Papers by Zhuosheng Zhang
Back to the Future: Bidirectional Information Decoupling Network for Multi-turn Dialogue Modeling (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing studies on dialogue modeling use pre-trained language models to encode dialogue history as successive tokens, which is insufficient in capturing the temporal characteristics of dialogues. |
| Approach: | They propose a bidirectional information decoupling network as a universal dialogue encoder which explicitly incorporates both the past and future contexts. |
| Outcome: | The proposed model incorporates past and future contexts and can be generalized to a wide range of dialogue-related tasks. |
Retrieval Augmentation for Commonsense Reasoning: A Unified Approach (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for retrieving encyclopedic knowledge lack a large corpus and effective commonsense retriever. |
| Approach: | They propose a framework for retrieval-augmented commonsense reasoning with a large commonsensense corpus and a commonseense retriever. |
| Outcome: | The proposed framework outperforms existing methods on commonsense reasoning tasks. |
EVA: Evolving Semantic Adversaries for Red-Teaming GUI Agents Against Environmental Injection Attacks (2026.findings-acl)
Copied to clipboard
Yijie Lu, Manman Zhao, Tianjie Ju, Zihe Yan, Xinbei Ma, Yuan Guo, Daizong Ding, Gongshen Liu, Zhuosheng Zhang
| Challenge: | Existing methods for red-teaming face a trade-off between requiring target-specific knowledge and incurring prohibitive computational costs. |
| Approach: | They propose a framework that evolves payloads exclusively on the semantic dimension via a discovery-deployment pipeline. |
| Outcome: | Experiments show that EVA outperforms baselines in terms of attack success rate while evolving benign seeds into successful attacks within 1.18 to 1.71 iterations. |
Self-Prompting Large Language Models for Zero-Shot Open-Domain QA (2024.naacl-long)
Copied to clipboard
| Challenge: | Open-Domain Question Answering (ODQA) aims to answer questions without explicitly providing specific background documents. |
| Approach: | They propose a framework to explicitly utilize the massive knowledge encoded in LLM parameters and their strong instruction understanding abilities. |
| Outcome: | The proposed framework surpasses state-of-the-art methods on three widely-used ODQA datasets and achieves comparable performance with customized fine-tuned models on full training data. |
Structural Characterization for Dialogue Disentanglement (2022.acl-long)
Copied to clipboard
| Challenge: | tangled multi-party dialogues lead to difficulties in understanding the dialogue history for both human and machine. |
| Approach: | They propose a model for disentangling multi-party dialogues using speaker property and reference dependency. |
| Outcome: | The proposed model achieves state-of-the-art on the Ubuntu IRC benchmark dataset and contributes to dialogue-related comprehension. |
GuideBench: Benchmarking Domain-Oriented Guideline Following for LLM Agents (2025.acl-long)
Copied to clipboard
| Challenge: | Large language models (LLMs) have been widely deployed as autonomous agents capable of following user instructions and making decisions in real-world applications. |
| Approach: | They propose a benchmark to evaluate LLMs' ability to follow domain-oriented guidelines . they evaluate Lms on three critical aspects: adherence to diverse rules, robustness to rule updates . |
| Outcome: | The proposed benchmark evaluates LLMs on three critical aspects: adherence to diverse rules, robustness to rule updates, and alignment with human preferences. |
Dialogue Graph Modeling for Conversational Machine Reading (2021.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for conversational machine reading (CMR) are not effective for capturing multiple objects in complex interactive scenarios. |
| Approach: | They propose a dialogue graph modeling framework that captures explicit and implicit interactions hidden in the rule documents and a model that asks clarification questions to the machine. |
| Outcome: | The proposed model exceeds the milestone accuracy score of 80% on the ShARC benchmark and achieves new state-of-the-art by first exceeding the milestone precision score of 90%. |
Modeling Hierarchical Reasoning Chains by Linking Discourse Units and Key Phrases for Reading Comprehension (2022.coling-1)
Copied to clipboard
| Challenge: | Existing methods of logical reasoning focus on entity-aware information but ignore hierarchical relations that may even have mutual effects. |
| Approach: | They propose a holistic graph network that deals with context at both discourse-level and word-level as the basis for logical reasoning. |
