Papers by Kexin Wang

31 papers
LLM-VA: Resolving the Jailbreak-Overrefusal Trade-off via Vector Alignment (2026.acl-long)

Copied to clipboard

Challenge: Existing vector steering methods adjust the magnitude of answer vectors, but this creates a fundamental trade-off—reducing jailbreak increases over-refusal.
Approach: They propose a method which aligns va with vb through closed-form weight updates, making the model’s willingness to respond causally dependent on its safety assessment.
Outcome: Experiments on 12 LLMs show that the proposed method achieves 11.45% higher F1 than the best baseline while preserving 95.92% utility.
Flames: Benchmarking Value Alignment of LLMs in Chinese (2024.naacl-long)

Copied to clipboard

Challenge: Existing benchmarks for large language models (LLMs) do not accurately uncover safety vulnerabilities in LLMs.
Approach: They propose a value alignment benchmark called Flames that encompasses both harmlessness principles and a unique morality dimension that integrates specific Chinese values such as harmony.
Outcome: The proposed model performs poorly on Flames, particularly in safety and fairness dimensions.
MOSPC: MOS Prediction Based on Pairwise Comparison (2023.acl-short)

Copied to clipboard

Challenge: et al., 2016a) show that MOS prediction model can improve ranking accuracy of speech quality.
Approach: They propose a general framework for MOS prediction based on pair comparison . they use C-Mixup algorithm to enhance generalization performance of MOSPC .
Outcome: The proposed model outperforms baselines on most correlation coefficient metrics . it also surpasses the strong baseline in ranking accuracy on each fine-grained segment.
LaMP-Val: Large Language Models Empower Personalized Valuation in Auction (2025.findings-emnlp)

Copied to clipboard

Challenge: Currently, most research focuses on the bidding algorithms used within auction mechanisms.
Approach: They propose a personalized valuation framework that integrates Large Language Models to incorporate personalized semantic preference into users valuation process.
Outcome: The proposed framework incorporates Large Language Models to incorporate personalized semantic preference into users valuation process.
Sticking to the Mean: Detecting Sticky Tokens in Text Embedding Models (2025.acl-long)

Copied to clipboard

Challenge: Sticky tokens, when repeatedly inserted into sentences, pull sentence similarity toward a certain value, disrupting the normal distribution of embedding distances and degrading downstream performance.
Approach: They propose a method to detect “sticky tokens” by sentence and token filtering and apply it to 40 checkpoints across 14 model families.
Outcome: The proposed method detects 868 sticky tokens across 14 models and shows that their presence does not correlate with model size or vocabulary size.
Parameter-Efficient Transfer Learning for End-to-end Speech Translation (2024.lrec-main)

Copied to clipboard

Challenge: Existing approaches to improve end-to-end speech translation are limited by the availability of labeled data.
Approach: They propose a method which utilizes two lightweight adaptation techniques to modulate Attention and the Feed-Forward Network while preserving the capabilities of pre-trained models.
Outcome: The proposed method outperforms baseline models and significantly improves performance in low-resource settings.
Fake Alignment: Are LLMs Really Aligned Well? (2024.naacl-long)

Copied to clipboard

Challenge: Existing studies on large language models have shown that they are poorly aligned in practice.
Approach: They propose a framework to evaluate safety in large language models . they propose two new metrics to quantify fake alignment and obtain corrected performance estimation.
Outcome: The proposed framework and two metrics show that some models with purported safety are poorly aligned in practice.
Semantic Similarity Models for Depression Severity Estimation (2023.emnlp-main)

Copied to clipboard

Challenge: Public health systems have limited capacity for case detection and diagnosis due to the widespread use of social media.
Approach: They propose to use social media content to generate semantic rankings for depressive symptoms and severity levels and use them to predict symptoms severity.
Outcome: The proposed pipeline improves on two Reddit-based benchmarks and shows that it is more efficient than state-of-the-art in terms of measuring depression level.
UKP-SQuARE v2: Explainability and Adversarial Attacks for Trustworthy QA (2022.aacl-demo)

