Papers by Junda Wang

24 papers
Reasoning-to-Defend: Safety-Aware Reasoning Can Defend Large Language Models from Jailbreaking (2025.emnlp-main)

Copied to clipboard

Challenge: Large Reasoning Models (LLMs) have demonstrated impressive performances across diverse domains, but how their safety benefits from enhanced reasoning capabilities against jailbreak queries remains unexplored.
Approach: They propose a safety-aware reasoning paradigm that integrates a pivot token-based safety-based reasoning mechanism into LLMs’ generation process.
Outcome: The proposed model improves the safety of large language models against jailbreak queries while minimizing attacks and maintaining the original performance.
MedQA-CS: Objective Structured Clinical Examination (OSCE)-Style Benchmark for Evaluating LLM Clinical Skills (2026.eacl-long)

Copied to clipboard

Challenge: Current clinical LLM benchmarks fail to evaluate advanced clinical skills in AI and large language models (LLMs).
Approach: They propose a framework to evaluate large language models (LLMs) using two instruction-following tasks designed to reflect real clinical scenarios.
Outcome: The proposed framework evaluates LLMs through two instruction-following tasks designed to reflect real clinical scenarios.
CachePrune: Teaching LLMs What Not to Follow via KV-Cache Editing (2026.acl-long)

Copied to clipboard

Challenge: Existing Large Language Models exhibit critical vulnerability to indirect prompt injection attacks, where instructions injected within in the prompt context can override the user's intent.
Approach: They propose a neural pruning algorithm that prunes neurons associated with instruction-following during KV cache encoding of the prompt context.
Outcome: The proposed approach significantly reduces the attack success rate while preserving the model's ability to follow user instructions.
Context-aware Information-theoretic Causal De-biasing for Interactive Sequence Labeling (2022.findings-emnlp)

Copied to clipboard

Challenge: Existing deep learning models for sequence labeling are expensive and time-consuming.
Approach: They propose an interactive sequence labeling that allows training directly with the user feedback . they identify context and feedback biases by formulating interactive sequence labels via a Structural Causal Model.
Outcome: The proposed approach can effectively alleviate the biases and can be learnt with the user feedback.
Mitigating Visual Knowledge Forgetting in MLLM Instruction-tuning via Modality-decoupled Gradient Descent (2025.findings-emnlp)

Copied to clipboard

Challenge: Existing fine-tuning and continual learning methods compress visual representations and emphasize task alignment over visual retention.
Approach: They propose a modality-decoupled gradient descent (MDGD) that regulates gradient updates to preserve effective rank of visual features and explicitly disentangles visual learning from task-specific alignment.
Outcome: The proposed model reduces visual forgetting and improves visual retention . it disentangles visual learning from task-specific alignment and preserves effective rank .
A Survey on LLM-based Conversational User Simulation (2026.eacl-long)

Copied to clipboard

Challenge: Recent advances in large language models (LLMs) have enabled high-fidelity generation of synthetic user conversation.
Approach: They propose a taxonomy covering user granularity and simulation objectives . they analyze core techniques and evaluation methodologies to help them understand the latest developments .
Outcome: The proposed model enables high-fidelity generation of synthetic user conversation.
From Selection to Generation: A Survey of LLM-based Active Learning (2025.acl-long)

Copied to clipboard

Challenge: Large Language Models (LLMs) have been used for selection and training of data for active learning.
Approach: They propose an intuitive taxonomy that categorizes LLM-based active learning techniques and discuss the transformative roles they can play in the active learning loop.
Outcome: The proposed model can generate entirely new data instances and provide more cost-effective annotations with fewer labeled data instances.
From Scores to Steps: Diagnosing and Improving LLM Performance in Evidence-Based Medical Calculations (2025.emnlp-main)

Copied to clipboard

Challenge: Existing benchmarks assess only the final answer with a wide numerical tolerance, overlooking systematic reasoning failures and potentially causing serious clinical misjudgments.
Approach: They propose a new step-by-step evaluation pipeline that assesses formula selection, entity extraction, and arithmetic computation.
Outcome: The proposed method improves the accuracy of large language models on medical benchmarks from 16.35% to 53.19%.
Few-Shot Dialogue Summarization via Skeleton-Assisted Prompt Transfer in Prompt Tuning (2024.eacl-long)

