Papers by Di Huang

32 papers
Know-MRI: A Knowledge Mechanisms Revealer&Interpreter for Large Language Models (2025.acl-demo)

Copied to clipboard

Challenge: Existing interpretation methods only support tasks with specific inputs, limiting their practical applications.
Approach: They propose an extensible module that matches different input data with interpretation methods and consolidates the interpreting outputs.
Outcome: The proposed module can match different input data with interpretation methods and consolidate the interpreting outputs.
Entity-aware Image Caption Generation (D18-1)

Copied to clipboard

Challenge: Existing image captioning approaches generate generic descriptions of visual content and ignore background information.
Approach: They propose a task which generates informative image captions using images and hashtags as input.
Outcome: The proposed model outperforms unimodal baselines significantly with evaluation metrics on a dataset from Flickr.
ShieldLM: Empowering LLMs as Aligned, Customizable and Explainable Safety Detectors (2024.findings-emnlp)

Copied to clipboard

Challenge: Existing tools for detecting safety issues in LLMs are expensive and inefficient.
Approach: They propose an LLM-based safety detector which annotates the safety of queries and provides explanations for its decisions.
Outcome: The proposed detector outperforms baselines on four sets of query-response pairs and is effective as a safety evaluator for advanced LLMs.
QiMeng-PRepair: Precise Code Repair via Edit-Aware Reward Optimization (2026.acl-long)

Copied to clipboard

Challenge: Existing approaches to program repair are based on correctness alone.
Approach: They propose a framework that mitigates over-editing and improves repair accuracy by generating buggy programs and re-edits.
Outcome: The proposed framework improves repair precision by 31.4% under fix1@1, a metric that considers repair correctness and extent, and significantly increases decoding throughput when combined with speculative editing.
Evaluating Readability and Faithfulness of Concept-based Explanations (2024.emnlp-main)

Copied to clipboard

Challenge: Existing methods for evaluating concepts from different perspectives lack a unified formalization.
Approach: They propose a formal definition of concepts generalizing to diverse concept-based explanations’ settings and apply it to other types of explanations or tasks.
Outcome: Extensive experimental analysis was carried out to determine the evaluation measures for explanation evaluation measures.
Private Language Models via Truncated Laplacian Mechanism (2024.emnlp-main)

Copied to clipboard

Challenge: Existing methods for word embedding are prone to privacy leakage, resulting in weaker relaxations of DP that are inferior to the canonical DP in terms of privacy strength.
Approach: They propose a method for private word embedding that uses a non-trivial extension of the truncated Laplacian mechanism and propose to test its effectiveness.
Outcome: The proposed method has lower variance compared to the previous methods.
Asynchronous Deep Interaction Network for Natural Language Inference (D19-1)

Copied to clipboard

Challenge: Existing methods have framed the reasoning problem as a semantic matching task.
Approach: They propose an asynchronous deep interaction network (ADIN) to deconstruct the reasoning process and implement asynchron and multi-step reasoning.
Outcome: The proposed model outperforms strong baselines on three popular benchmarks: SNLI, MultiNLI, and SciTail.
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.
Tell Me What You Don’t Know: Enhancing Refusal Capabilities of Role-Playing Agents via Representation Space Analysis and Editing (2025.findings-acl)

Copied to clipboard

Challenge: Role-playing Agents (RPAs) struggle to recognize and respond to hard queries that conflict with their role-play knowledge.
Approach: They propose a lightweight representation editing approach that conveniently shifts conflicting requests to the rejection region, thereby enhancing the model’s refusal accuracy.
Outcome: The proposed model improves RPAs’ refusal ability of conflicting requests while maintaining their general role-playing capabilities.
Revisiting Entropy Regularization: Adaptive Coefficient Unlocks Its Potential for LLM Reinforcement Learning (2026.findings-acl)

Copied to clipboard

Challenge: Reasoning ability is a defining capability of Large Language Models (LLMs), but RLVR training suffers from policy entropy collapse, hindering exploration and limiting reasoning performance.
Approach: They propose a framework that dynamically balances exploration and exploitation via three components: difficulty-aware coefficient allocation, initial-anchored target entropy, and dynamic global coefficient adjustment.
Outcome: The proposed framework outperforms baselines on multiple mathematical reasoning benchmarks.
Reinforcement Learning on Pre-Training Data (2026.acl-long)

