Papers by Yuan Zeng

25 papers
CRITICTOOL: Evaluating Self-Critique Capabilities of Large Language Models in Tool-Calling Error Scenarios (2025.emnlp-main)

Copied to clipboard

Challenge: a number of tools are used to perform complex tasks, but the tool utilization process can cause errors.
Approach: They propose a critique evaluation benchmark for tool learning that analyzes function-calling errors on tool evaluation benchmarks.
Outcome: The proposed critique evaluation benchmark holds diverse tool-use errors with varying complexities, which better reflects real-world scenarios.
AdDriftBench: A Benchmark for Detecting Data Drift and Label Drift in Short Video Advertising (2025.findings-emnlp)

Copied to clipboard

Challenge: Short video advertising scenarios present unique challenges due to data drift (DD) and label drift (LD).
Approach: They propose to use data drift and label drift to evaluate models under rapidly shifting content distributions and labeling scenarios to assess their generalization capabilities.
Outcome: The proposed model performs moderately in short video advertising contexts, particularly in handling fine-grained semantics and adapting to shifting instructions.
Modeling Conceptual Attribute Likeness and Domain Inconsistency for Metaphor Detection (2023.emnlp-main)

Copied to clipboard

Challenge: Metaphor detection aims to distinguish between metaphorical and literal expressions in text.
Approach: They propose an attribute likeness and domain inconsistency learning framework for wordpair metaphor detection based on conceptual metaphor theory . they model attribute likeity with an attribute siamese network and devise a domain contrastive learning strategy to learn semantic inconsistentness of concepts in source and target domains .
Outcome: The proposed framework outperforms existing word-pair and token-level methods on four datasets.
DRAMA: Dynamic Multi-Granularity Graph Estimate Retrieval over Tabular and Textual Question Answering (2024.lrec-main)

Copied to clipboard

Challenge: TableTextQA tasks require tabular and textual data, gaining increasing attention . however, row-based approaches suffer from limitations such as lack of interaction between rows .
Approach: They propose a method that incorporates an interaction mechanism among multiple rows . Empirical results demonstrate that the proposed method is effective .
Outcome: Empirical results show that the proposed model is effective on tabFact and HybridQA datasets.
Data Efficient RLVR via Off-Policy Influence Guidance (2026.acl-long)

Copied to clipboard

Challenge: Existing data selection methods for RLVR are heuristic-based, lacking theoretical guarantees and generalizability.
Approach: They propose an off-policy influence estimation method that approximates data influence using offline trajectories.
Outcome: The proposed method reduces the computational cost of policy rollouts and improves storage and computation efficiency.
i-Code V2: An Autoregressive Generation Framework over Vision, Language, and Speech Data (2024.findings-naacl)

Copied to clipboard

Challenge: i-Code V2 is one of the first models capable of generating natural language from any combination of Vision, Language, and Speech data.
Approach: They propose to create a model that can generate natural language from any combination of Vision, Language, and Speech data.
Outcome: i-Code V2 matches or outperforms state-of-the-art single- and dual-modality baselines on 7 multimodal tasks.
Beyond Black-Box Interventions: Latent Probing for Faithful Retrieval-Augmented Generation (2026.findings-acl)

Copied to clipboard

Challenge: Existing approaches to improve contextual faithfulness treat the LLM as a black box, generating responses that are inconsistent with the provided context.
Approach: They propose a framework for faithful RAG that operates in three stages: (i) fine-grained knowledge pruning to filter irrelevant context, (ii) latent conflict probing to identify hard conflicts in the model’s latent space, and (iv) conflict-aware attention to modulate attention heads toward faithful context integration.
Outcome: Experiments show that ProbeRAG significantly improves both accuracy and contextual faithfulness.
Whispers that Shake Foundations: Analyzing and Mitigating False Premise Hallucinations in Large Language Models (2024.emnlp-main)

Copied to clipboard

Challenge: Large language models generate hallucinated text when confronted with false premise questions . authors propose a method to mitigate false premises hallucinosity .
Approach: They propose a method to constrain false premise attention heads during the model inference process.
Outcome: The proposed method improves performance by constraining false premise attention heads . it yields a notable increase of nearly 20% of model performance .
Dr. Assistant: Enhancing Clinical Diagnostic Inquiry via Structured Diagnostic Reasoning Data and Reinforcement Learning (2026.findings-acl)