Copied to clipboard

Challenge: Question Answering (QA) systems rely on deep neural networks, which are difficult to interpret by humans.
Approach: They propose an interpretable model that provides an explanation infrastructure for comparing models based on saliency maps and graph-based explanations.
Outcome: The proposed methods can be used to compare models based on saliency maps and graph-based explanations.
ACORD: An Expert-Annotated Retrieval Dataset for Legal Contract Drafting (2025.acl-long)

Copied to clipboard

Challenge: Contract clause retrieval is critical to contract drafting because of its high quality and complexity.
Approach: They propose the first expert-annotated benchmark specifically designed for contract clause retrieval . ACORD focuses on complex contract clauses such as Limitation of Liability, Indemnification, Change of Control .
Outcome: The atticus clause retrieval dataset shows promising results but needs improvement . the benchmark can be used as an IR benchmark for the NLP community .
MMDEND: Dendrite-Inspired Multi-Branch Multi-Compartment Parallel Spiking Neuron for Sequence Modeling (2025.acl-long)

Copied to clipboard

Challenge: Vanilla spiking neurons are simplified from complex biological neurons with dendrites, soma, and synapses into single somatic compartments.
Approach: They propose a multi-branch, multi-compartment parallel spiking dendritic neuron with a proportion-adjustable multi-branched structure that enables long-term temporal dependencies.
Outcome: The proposed model achieves better long-sequence modeling capability with fewer parameters and lower energy consumption.
TSDAE: Using Transformer-based Sequential Denoising Auto-Encoderfor Unsupervised Sentence Embedding Learning (2021.findings-emnlp)

Copied to clipboard

Challenge: Existing methods to learn sentence embeddings require labeled data, but it is expensive.
Approach: They propose an unsupervised method which learns sentence embeddings using unlabeled data . they propose a transformer-based sequence denoising auto-encoder which can be used for training .
Outcome: The proposed method outperforms existing methods on four datasets from heterogeneous domains.
PLAWBENCH: A Rubric-Based Benchmark for Evaluating LLMs in Real-World Legal Practice (2026.acl-long)

Copied to clipboard

Challenge: Existing benchmarks for large language models (LLMs) are coarse, single-dimensional metrics and do not explicitly assess fine-grained legal reasoning.
Approach: They propose a Practical Law Benchmark to evaluate large language models in real-world legal practice scenarios.
Outcome: The proposed model is based on 850 questions and 13 scenarios with expert-designed evaluation rubrics.
Invisible Entropy: Towards Safe and Efficient Low-Entropy LLM Watermarking (2025.emnlp-main)

Copied to clipboard

Challenge: Existing methods to watermark low-entropy content are expensive and risky . IE reduces parameter size by 99% while achieving performance on par with state-of-the-art methods .
Approach: They propose a logit-based watermarking paradigm that uses entropy-based features to predict whether the next token is high or low.
Outcome: The proposed method reduces parameter size by 99% while achieving performance on par with state-of-the-art methods.
LLMEval-Fair: A Large-Scale Longitudinal Study on Robust and Fair Evaluation of Large Language Models (2026.acl-long)

Copied to clipboard

Challenge: Existing evaluation of Large Language Models on static benchmarks is vulnerable to data contamination and leaderboard overfitting.
Approach: LLMEval-Fair framework provides a framework for dynamic evaluation of Large Language Models . evaluators use a proprietary bank of 220k graduate-level questions to analyze model data .
Outcome: LLMEval-Fair provides robust and credible evaluation framework for Large Language Models . it provides a strong empirical validation for the dynamic evaluation paradigm .
Probing the Safety Robustness of LLMs in Latent Space (2026.acl-long)

Copied to clipboard

Challenge: Despite substantial progress in safety alignment techniques, aligned large language models can still produce unsafe responses under minor internal perturbations.
Approach: They introduce Activation Steering Attack (ASA) and leverage the Negative Log-Likelihood (NLL) as a diagnostic signal to probe the local sensitivity of safety behaviors in latent space.
Outcome: The proposed method is model-agnostic and supervision-free, enabling a general and reproducible diagnostic metric for analyzing safety robustness.
A Survey of Large Language Models for Text-Guided Molecular Discovery: From Molecule Generation to Optimization (2026.acl-long)