| Outcome: | The proposed method improves on logical reasoning QA datasets and natural language inference datasets. |
The Confidence Paradox: Unveiling the Latent Discriminative Power of Diffusion Large Language Models in Mathematical Reasoning (2026.findings-acl)
Copied to clipboard
| Challenge: | Diffusion large language models (DLLMs) are a promising alternative to autoregressive (AR) generation, offering token-level probabilities under bidirectional context. |
| Approach: | They propose to use diffusion large language models to generate token-level probabilities under bidirectional context and to examine the calibration paradox inherent to their native uncertainty estimates. |
| Outcome: | The proposed model outperforms AR baselines on mathematical reasoning benchmarks and is highly miscalibrated on reasoning benchmark. |
Instance Regularization for Discriminative Language Model Pre-training (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing studies have optimized independent strategies of ennoising or denosing . Existing methods treat training instances equally throughout the training process . |
| Approach: | They propose to use ennoising and denoising to train discriminative pre-trained language models . they propose to model the complexity of restoring the original sentences from corrupted ones . |
| Outcome: | Experimental results show that the proposed method improves pre-training efficiency, effectiveness, and robustness. |
MedAgents: Large Language Models as Collaborators for Zero-shot Medical Reasoning (2024.findings-acl)
Copied to clipboard
Xiangru Tang, Anni Zou, Zhuosheng Zhang, Ziming Li, Yilun Zhao, Xingyao Zhang, Arman Cohan, Mark Gerstein
| Challenge: | Large language models face unique challenges such as domain-specific terminologies and reasoning over specialized knowledge. |
| Approach: | They propose a multi-disciplinary collaboration framework that leverages LLM-based agents in a role-playing setting. |
| Outcome: | The proposed framework excels at mining and harnessing medical expertise within LLMs, as well as extending its reasoning abilities. |
Smoothing Dialogue States for Open Conversational Machine Reading (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing studies train independent or pipeline systems for the two subtasks but are trivial by using hard-label decisions to activate question generation. |
| Approach: | They propose a method to smooth two dialogue states in one decoder and bridge decision making and question generation to provide a richer dialogue state reference. |
| Outcome: | The proposed method achieves state-of-the-art on the OR-ShARC dataset. |
Meta-Reasoning: Semantics-Symbol Deconstruction for Large Language Models (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing methods rely on syntactically mapping natural languages to complete formal languages like Python and SQL. |
| Approach: | They propose to deconstruct reasoning-independent semantic information into generic symbolic representations, thereby efficiently capturing more generalized reasoning knowledge. |
| Outcome: | The proposed method improves in-context reasoning accuracy, learning efficiency, out-of-domain generalization, and output stability compared to the Chain-of thought technique. |
Tracing Origins: Coreference-aware Machine Reading Comprehension (2022.acl-long)
Copied to clipboard
| Challenge: | a recent study has enriched pre-trained language models with syntactic, semantic and other linguistic information to improve their performance. |
| Approach: | They use a pre-trained language model to leverage coreference information to enhance word embeddings . they use additional encoder layers to focus on coreference mentions or a relational graph convolutional network to model the coreference relations. |
| Outcome: | The proposed model imitates the human reading process and leverages coreference information to enhance word embeddings. |
Is ChatGPT a General-Purpose Natural Language Processing Task Solver? (2023.emnlp-main)
Copied to clipboard
| Challenge: | Recent advances in scale have enabled large language models to perform NLP tasks zero-shot . however, it is not known whether ChatGPT can serve as a generalist model that can perform many NLP jobs zero- shot. |
| Approach: | They empirically evaluate ChatGPT's zero-shot learning ability on 20 popular NLP datasets . they find it performs well on many tasks favoring reasoning abilities . |
| Outcome: | The proposed model can perform many NLP tasks zero-shot without adaptation on downstream data. |
Investigating Multi-Hop Factual Shortcuts in Knowledge Editing of Large Language Models (2024.acl-long)
Copied to clipboard
| Challenge: | Recent work has demonstrated the power of large language models in recalling knowledge and reasoning. |