Copied to clipboard

Challenge: Recent progress in large language models is driven by scaling of training compute through pre-training with nexttoken prediction (NTP) or post-training (RL) Pre-training using NTP enables models to acquire extensive knowledge and skills from general data, but it suffers from data inefficiency and catastrophic forgetting in continual learning settings.
Approach: They propose to scale training compute through pre-training with next-token prediction (NTP) or post-training by scaling reinforcement learning (RL) to improve learning from general data.
Outcome: Experiments on multiple benchmarks and models show that the proposed approach improves continual pre-training and provides a strong foundation for post-training on Qwen3-8B-Base.
Speculating LLMs’ Chinese Training Data Pollution from Their Tokens (2025.emnlp-main)

Copied to clipboard

Challenge: Experiments on GPT and other 23 LLMs indicate that tokens widely exist while GPT’s vocabulary behaves the worst: more than 23% long Chinese tokens (i.e., a token with more than two Chinese characters) are either porn or online gambling.
Approach: They propose to locate Polluted Chinese (PoC) tokens in LLMs and build a PoC token detector to label them in vocabularies by considering each token’s semantics and related contents from the search engines.
Outcome: The proposed method predicts that the ratio of “*” related webpages in GPT-4o's training data is around 0.5%.
CQG: A Simple and Effective Controlled Generation Framework for Multi-hop Question Generation (2022.acl-long)

Copied to clipboard

Challenge: Current models can not ensure the complexity of generated questions, so they may generate shallow questions that can be answered without multi-hop reasoning.
Approach: They propose a controlled framework to generate multi-hop questions that contain key entities in multi- hop reasoning chains and a novel Transformer-based decoder to guarantee that key entities appear in the questions.
Outcome: The proposed model outperforms the state-of-the-art model 25% on HotpotQA.
Transferring from Formal Newswire Domain with Hypernet for Twitter POS Tagging (D18-1)

Copied to clipboard

Challenge: Existing POS tagging methods for Twitter use labeled newswire text . however, Twitter users tend to mimic formal media expressions and develop linguistically informal styles.
Approach: They propose to use newswire text to learn POS tagging for Twitter while twitter users are developing linguistically informal styles.
Outcome: The proposed method achieves better performance than state-of-the-art methods on three different datasets.
The Art of SOCRATIC QUESTIONING: Recursive Thinking with Large Language Models (2023.emnlp-main)

Copied to clipboard

Challenge: Chain-of-Thought (CoT) prompting relies on the initial decisions, causing errors in early steps to accumulate and impact the final answers.
Approach: They propose a divide-and-conquer style algorithm that leverages large language models to raise and answer sub-questions until collecting enough information to tackle the original one.
Outcome: The proposed algorithm is more robust to errors and errors than CoT prompting and Tree-of-Thought prompting methods.
Understanding the Dark Side of LLMs’ Intrinsic Self-Correction (2025.acl-long)

Copied to clipboard

Challenge: Recent studies show that LLMs’ intrinsic self-correction fails without oracle labels as feedback.
Approach: They propose to use one simple task and three complex tasks with state-of-the-art LLMs like ChatGPT, Llama, and DeepSeek to interpret LLM's intrinsic self-correction.
Outcome: The proposed methods reveal the dark side of LLMs’ intrinsic self-correction for different tasks, especially for those failure cases.
Robust Lottery Tickets for Pre-trained Language Models (2022.acl-long)

Copied to clipboard

Challenge: Recent studies have shown that pre-trained language models contain smaller matching subnetworks that are not robust to adversarial examples.
Approach: They propose a method to find robust tickets hidden in pre-trained language models by learning binary weight masks and an adversarial loss objective to guide the search.
Outcome: The proposed method improves on previous work on adversarial robustness evaluation.
Towards Effective and Efficient Continual Pre-training of Large Language Models (2025.acl-long)

Copied to clipboard

Challenge: Continual pre-training (CPT) has been an important approach for adapting language models to specific domains or tasks.
Approach: They propose a Continual pre-training method that can greatly improve Chinese language ability and scientific reasoning ability of LLMs.
Outcome: The proposed method can greatly improve Chinese language ability and scientific reasoning ability of LLMs.
Improving Translation Quality Estimation with Bias Mitigation (2023.acl-long)