Copied to clipboard

Challenge: Large language models (LLMs) are introducing a paradigm shift in molecular discovery by enabling text-guided interaction with chemical spaces through natural language and symbolic notations.
Approach: They analyze the current LLM learning paradigms to tackle four critical evaluation dimensions that have emerged as critical dimensions in recent studies.
Outcome: The proposed models are able to interact with chemical spaces through natural language and symbolic notations, and have emerging extensions to incorporate multi-modal inputs.
GPL: Generative Pseudo Labeling for Unsupervised Domain Adaptation of Dense Retrieval (2022.naacl-main)

Copied to clipboard

Challenge: Dense retrieval approaches suffer from the lexical gap and require large amounts of training data.
Approach: They propose an unsupervised method for domain adaptation that uses query generator and pseudo labeling from a cross-encoder to improve retrieval performance.
Outcome: The proposed method outperforms state-of-the-art retrieval methods on domain-specialized datasets by 9.3 points nDCG@10 on six tasks.
From Evasion to Concealment: Stealthy Knowledge Unlearning for LLMs (2025.findings-acl)

Copied to clipboard

Challenge: Existing approaches to unlearning often treat nonsensical responses or template-based refusals as the unlearning target, making the process even more vulnerable to attacks and jailbreaks.
Approach: They propose a method that uses inverted facts to remove the need for auxiliary models or retaining data while avoiding leakage.
Outcome: Evaluated on the ToFU Knowledge Unlearning dataset using Llama2-7B-Chat and Phi-1.5, MEOW outperforms baselines in forgetting quality while preserving model utility.
DAPR: A Benchmark on Document-Aware Passage Retrieval (2024.acl-long)

Copied to clipboard

Challenge: Recent approaches to retrieve short passages are limited due to expensive operations such as selfattention.
Approach: They propose to use a task to find passages from a large corpus of documents . they extend the SoTA passage retrievers with document context to achieve this task .
Outcome: The proposed task is called Document-Aware Passage Retrieval (DAPR) it performs best on easy and hard queries, but fails on hard queries requiring document-context understanding .
MAS-Bench: A Unified Benchmark for Shortcut-Augmented Hybrid Mobile GUI Agents (2026.acl-long)

Copied to clipboard

Challenge: Shortcuts such as APIs and deep-links have emerged as efficient complements to flexible GUI operations, but systematic evaluation of GUI–shortcut hybrid agents remains underexplored.
Approach: They propose a benchmark that evaluates GUI-shortcut hybrid agents with a specific focus on the mobile domain.
Outcome: MAS-Bench evaluates agent's ability to generate shortcuts by discovering and creating reusable, low-cost workflows.
Modular Sentence Encoders: Separating Language Specialization from Cross-Lingual Alignment (2025.acl-long)

Copied to clipboard

Challenge: Multilingual sentence encoders are often trained to map sentences from different languages into a shared semantic vector space . cross-lingual alignment training distorts optimal monolingual structure of semantic spaces of individual languages . a modular solution can be used for cross-linguistic tasks such as cross-language semantic similarity and zero-shot transfer .
Approach: They propose a modular training system that embeds sentences from different languages into a shared semantic vector space.
Outcome: The proposed solution achieves better performance across all tasks compared to monolithic models.
ESC-Eval: Evaluating Emotion Support Conversations in Large Language Models (2024.emnlp-main)

Copied to clipboard

Challenge: Emotion Support Conversation (ESC) is a crucial application for reducing stress and providing emotional guidance.
Approach: They re-organize 2,801 role-playing cards to define roles of role-players . they train a specific role- playing model called ESC-Role which behaves more like a confused person than GPT-4 .
Outcome: The proposed model behaves more like a confused person than GPT-4, and the model performs better than GPLs.
UKP-SQUARE: An Online Platform for Question Answering Research (2022.acl-demo)

