Papers by Mukai Li

14 papers
Verified Critical Step Optimization for LLM Agents (2026.findings-acl)

Copied to clipboard

Challenge: Critical Step Optimization (CSO) focuses preference learning on verified critical steps where alternative actions demonstrably flip task outcomes from failure to success.
Approach: They propose a method which focuses preference learning on verified critical steps where alternative actions demonstrably flip task outcomes from failure to success.
Outcome: The proposed method outperforms the existing methods on GAIA-Text-103 and XBench-DeepSearch while requiring supervision at only 16% of trajectory steps.
VLFeedback: A Large-Scale AI Feedback Dataset for Large Vision-Language Models Alignment (2024.emnlp-main)

Copied to clipboard

Challenge: Large vision-language models (LVLMs) are evolving rapidly and require data with human supervision to achieve better alignment.
Approach: They introduce VLFeedback, the first large-scale vision-language feedback dataset . they train Silkie, an LVLM fine-tuned via direct preference optimization .
Outcome: The proposed model outperforms its base model in helpfulness, visual faithfulness, and safety metrics and exhibits enhanced resilience against red-teaming attacks.
ONION: A Simple and Effective Defense Against Textual Backdoor Attacks (2021.emnlp-main)

Copied to clipboard

Challenge: Backdoor attacks can manipulate the output of deep neural networks and possess high insidiousness.
Approach: They propose a textual backdoor defense based on outlier word detection that can handle all the textual attacks.
Outcome: The proposed method can handle all the textual backdoor attack situations.
Design Choices for Extending the Context Length of Visual Language Models (2025.acl-long)

Copied to clipboard

Challenge: Existing open-source Visual Language Models lack systematic exploration into extending their context length, and commercial models often provide limited details.
Approach: They propose to extend Visual Language Models (VLMs) to 128K lengths and open-source the code, data, and models.
Outcome: The proposed model is based on the Qwen-VL series model and is competitive with commercial model GPT-4V.
MMEKG: Multi-modal Event Knowledge Graph towards Universal Representation across Modalities (2022.acl-demo)

Copied to clipboard

Challenge: Recent Knowledge Graphs (KGs) store billions of world facts in a directed graph, but expression ability of such entity-centric KGs is limited.
Approach: They propose a large-scale multi-modal event knowledge graph named MMEKG that unifies different modalities of knowledge via events.
Outcome: The proposed system unifies different modalities of knowledge via events, which complement and disambiguate each other.
ERGO: Event Relational Graph Transformer for Document-level Event Causality Identification (2022.coling-1)

Copied to clipboard

Challenge: Existing methods to identify event-event causal relations in a document are noisy and require heuristic rules or external tools.
Approach: They propose a document-level event-event causality identification framework that uses heuristic rules to design edges between events.
Outcome: The proposed framework outperforms existing state-of-the-art methods on two benchmark datasets.
Hidden Killer: Invisible Textual Backdoor Attacks with Syntactic Trigger (2021.acl-long)

Copied to clipboard

Challenge: Existing methods for textual backdoor attacks insert additional contents into normal samples as triggers, causing detection and blocking of backdoors.
Approach: They propose to use syntactic structure as trigger in textual backdoor attacks . they propose to achieve similar attack performance but have higher invisibility .
Outcome: The proposed method achieves almost 100% success rate but has higher invisibility and stronger resistance to defenses than the insertion-based methods.
Mind the Style of Text! Adversarial and Backdoor Attacks Based on Text Style Transfer (2021.emnlp-main)

Copied to clipboard

Challenge: Experimental results show that popular NLP models are vulnerable to both adversarial and backdoor attacks based on text style transfer.
Approach: They propose to conduct adversarial and backdoor attacks based on text style transfer . the authors propose to use text style to alter the style of a sentence .
Outcome: The proposed methods show that popular models are vulnerable to both attacks based on text style transfer . the results show that the proposed methods perform better than baselines in many aspects .
Red Teaming Visual Language Models (2024.findings-acl)

Copied to clipboard

Challenge: VLMs (Vision-Language Models) can be induced to generate harmful or inaccurate content through specific test cases.
Approach: They propose a red teaming dataset which encompasses 12 subtasks under 4 primary aspects (faithfulness, privacy, safety, fairness) this dataset is the first to benchmark current VLMs in terms of these 4 aspects .
Outcome: The proposed dataset shows that 10 open-source VLMs struggle with red teaming in different degrees and have up to 31% performance gap with GPT-4V.
OS-Sentinel: Towards Safety-Enhanced Mobile GUI Agents via Hybrid Validation in Realistic Workflows (2026.acl-long)

Copied to clipboard

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.
EconProver: Towards More Economical Test-Time Scaling for Automated Theorem Proving (2026.acl-long)

Copied to clipboard

Challenge: Large Language Models (LLMs) have recently advanced the field of Automated Theorem Proving (ATP) Existing cost analyses regulate only the number of sampling passes, ignoring the substantial disparities in sampling costs.
Approach: They propose to integrate two complementary methods into a unified EconRL pipeline to increase pass rates under constrained sampling passes.
Outcome: The proposed method reduces token usage and sample passes while maintaining the original performance.
Prompt for Extraction? PAIE: Prompting Argument Interaction for Event Argument Extraction (2022.acl-long)

Copied to clipboard

Challenge: Using a prompt-based model, we find that event argument extraction is efficient and generalized well to few-shot settings.
Approach: They propose a model PAIE for event argument extraction using prompt tuning for extractive objectives.
Outcome: The proposed model can extract arguments with the same role instead of heuristic threshold tuning.
DiffuSeq-v2: Bridging Discrete and Continuous Text Spaces for Accelerated Seq2Seq Diffusion Models (2023.findings-emnlp)

Copied to clipboard

Challenge: Existing approaches to text generation use discrete text within a continuous diffusion space, which incurs substantial computational overhead during training and results in slower sampling speeds.
Approach: They propose a soft absorbing state that facilitates diffusion models in learning to reconstruct discrete mutations based on the underlying Gaussian space.
Outcome: The proposed method accelerates training convergence by 4x and generates samples of similar quality 800x faster, rendering it closer to practical application.
AlphaQT-Bench: Diagnosing the Gap between Financial Code Generation and Quantitative Reasoning in LLMs (2026.findings-acl)

Copied to clipboard

Challenge: Existing benchmarks rely on outcome-driven metrics such as profitability and look-ahead bias.
Approach: They propose a diagnostic benchmark for instruction-grounded financial code generation under strict semantic and temporal constraints.
Outcome: The proposed benchmarks show that the models fail under causal, structural, or functional constraints.

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