| Approach: | They propose to erase shortcut neurons to mitigate the associated risks . 20% of the failures are attributed to shortcuts, they find . |
| Outcome: | The proposed approach reduces failures in multi-hop knowledge editing caused by shortcuts by 20% . |
Improving Machine Translation with Human Feedback: An Exploration of Quality Estimation as a Reward Model (2024.naacl-long)
Copied to clipboard
| Challenge: | Existing methods to improve translation quality using human feedback have not been validated. |
| Approach: | They propose to use quality estimation to predict human preferences for feedback training . they propose to detect incorrect translations and assign a penalty term to the reward scores . |
| Outcome: | The proposed method outperforms systems using larger parallel corpora by a small amount of monolingual data. |
Towards End-to-End Open Conversational Machine Reading (2023.findings-eacl)
Copied to clipboard
| Challenge: | Existing approaches to the problem of open-retrieval conversational machine reading (OR-CMR) use two separate modules to approach the problem's two successive sub-tasks. |
| Approach: | They propose to model OR-CMR as a unified text-to-text task in a fully end-to end style and propose to use a text-based approach to solve the problem. |
| Outcome: | Experiments on the ShARC and OR-ShARC dataset show that the proposed framework can generalize to different backbone models. |
ParaCook: On Time-Efficient Planning for Multi-Agent Systems (2026.findings-acl)
Copied to clipboard
Shiqi Zhang, Xinbei Ma, Yunqing Xu, Zouying Cao, Pengrui Lu, Haobo Yuan, Tiancheng Shen, Zhuosheng Zhang, Hai Zhao, Ming-Hsuan Yang
| Challenge: | Existing agent benchmarks focus on task completion while neglecting time efficiency in parallel and asynchronous operations. |
| Approach: | They propose a framework for large language models that allows agents to plan long-horizon tasks in a scalable way. |
| Outcome: | The proposed framework is based on the Overcooked game and can be used to evaluate time efficiency-aware multi-agent planning. |
CoCo-Agent: A Comprehensive Cognitive MLLM Agent for Smartphone GUI Automation (2024.findings-acl)
Copied to clipboard
| Challenge: | Current vital challenges for autonomous agents lie in two aspects: dependence on strong (M)LLMs and insufficient GUI environment modeling. |
| Approach: | They propose a comprehensive cognitive LLM agent with two novel approaches to improve GUI automation performance. |
| Outcome: | The proposed agent achieves state-of-the-art performance on AITW and META-GUI benchmarks. |
From Multimodal LLM to Human-level AI: Modality, Instruction, Reasoning, Efficiency and beyond (2024.lrec-tutorials)
Copied to clipboard
| Challenge: | This tutorial aims to deliver a comprehensive review of cutting-edge research in MLLMs. |
| Approach: | This tutorial will review cutting-edge research in MLLMs and examine the impact of ML in learning and reasoning. |
| Outcome: | This course will review cutting-edge research in MLLMs and examine the impact of ML models on learning, learning, and multimodal reasoning. |
One-shot Learning for Question-Answering in Gaokao History Challenge (C18-1)
Copied to clipboard
| Challenge: | Existing work on deep question-answering tasks from admission exams is challenging since it requires effective representation to capture complicated semantic relations between questions and answers. |
| Approach: | They propose a hybrid neural model for deep question-answering task from history examinations using a gated network and a machine labeler. |
| Outcome: | The proposed model obtains substantial performance gains over baseline models in terms of multiple evaluation metrics. |
Measuring Bargaining Abilities of LLMs: A Benchmark and A Buyer-Enhancement Method (2024.findings-acl)
Copied to clipboard
| Challenge: | Using a novel approach, we can evaluate an agent’s bargaining abilities as an asymmetric incomplete information game. |
| Approach: | They propose an approach that integrates a deterministic Offer Generator and an LLM Narrator to create natural language sentences for generated offers. |
| Outcome: | The proposed approach improves the buyer’s deal rates from 26.67% to 88.88% and brings a ten times multiplication of profits on all baselines, even a model that has not been aligned. |
OS-Kairos: Adaptive Interaction for MLLM-Powered GUI Agents (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing efforts to build GUI agents focused on the autonomous mode have failed to address the problem of over-execution. |
| Approach: | They propose an adaptive GUI agent that predicts confidence levels at each interaction step and elicits adaptive interaction. |