Copied to clipboard

Challenge: Existing prompt transfer techniques lack consideration for dialogue-specific information.
Approach: They propose a method which leverages skeleton generation as extra supervision that functions as a medium connecting the distinct source and target task.
Outcome: The proposed method significantly outperforms baselines on two dialogue summarization benchmarks.
Large Language Models and Causal Inference in Collaboration: A Comprehensive Survey (2025.findings-naacl)

Copied to clipboard

Challenge: Large Language Models (LLMs) have shown great potential to enhance Natural Language Processing (NLP) models in areas such as predictive accuracy, fairness, robustness, and explainability.
Approach: They evaluate or improve generative Large Language Models from a causal perspective in areas such as reasoning capacity, fairness and safety issues, explainability, and handling multimodality.
Outcome: The proposed models can be used to perform causal relationship discovery and causal effect estimation tasks.
AISFG: Abundant Information Slot Filling Generator (2022.naacl-main)

Copied to clipboard

Challenge: Existing approaches to zero/few-shot slot filling focus on slot descriptions and examples . AISFG model is based on domain-specific labels, which is not capable of transferring to new domains with little or no data.
Approach: They propose a model with a query template that incorporates domain descriptions, slot descriptions, and examples with context.
Outcome: Experimental results show that the proposed model outperforms state-of-the-art approaches in zero/few-shot slot filling task.
Personalized Federated Learning for Text Classification with Gradient-Free Prompt Tuning (2024.findings-naacl)

Copied to clipboard

Challenge: Pretrained language models (PLMs) are used for personalized federated learning . communication costs are high with large PLMs, and local training is expensive .
Approach: They propose a framework for federated learning with pretrained language models . they propose 'discrete local search' and compression mechanism for local training .
Outcome: The proposed framework achieves superior performance compared with baselines.
NoteChat: A Dataset of Synthetic Patient-Physician Conversations Conditioned on Clinical Notes (2024.findings-acl)

Copied to clipboard

Challenge: NoteChat is a cooperative multi-agent framework for generating patient-physician dialogues . evaluator finds it outperforms state-of-the-art models for generating clinical notes . clinical documentation is largely done by physicians at both steps .
Approach: They propose a cooperative multi-agent framework leveraging Large Language Models to generate patient-physician dialogues.
Outcome: The proposed framework outperforms state-of-the-art models for generating clinical notes . it can engage patients directly and help clinical documentation, a leading cause of physician burnout .
DiffusionAttacker: Diffusion-Driven Prompt Manipulation for LLM Jailbreak (2025.emnlp-main)

Copied to clipboard

Challenge: Large Language Models (LLMs) are susceptible to generating harmful content when prompted with carefully crafted inputs, a vulnerability known as LLM jailbreaking.
Approach: They propose an end-to-end generative approach for jailbreak rewriting inspired by diffusion models that uses a sequence-tosequence (seq2sequ) diffusion model as a generator, conditioning on the original prompt and guiding the denoising process with a novel attack loss.
Outcome: Experiments on Advbench and Harmbench show that the proposed method outperforms autoregressive jailbreak models across evaluation metrics including ASR, fluency, diversity and diversity.
RARE: Retrieval-Augmented Reasoning Enhancement for Large Language Models (2025.acl-long)

Copied to clipboard

Challenge: Existing work aims to improve reasoning accuracy and factual integrity across large language models for knowledge-intensive tasks such as medical and commonsense reasoning.
Approach: They propose a versatile extension to the mutual reasoning framework (rStar) that enhances reasoning accuracy and factual integrity across large language models.
Outcome: The proposed extension to the mutual reasoning framework improves reasoning accuracy and factual integrity across large language models for complex, knowledge-intensive tasks.
CoMMIT: Coordinated Multimodal Instruction Tuning (2025.emnlp-main)

Copied to clipboard

Challenge: et al., 2024) show that multimodal instruction tuning is more effective than baselines.
Approach: They propose a multimodal balance coefficient that enables quantitative measurement of the balance of learning . they propose auxiliary regularization on the gradient to promote updating with larger step sizes .
Outcome: The proposed method is more effective than baselines in MLLM instruction tuning.
LEAF: Learning and Evaluation Augmented by Fact-Checking to Improve Factualness in Large Language Models (2025.emnlp-industry)