Copied to clipboard

Challenge: Clinical Decision Support Systems (CDSSs) provide reasoning and inquiry guidance for physicians, yet they face high maintenance costs and low generalization capability.
Approach: They propose a clinical diagnostic model with clinical reasoning and inquiry skills, the Dr. Assistant, and a pipeline to capture abstract reasoning logic.
Outcome: The proposed model outperforms open-source models and achieves competitive performance to closed-source model.
History-Aware Hierarchical Transformer for Multi-session Open-domain Dialogue System (2022.findings-emnlp)

Copied to clipboard

Challenge: Existing open-domain dialogue systems conduct one-session conversations, but multi-session MSCs are under-investigated.
Approach: They propose a History-Aware Hierarchical Transformer for multi-session open-domain dialogue . they propose to encode history conversations into a history memory and leverage historical information to generate well-informed responses.
Outcome: The proposed model outperforms baseline models on a large-scale MSC dataset.
BizCompass: Benchmarking the Reasoning Capabilities of LLMs in Business Knowledge and Applications (2026.findings-acl)

Copied to clipboard

Challenge: Existing benchmarks focus on narrow tasks and leave a fundamental question unanswered . Existing models only focus on specific tasks, requiring rigorous reasoning and knowledge .
Approach: They propose a benchmark to connect theoretical foundations with practical business knowledge and applications.
Outcome: The benchmark systematically evaluates both open-source and commercial LLMs . it reveals how theoretical knowledge translates into practical performance in business .
OpenAttack: An Open-source Textual Adversarial Attack Toolkit (2021.acl-demo)

Copied to clipboard

Challenge: Various attack models are distinct and implemented with different programming frameworks and settings, which hinders quick utilization and fair comparison of attack models.
Approach: They propose an open-source textual adversarial attack toolkit to solve these issues by combining 15 typical attack models into one toolkit.
Outcome: The proposed toolkit supports all attack types, multilinguality, and parallel processing.
Multi-View Incongruity Learning for Multimodal Sarcasm Detection (2025.coling-main)

Copied to clipboard

Challenge: Existing methods for multimodal sarcasm detection rely on spurious correlations, demonstrating poor generalizability beyond training environments.
Approach: They propose a method that integrates multimodal incongruities via contrastive learning for multimodal sarcasm detection by using three views to drive multi-view learning.
Outcome: The proposed method outperforms existing methods on benchmark datasets and shows that it is more generalizable than existing methods.
Protecting Privacy in Multimodal Large Language Models with MLLMU-Bench (2025.naacl-long)

Copied to clipboard

Challenge: Large Language Models (LLMs) and Multimodal Large Language models (MLLMs) trained on vast web corpora can memorize and disclose individuals’ confidential and private data, raising legal and ethical concerns.
Approach: They propose a benchmark to assess unlearning algorithms from multiple perspectives and provide a baseline for existing generative models.
Outcome: The proposed benchmark consists of 500 fictitious profiles and 153 profiles for public celebrities, evaluated from both multimodal (image+text) and unimodal (text) perspectives.
Grounded Multimodal Procedural Entity Recognition for Procedural Documents: A New Dataset and Baseline (2024.lrec-main)

Copied to clipboard

Challenge: Existing methods to extract procedural knowledge from documents focus on text-only settings, which is insufficient for entity disambiguation.
Approach: They propose a model to detect the entity and the corresponding bounding box groundings in images.
Outcome: The proposed model detects the entity and the corresponding bounding box groundings in image (i.e., visual entities) it is based on a dataset of a WikiHow 1 and EHow 2 document and the results are compared with existing models.
Dirichlet Latent Variable Hierarchical Recurrent Encoder-Decoder in Dialogue Generation (D19-1)

Copied to clipboard

Challenge: Existing work assumes the Gaussian priors of the latent variable, which are incapable of representing complex latent variables effectively.
Approach: They propose to use the Dirichlet distribution with flexible structures to characterize latent variables in place of the Gaussian priors.
Outcome: The proposed model outperforms existing models on the dialogue generation task.
RISER: Orchestrating Latent Reasoning Skills for Adaptive Activation Steering (2026.findings-acl)