Copied to clipboard

Challenge: Recent advances in NLP and information retrieval have given rise to a diverse set of question answering tasks that are of different formats and require different model architectures and setups.
Approach: They propose an extensible online QA platform that allows users to query and analyze a large collection of modern Skills via a user-friendly web interface and integrated behavioural tests.
Outcome: The proposed tool allows users to query and analyze a large collection of modern Skills via a user-friendly web interface and integrated behavioural tests.
Conflict-Aware Memory for Embodied Agents: Enhancing Vector Data Quality via Detection Rules (2026.acl-long)

Copied to clipboard

Challenge: Embodied agents have successfully leveraged large language models (LLMs) to better transform human instructions and images into executable task plans.
Approach: They propose Conflict Detection Rules to identify and manage data quality issues in vector knowledge bases and correct the index structure.
Outcome: Experimental results show that planners with Conflict Detection Rules exceed the basic LLM planner by 15.25% and 14.25% in grammatical accuracy (GA) and interpretation accuracy (IA) on average.
GameplayQA: A Benchmarking Framework for Decision-Dense POV-Synced Multi-Video Understanding of 3D Virtual Agents (2026.acl-long)

Copied to clipboard

Challenge: Multimodal Large Language Models are emerging as a backbone for autonomous agents in 3D environments.
Approach: They propose a framework for evaluating agentic-centric perception and reasoning through video understanding.
Outcome: The proposed framework evaluates agentic-centric perception and reasoning through video understanding.
Measuring Social Norms of Large Language Models (2024.findings-naacl)

Copied to clipboard

Challenge: Existing datasets that evaluate a general understanding of social science are inadequate to understand social norms.
Approach: They propose a multi-agent framework to improve large language models’ ability to understand social norms by comparing them to elementary students.
Outcome: The proposed framework improves large language models to be on par with humans.
AdaSent: Efficient Domain-Adapted Sentence Embeddings for Few-Shot Classification (2023.emnlp-main)

Copied to clipboard

Challenge: Recent work has found that few-shot sentence classification based on pre-trained Sentence Encoders (SEs) is efficient, robust, and effective.
Approach: They propose a domain-specialization approach that decouples SEPT from DAPT by training a SEPT adapter on a pre-trained PLM.
Outcome: The proposed approach matches or surpasses the performance of full SEPT on DAPT-ed PLMs while significantly reducing training costs.
Context-Driven Index Trimming: A Data Quality Perspective to Enhancing Precision of RALMs (2024.findings-emnlp)

Copied to clipboard

Challenge: Existing research often overlooks the data quality issues within retrieval results, often caused by inaccurate existing vector-distance-based retrieval methods.
Approach: They propose to use Context-Driven Index Trimming framework to capture and regulate consistency between retrieved contexts and modify indexes in the database.
Outcome: Experiments show that the proposed framework can improve answer quality by 3.75% on open-domain question-answering tasks.
UKP-SQuARE v3: A Platform for Multi-Agent QA Research (2023.acl-demo)

Copied to clipboard

Challenge: Current approaches to QA models are multi-dataset models, but combining expert agents can yield large performance gains over multi-agent models.
Approach: They extend an online platform for QA research to support three families of multi-agent systems: agent selection, early-fusion of agents, and late-fusion.
Outcome: The proposed model can be compared with multi-dataset models and achieve high inference speed and performance.
SpikeVoice: High-Quality Text-to-Speech Via Efficient Spiking Neural Network (2024.acl-long)

Copied to clipboard

Challenge: SpikeVoice performs high-quality Text-To-Speech (TTS) via SNN . major obstacle to using SNN for such generative tasks lies in the demand for models to grasp long-term dependencies.
Approach: They propose a brain-inspired Spiking Neural Network (SNN) which performs high-quality Text-To-Speech (TTS) via SNN and explores the potential of SNN to "speak".
Outcome: The proposed model achieves comparable results to Artificial Neural Networks (ANN) with only 10.5% energy consumption of ANN.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations