsupervised segmentation methods for subword acquisition and different subword-augmented strategies for text understanding, showing that subword-augmented embedding significantly improves our baselines in various types of text understanding tasks on both English and Chinese benchmarks.

Representation learning is the foundation of machine reading comprehension. In state-of-the-art models, deep learning methods broadly use word and character level representations. However, character is not naturally the minimal linguistic unit. In addition, with a simple concatenation of character and word embedding, previous models actually give suboptimal solution. In this paper, we propose to use subword rather than character for word embedding enhancement. We also empirically explore different augmentation strategies on subword-augmented embedding to enhance the cloze-style reading comprehension model (reader). In detail, we present a reader that uses subword-level representation to augment word embedding with a short list to handle rare words effectively. A thorough examination is conducted to evaluate the comprehensive performance and generalization ability of the proposed reader. Experimental results show that the proposed approach helps the reader significantly outperform the state-of-the-art baselines on various public datasets.

Multi-turn conversation understanding is a major challenge for building intelligent dialogue systems. This work focuses on retrieval-based response matching for multi-turn conversation whose related work simply concatenates the conversation utterances, ignoring the interactions among previous utterances for context modeling. In this paper, we formulate previous utterances into context using a proposed deep utterance aggregation model to form a fine-grained context representation. In detail, a self-matching attention is first introduced to route the vital information in each utterance. Then the model matches a response with each refined utterance and the final matching score is obtained after attentive turns aggregation. Experimental results show our model outperforms the state-of-the-art methods on three multi-turn conversation benchmarks, including a newly introduced e-commerce dialogue corpus.

Answering questions from university admission exams (Gaokao in Chinese) is a challenging AI task since it requires effective representation to capture complicated semantic relations between questions and answers. In this work, we propose a hybrid neural model for deep question-answering task from history examinations. Our model employs a cooperative gated neural network to retrieve answers with the assistance of extra labels given by a neural turing machine labeler. Empirical study shows that the labeler works well with only a small training dataset and the gated mechanism is good at fetching the semantic representation of lengthy answers. Experiments on question answering demonstrate the proposed model obtains substantial performance gains over various neural model baselines in terms of multiple evaluation metrics.

Semantic role labeling (SRL) aims to recognize the predicate-argument structure of a sentence.Syntactic information has been paid a great attention over the role of enhancing SRL.However, the latest advance shows that syntax would not be so important for SRL with the emerging much smaller gap between syntax-aware and syntax-agnostic SRL.To comprehensively explore the role of syntax for SRL task, we extend existing models and propose a unified framework to investigate more effective and more diverse ways of incorporating syntax into sequential neural networks.Exploring the effect of syntactic input quality on SRL performance, we confirm that high-quality syntactic parse could still effectively enhance syntactically-driven SRL.Using empirically optimized integration strategy, we even enlarge the gap between syntax-aware and syntax-agnostic SRL.Our framework achieves state-of-the-art results on CoNLL-2009 benchmarks both for English and Chinese, substantially outperforming all previous models.