| Outcome: | The proposed GUI agent outperforms existing models on a complex dataset and on established benchmarks. |
Gracefully Filtering Backdoor Samples for Generative Large Language Models without Retraining (2025.coling-main)
Copied to clipboard
| Challenge: | Existing backdoor defense methods are ineffective for generative large language models . generative LLMs output sequences of high-dimensional token logits instead of low-dimensional classification logits . |
| Approach: | They propose a method that leverages sample-wise gradients to identify backdoor samples without retraining LLMs. |
| Outcome: | The proposed method outperforms baselines significantly in identifying backdoor samples without retraining LLMs. |
Agent-Dice: Disentangling Knowledge Updates via Geometric Consensus for Agent Continual Learning (2026.findings-acl)
Copied to clipboard
| Challenge: | Large Language Model (LLM)-based agents extend the utility of LLMs by interacting with dynamic environments. |
| Approach: | They propose a parameter fusion framework based on directional consensus evaluation that disentangles knowledge updates through a two-stage process. |
| Outcome: | The proposed framework disentangles knowledge updates through a two-stage process with minimal computational overhead and parameter updates. |
A Unified Syntax-aware Framework for Semantic Role Labeling (D18-1)
Copied to clipboard
| Challenge: | Syntactic information has been paid a great attention over the role of enhancing SRL . but the gap between syntax-aware and syntax-gnostic SRL is smaller . a new framework proposes syntax-based SRL for a wide range of NLP tasks . |
| Approach: | They propose to extend existing models to investigate more effective ways of incorporating syntax into sequential neural networks. |
| Outcome: | The proposed framework outperforms existing models on CoNLL-2009 benchmarks in English and Chinese. |
Decker: Double Check with Heterogeneous Knowledge for Commonsense Fact Verification (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing studies focus on grasping unstructured evidence or potential reasoning paths from structured knowledge bases, yet failing to exploit the benefits of heterogeneous knowledge simultaneously. |
| Approach: | They propose a commonsense fact verification model that bridging heterogeneous knowledge by uncovering latent relationships between structured and unstructured knowledge. |
| Outcome: | The proposed model can bridge heterogeneous knowledge by uncovering latent relationships between structured and unstructured knowledge. |
Moon IME: Neural-based Chinese Pinyin Aided Input Method with Customizable Association (P18-4)
Copied to clipboard
| Challenge: | a pinyin input method engine (IME) allows users to input Chinese into a computer by typing pinyan through the common keyboard. |
| Approach: | They present a pinyin IME that integrates neural machine translation and IR to offer amusive and customizable association ability. |
| Outcome: | The Moon IME integrates neural machine translation and IR to offer amusive association ability. |
SynGhost: Invisible and Universal Task-agnostic Backdoor Attack via Syntactic Transfer (2025.findings-naacl)
Copied to clipboard
| Challenge: | Existing attacks are classified into end-to-end and pre-training types based on the attack phase . Existing backdoor attacks are based upon perplexity, fine-pruning, and maxEntropy. |
| Approach: | They propose an entropy-based poisoning filter that mitigates backdoor attacks . they propose an invisible and universal task-agnostic backdoor attack via syntactic transfer . |
| Outcome: | The proposed attack can transfer backdoors to various downstream tasks while preserving pre-trained language models' pre-training capabilities. |
Hidden Ghost Hand: Unveiling Backdoor Vulnerabilities in MLLM-Powered Mobile GUI Agents (2025.findings-emnlp)
Copied to clipboard
Pengzhou Cheng, Haowen Hu, Zheng Wu, Zongru Wu, Tianjie Ju, Daizong Ding, Zhuosheng Zhang, Gongshen Liu
| Challenge: | MLLM-powered GUI agents expose multiple interaction-level triggers, causing backdoor attacks . backdoor injection maximizes feature difference across sample classes, improving flexibility . |
| Approach: | They propose a framework for red-teaming backdoor attacks using MLLMs . they construct composite triggers by combining goal and interaction levels . |
| Outcome: | The proposed framework is effective and stealthy for red-teaming backdoor attacks. |
Dynamic Planning for LLM-based Graphical User Interface Automation (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing approaches to planning for GUI tasks are limited due to long historical dialogues. |