Copied to clipboard

Challenge: Large language models (LLMs) struggle with factual accuracy in knowledge-intensive domains like healthcare.
Approach: They propose a framework for improving LLM factuality in medical question answering . RAFE, Fact-Check-then-RAG and Learning from Fact Check are components .
Outcome: Experimental results show that LEAF outperforms Factcheck-GPT in detecting inaccuracies and corrects errors without labeling . the framework provides a scalable solution for industrial applications requiring high factuality scores.
InstructGraph: Boosting Large Language Models via Graph-centric Instruction Tuning and Preference Alignment (2024.findings-acl)

Copied to clipboard

Challenge: Existing large language models (LLMs) can solve graph reasoning and generation tasks with parameter updates without sacrificing performance.
Approach: They propose a structured format verbalizer to unify all graph data into a universal code-like format, which can simply represent the graph without any external graph-specific encoders.
Outcome: The proposed framework outperforms GPT-4 and LLaMA2 in graph reasoning and generation tasks by more than 13% and 38%, respectively.
GUI Agents: A Survey (2025.findings-acl)

Copied to clipboard

Challenge: Large Foundation Models (LFMs) have transformed the landscape of AI research and day-to-day life.
Approach: They propose a framework that delineates GUI agents' perception, reasoning, planning, and acting capabilities.
Outcome: The proposed framework delineates their perception, reasoning, planning, and acting capabilities.
Unleashing the Native Recommendation Potential: LLM-Based Generative Recommendation via Structured Term Identifiers (2026.findings-acl)

Copied to clipboard

Challenge: Existing methods for constructing item identifiers face bottlenecks due to their large output space and expensive vocabulary expansion and alignment training.
Approach: They propose to use Large Language Models to develop general-purpose, semantically-aware recommender systems that can be generalized and reusable.
Outcome: Experiments on real-world datasets show that GRAM outperforms baselines and significantly outperformed baselines.
DeCoT: Debiasing Chain-of-Thought for Knowledge-Intensive Tasks in Large Language Models via Causal Intervention (2024.acl-long)

Copied to clipboard

Challenge: In large language models, external knowledge is required to augment their internal knowledge through prompts, but this does not guarantee that LLMs can identify and use relevant information in the prompts to conduct chain-of-thought reasoning.
Approach: They propose a structural causal model to formally explain the internal knowledge bias of large language models (LLMs) they review the chain-of-thought (CoT) prompting from a causal perspective and find that biased information from pretrained models can impair LLMs’ reasoning abilities.
Outcome: The proposed model enables more accurate CoT reasoning and enhances LLM generation on knowledge-intensive tasks.
Federated Domain Adaptation for Named Entity Recognition via Distilling with Heterogeneous Tag Sets (2023.findings-acl)

Copied to clipboard

Challenge: a federated domain adaptation approach is used to learn with NER datasets from multiple platforms while not violating data privacy.
Approach: They propose to use a distillation approach to facilitate knowledge transfer across platforms.
Outcome: The proposed model performs better in the clinic domain.
SceneAlign: Aligning Multimodal Reasoning to Scene Graphs in Complex Visual Scenes (2026.acl-long)

Copied to clipboard

Challenge: Existing preference-based approaches fail to address this challenge by exploiting language priors to bypass visual grounding.
Approach: They propose a framework that leverages scene graphs as structured visual information to perform controllable structural interventions.
Outcome: The proposed framework improves answer accuracy and reasoning faithfulness across seven visual reasoning benchmarks.
Knowledge-Aware Query Expansion with Large Language Models for Textual and Relational Retrieval (2025.naacl-long)

Copied to clipboard

Challenge: Existing methods to generate query expansions focus on enhancing textual similarities between search queries and document corpus, overlooking document relations.
Approach: They propose a knowledge-aware query expansion framework augmenting LLMs with structured document relations from knowledge graph (KG) they leverage document texts as rich KG node representations and use document-based relation filtering for their method.
Outcome: The proposed framework augments LLMs with structured document relations from knowledge graph (KG) Extensive experiments on three datasets of diverse domains show the advantages compared against state-of-the-art methods on textual and relational semi-structured retrieval.

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