Copied to clipboard

Challenge: Existing methods for domain-specific reasoning with large language models require updating parameter updates.
Approach: They propose a plug-and-play intervention framework that adaptively steers LLM reasoning in activation space.
Outcome: The proposed framework achieves zero-shot accuracy improvements of 3.4–6.5% over the base model while outperforming chain-of-thought-style reasoning with 2–3 higher token efficiency and robust accuracy gains.
KGA: A General Machine Unlearning Framework Based on Knowledge Gap Alignment (2023.acl-long)

Copied to clipboard

Challenge: Recent legislation of the "right to be forgotten" has led to the interest in machine unlearning . MU can be used to forget specific training instances as if they have never existed .
Approach: They propose a general unlearning framework called KGA to induce forgetfulness . they propose several unlearning evaluation metrics with pertinence .
Outcome: The proposed framework improves on large-scale datasets and provides insight into unlearning for NLP tasks.
Global Context or Local Detail? Adaptive Visual Grounding for Hallucination Mitigation (2026.findings-acl)

Copied to clipboard

Challenge: Large vision–language models suffer from object-existence hallucinations when multi-step deliberation decouples from visual evidence.
Approach: They propose a framework that allocates visual computation by uncertainty . they propose highlighting retains global context, while selective zoom-in performs local verification.
Outcome: The proposed framework reduces the complexity of multimodal reasoning by minimizing the operator trade-off.
Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention (2025.acl-long)

Copied to clipboard

Challenge: Long-context modeling is crucial for next-generation language models, but high computational cost of standard attention mechanisms poses significant computational challenges.
Approach: They propose a natively trained Sparse Attention mechanism that integrates algorithms with hardware-aligned optimizations to achieve efficient long-context modeling.
Outcome: The proposed model maintains or exceeds Full Attention models across general benchmarks, long-context tasks, and instruction-based reasoning.
Detecting AI-Generated Content on Social Media with Multi-modal Language Models (2026.acl-industry)

Copied to clipboard

Challenge: Existing methods for AI-generated content detection face poor generalization to newer models, reliance on single modalities, and lack of interpretable explanations.
Approach: They propose a model that curates diverse social media data and trains a vision-language model for detection and explanation.
Outcome: The proposed model achieves state-of-the-art detection performance on public benchmarks and observes positive downstream impacts on user engagement.
CharTide: Data-Centric Chart-to-Code Generation via Tri-Perspective Tuning and Inquiry-Driven Evolution (2026.acl-long)

Copied to clipboard

Challenge: Existing approaches to chart-to-code generation are constrained by data-centric limitations . authors present a new framework that redesigns both training and alignment data .
Approach: They propose a data-centric framework that redesigns both training and alignment data for chart-to-code generation.
Outcome: The proposed framework outperforms open-source baselines and is competitive with GPT-5.
i-Code Studio: A Configurable and Composable Framework for Integrative AI (2024.emnlp-demo)

Copied to clipboard

Challenge: Existing frameworks for Integrative AI lack flexibility and composability to handle multimodal tasks.
Approach: They propose a configurable framework for Integrative AI that orchestrates multiple pre-trained models to conduct complex multimodal tasks.
Outcome: The proposed framework achieves impressive results on zero-shot multimodal tasks . it can communicate and personalize for users, and it can be used in a multimodal agent .
An LLM-Enabled Knowledge Elicitation and Retrieval Framework for Zero-Shot Cross-Lingual Stance Identification (2024.findings-emnlp)

Copied to clipboard

Challenge: Existing research is conducted in monolingual setting on English datasets, whereas in other low-resource languages, it lacks sufficient data for training quality stance detection models.
Approach: They propose a knowledge elicitation and retrieval framework that leverages the capability of large language models for stance knowledge acquisition and matches the target language input to the most relevant stance information.
Outcome: The proposed framework improves on multilingual datasets and competitive baselines.
MMEvalPro: Calibrating Multimodal Benchmarks Towards Trustworthy and Efficient Evaluation (2025.naacl-long)

Copied to clipboard

Challenge: Large Multimodal Models (LMMs) exhibit impressive cross-modal understanding and reasoning abilities, but many benchmarks suffer from systematic biases.
Approach: They propose a benchmark to avoid Type-I errors by creating one perception question and one knowledge anchor question through a meticulous annotation process.
Outcome: The proposed benchmark avoids Type-I errors while maintaining reliability of MCQ evaluations.

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