| Approach: | They propose a novel approach to dynamic planning based on environmental feedback and execution history to guide action prediction in GUI tasks. |
| Outcome: | The proposed approach surpasses the strong GPT-4V baseline by +12.7% in accuracy. |
Sentence-aware Contrastive Learning for Open-Domain Passage Retrieval (2022.acl-long)
Copied to clipboard
| Challenge: | Existing studies focus on improving negative sampling strategy or extra pretraining for dense passage representations, but these studies are not capturing passage with internal representation conflicts. |
| Approach: | They propose a model with a smaller granularity to capture internal representation conflicts . they introduce a negative sampling strategy to encourage a diverse generation of sentence representations within the same passage. |
| Outcome: | The proposed model can be trained on three benchmark datasets to alleviate internal representation conflicts. |
Modeling Multi-turn Conversation with Deep Utterance Aggregation (C18-1)
Copied to clipboard
| Challenge: | Existing work on retrieval-based context modeling for multi-turn conversation ignores interactions among previous utterances. |
| Approach: | They propose retrieval-based response matching for multi-turn conversation . they propose to combine previous utterances into context using a deep utterrance aggregation model . |
| Outcome: | The proposed model outperforms state-of-the-art methods on three multi-turn conversation benchmarks including an e-commerce dialogue corpus. |
Acquiring Clean Language Models from Backdoor Poisoned Datasets by Downscaling Frequency Space (2024.acl-long)
Copied to clipboard
| Challenge: | Prior work attempts to mitigate backdoor learning during training LMs on poisoned datasets . backdoor attack poisons a small portion of training data by implanting specific text patterns . |
| Approach: | They propose a multi-scale low-rank adaptive model that prioritizes learning of clean mapping . they propose radial scalings to reduce the success rate of diverse backdoor attacks . |
| Outcome: | The proposed model outperforms baselines significantly in the frequency space . it reduces the success rate of diverse backdoor attacks to below 15% across datasets . |
Span Fine-tuning for Pre-trained Language Models (2021.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods to fine-tune pre-trained language models are time-consuming and lack flexibility. |
| Approach: | They propose a span fine-tuning method which allows for a more efficient and efficient way of incorporating span-level information into pre-training. |
| Outcome: | Experiments on GLUE benchmark show that the proposed method significantly enhances the PrLM and offers more flexibility in an efficient way. |
LIMIT-BERT : Linguistics Informed Multi-Task BERT (2020.findings-emnlp)
Copied to clipboard
| Challenge: | Existing language models are usually trained on large amounts of unlabeled text data. |
| Approach: | They propose a multi-task language representations learning framework for multi-linguistics tasks by Multi-Task Learning. |
| Outcome: | The proposed model outperforms the baseline Whole Word Masking BERT on both dependency and constituent syntactic/semantic parsing, GLUE benchmark, and SNLI task. |
Structural Pre-training for Dialogue Comprehension (2021.acl-long)
Copied to clipboard
| Challenge: | Recent advances in large-scale pre-training language models (PrLMs) have achieved remarkable successes in a variety of natural language processing tasks. |
| Approach: | They propose to use SPIDER to capture dialogue exclusive features from dialogue texts. |
| Outcome: | The proposed model performs well on widely used dialogue benchmarks. |
OS-Sentinel: Towards Safety-Enhanced Mobile GUI Agents via Hybrid Validation in Realistic Workflows (2026.acl-long)
Copied to clipboard
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
| Challenge: | Existing methods for detecting unsafe mobile GUI agents are underexplored. |
| Approach: | They propose a mobile agent safety detection framework that integrates a formal verifier and a VLM-based contextual judge to detect system-level violations. |
| Outcome: | The proposed framework achieves 10%–30% improvements over existing approaches across multiple metrics. |
R-Judge: Benchmarking Safety Risk Awareness for LLM Agents (2024.findings-emnlp)
Copied to clipboard
Tongxin Yuan, Zhiwei He, Lingzhong Dong, Yiming Wang, Ruijie Zhao, Tian Xia, Lizhen Xu, Binglin Zhou, Fangqi Li, Zhuosheng Zhang, Rui Wang, Gongshen Liu
| Challenge: | Large language models (LLMs) have shown compelling abilities in reasoning, decision-making, and instruction following. |
| Approach: | They propose a benchmark to evaluate the proficiency of large language models (LLMs) in judging and identifying safety risks given agent interaction records. |
| Outcome: | The proposed model outperforms the best-performing model, GPT-4o, while no other models significantly exceed the random. |
On the Robustness of Editing Large Language Models (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing studies have exhibited impressive success and significant potential. |
| Approach: | They propose to modify the knowledge memory with minimum computational cost while preserving the performance on the retained knowledge. |
| Outcome: | The proposed methods avoid retraining to update the model parameters and have demonstrated promising performance and efficiency. |
AuRoRA: A One-for-all Platform for Augmented Reasoning and Refining with Task-Adaptive Chain-of-Thought Prompting (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing methods for large language models (LLMs) lean on handcrafted or task-specific demonstrations and lack reliable knowledge base. |
| Approach: | They propose a one-for-all platform for augmented reasoning and refining based on chain-of-thought prompting that excels in adaptability, reliability, integrity, and interpretability. |
| Outcome: | The proposed system exhibits superior performances across six reasoning tasks and offers real-time visual analysis. |
Learning Better Masking for Better Language Model Pre-training (2023.acl-long)
Copied to clipboard
| Challenge: | Existing PrLMs adopt a Random-Token Masking strategy with a fixed masking ratio and different contents are masked by an equal probability throughout the training. |
| Approach: | They propose two scheduled masking approaches that adaptively tune masking ratio and masked content in different training stages, which improves pre-training efficiency and effectiveness. |
| Outcome: | The proposed methods improve the pre-training efficiency and effectiveness on the downstream tasks. |
You Only Look at Screens: Multimodal Chain-of-Action Agents (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to creating autonomous graphical user interfaces rely on external tools and application-specific APIs to interpret the environment. |
| Approach: | They propose a multimodal solution that directly interacts with the user interface without environment parsing. |
| Outcome: | The proposed solution bypasses environment parsing and reliance on application-dependent APIs. |
Can Watermarks Survive Translation? On the Cross-lingual Consistency of Text Watermark for Large Language Models (2024.acl-long)
Copied to clipboard
| Challenge: | Existing text watermarking technologies lack consistency when texts are translated into different languages. |
| Approach: | They propose a cross-lingual watermark removal attack to bypass watermarking by first obtaining a response from an LLM in a pivot language and then translating it into the target language. |
| Outcome: | The proposed method can remove watermarks without performance loss by obtaining a response from an LLM in a pivot language and then translating it into the target language. |
Distinguishing Non-natural from Natural Adversarial Samples for More Robust Pre-trained Language Model (2022.findings-acl)
Copied to clipboard
| Challenge: | Recent studies on adversarial attacks achieve high success rates against PrLMs, claiming that they are not robust. |
| Approach: | They propose to use anomaly detector to evaluate PrLMs with more natural adversarial samples to evaluate their robustness. |
| Outcome: | The proposed method can be used to defend all types of attacks and achieve higher accuracy on adversarial samples and compliant samples than other defense frameworks. |
Mitigating Misleading Chain-of-Thought Reasoning with Selective Filtering (2024.lrec-main)
Copied to clipboard
| Challenge: | Large language models have demonstrated remarkable capabilities by leveraging chain-of-thought reasoning techniques to solve complex questions. |
| Approach: | They propose a method that assesses the entailment relationship between the question and the candidate reasoning chain and uses it to predict the answer. |
| Outcome: | The proposed approach improves the fine-tuned T5 baseline over the ScienceQA, ECQA, and LastLetter tasks. |
Caution for the Environment: Multimodal LLM Agents are Susceptible to Environmental Distractions (2025.acl-long)
Copied to clipboard
| Challenge: | Experimental results show that multimodal GUI agents are susceptible to environmental distractions. |
| Approach: | They propose a scenario where both user and agent are benign and environment is not malicious . they implement an adversarial environment injection and analyze the approach to improve faithfulness . |
| Outcome: | The proposed approach improves faithfulness of multimodal large language model agents in a graphical user interface environment. |
ColorBrowserAgent: Complex Long-Horizon Browser Agent with Adaptive Knowledge Evolution (2026.acl-industry)
Copied to clipboard
Jihong Wang, Jiamu Zhou, Weiming Zhang, Teng Wang, Weiwen Liu, Zhuosheng Zhang, Xingyu Lou, Weinan Zhang, Huarong Deng, Jun Wang
| Challenge: | Xue et al., 2025): deploying autonomous web agents in production remains difficult due to site heterogeneity and long-horizon instability. |
| Approach: | They propose a knowledge-evolving agent that can be used to automate web workflows . they use human-in-the-loop knowledge adaptation and knowledge-aligned progressive summarization . |
| Outcome: | Experiments on WebArena, WebChoreAren and industrial deployment show it outperforms baselines. |
Subword-augmented Embedding for Cloze Reading Comprehension (C18-1)
Copied to clipboard
| Challenge: | Existing models for machine reading comprehension use word and character representations, but character is not the minimal unit. |
| Approach: | They propose to use subword rather than character for word embedding enhancement . they also empirically explore different augmentation strategies on subword-augmented embedded embedders . |
| Outcome: | The proposed model outperforms state-of-the-art models on public datasets. |
Open Vocabulary Learning for Neural Chinese Pinyin IME (P19-1)
Copied to clipboard
| Challenge: | Pinyin-to-character conversion is the core component of pinyin based Chinese input method engine (IME). |
| Approach: | They propose a neural P2C conversion model augmented by an online updated vocabulary to support open vocabulary learning during IME working. |
| Outcome: | The proposed model outperforms commercial IMEs and state-of-the-art models on standard corpus and true inputting history dataset in terms of multiple metrics and the online updated vocabulary helps it follow user inputting behavior. |
Task Compass: Scaling Multi-task Pre-training with Task Prefix (2022.findings-emnlp)
Copied to clipboard
Zhuosheng Zhang, Shuohang Wang, Yichong Xu, Yuwei Fang, Wenhao Yu, Yang Liu, Hai Zhao, Chenguang Zhu, Michael Zeng
| Challenge: | Existing studies show that multi-task learning with large-scale supervised tasks suffers from negative effects across tasks. |
| Approach: | They propose a task prefix guided multi-task pre-training framework to explore the relationships among tasks. |
| Outcome: | The proposed model can be used as a foundation backbone for a wide range of tasks and as augmentation tool for data augmentation with complementary tasks. |
GLaPE: Gold Label-agnostic Prompt Evaluation for Large Language Models (2024.emnlp-main)
Copied to clipboard
| Challenge: | Recent studies have explored leveraging the LLM itself as an optimizer to identify optimal prompts that maximize task accuracy. |
| Approach: | They propose a gold label-agnostic prompt evaluation method to reduce dependence on gold labels. |
| Outcome: | The proposed method produces more effective prompts even without gold labels. |
Element-aware Summarization with Large Language Models: Expert-aligned Evaluation and Chain-of-Thought Method (2023.acl-long)
Copied to clipboard
| Challenge: | Experimental results show that automatic summarization generates concise summaries that contain key ideas of source documents. |
| Approach: | They propose to use Element-aware test sets to annotate news-related reference summaries to focus on more fine-grained news elements objectively and comprehensively. |
| Outcome: | The proposed method outperforms state-of-the-art fine-tuned PLMs and zero-shot LLMs by +4.33/+4.77 on the two datasets, respectively. |
Lingke: a Fine-grained Multi-turn Chatbot for Customer Service (C18-2)
Copied to clipboard
| Challenge: | e-commerce chatbots usually need a mass of human dialogue data to train, but for multi-turn conversations, the performance is poor. |
| Approach: | They propose an information retrieval augmented multi-turn chatbot which can answer questions based on unstructured documents and deal with multi-turned conversations. |
| Outcome: | The proposed solution outperforms all other models in multi-turn conversations and can learn from conversation records. |
MEGen: Generative Backdoor into Large Language Models via Model Editing (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for training large language models are limited to yes-or-no discriminative tasks, leading users to underestimate the potential risks. |
| Approach: | They propose an editing-based generative backdoor that expands the backdoor to generative tasks in a unified format of any text-to-any text. |
| Outcome: | The proposed model achieves high attack success rate by adjusting only a small set of local parameters with few-shot samples. |