Copied to clipboard

Challenge: State-of-the-art translation Quality Estimation models are biased, relying on monolingual features while ignoring the bilingual semantic alignment.
Approach: They propose a method to mitigate the bias of translation quality estimation models by contrastive learning between clean and noisy sentence pairs.
Outcome: The proposed method improves the estimation performance while mitigating the bias.
KMatrix-2: A Comprehensive Heterogeneous Knowledge Collaborative Enhancement Toolkit for Large Language Model (2025.emnlp-demos)

Copied to clipboard

Challenge: Existing studies on K-LLMs systems focus on declarative knowledge and procedural knowledge (rules) .
Approach: They propose to build a toolkit that supports comprehensive heterogeneous knowledge collaborative enhancement for Large Language Models (LLMs).
Outcome: The proposed toolkit provides unified knowledge integration and joint knowledge retrieval methods to achieve more comprehensive heterogeneous knowledge collaborative enhancement.
LiCoMemory: Lightweight and Cognitive Agentic Memory for Efficient Long-Term Reasoning (2026.findings-acl)

Copied to clipboard

Challenge: Large Language Models are constrained by limited context windows and lack of persistent memory . recent efforts address these limitations via external memory architectures .
Approach: They propose an end-to-end agentic memory framework for real-time updating and retrieval that integrates hierarchical and temporal indexing layers.
Outcome: The proposed framework outperforms established benchmarks in temporal reasoning, multi-session consistency, and retrieval efficiency.
Breaking the Stage Barrier: A Novel Single-Stage Approach to Long Context Extension for Large Language Models (2025.coling-main)

Copied to clipboard

Challenge: Recent studies show that Large language models struggle with handling long token sequences due to limited training context size.
Approach: They propose a single-stage continual pretraining method to equip LLMs with long context modeling capabilities.
Outcome: The proposed method outperforms existing methods on 4 language modeling benchmarks.
Towards Stable and Effective Reinforcement Learning for Mixture-of-Experts (2026.acl-long)

Copied to clipboard

Challenge: Reinforcement learning with verifiable rewards (RLVR) training with Mixture-of-Experts policies remains fragile and prone to reward collapse.
Approach: They propose a router shift-based policy optimization method that computes a per-token router-shift ratio conditioned on the previously activated experts and applies stop-gradient and a lower-bound floor.
Outcome: The proposed method achieves better performance and greater stability than previous methods.
Vision-Flan: Scaling Human-Labeled Tasks in Visual Instruction Tuning (2024.findings-acl)

Copied to clipboard

Challenge: Recent vision-language models (VLMs) have shown impressive capabilities as general visual assistants, but there are two challenges to their performance: (1) lacking task diversity in pretraining and visual instruction tuning; (2) annotation error and bias in GPT-4 synthesized instruction tuning data.
Approach: They propose a two-stage instruction tuning framework that fine tunes VLMs firstly and further tuned on GPT-4 synthesized data.
Outcome: The proposed framework outperforms the traditional single-stage visual instruction tuning framework and achieves state-of-the-art performance across a wide range of multi-modal evaluation benchmarks.
Ted-Tok: Maintaining an Evolving Vocabulary for Lifelong Learning (2026.acl-long)

Copied to clipboard

Challenge: a static tokenizer fragments newly emerging lexical items as language evolves . as language grows, a dynamic tokenizer reduces compression efficiency and performance .
Approach: They propose a Temporal Drift Tokenizer that maintains an evolving vocabulary that adapts to emerging linguistic patterns over time.
Outcome: The proposed tokenizer maintains an evolving vocabulary that adapts to emerging linguistic patterns over time.
Autonomous Workflow for Multimodal Fine-Grained Training Assistants Towards Mixed Reality (2024.findings-acl)

Copied to clipboard

Challenge: a fine-grained, comprehensive understanding of multimodal environments remains under-explored.
Approach: They propose an automated workflow for integrating AI agents into extended reality (XR) they propose a cerebral language agent that integrates LLM with memory, planning, and interaction with XR tools and a vision-language agent .
Outcome: The proposed workflow integrates AI agents seamlessly into extended reality (XR) applications for fine-grained training.
LLaMA-Berry: Pairwise Optimization for Olympiad-level Mathematical Reasoning via O1-like Monte Carlo Tree Search (2025.naacl-long)

Copied to clipboard

Challenge: LLaMA-Berry is an advanced mathematical reasoning framework to enhance the problem-solving ability of large language models (LLMs).
Approach: They propose a Monte Carlo Tree Search and Self-Refine framework to optimize reasoning paths and a pairwise reward model to evaluate different paths globally.
Outcome: The proposed framework overcomes inefficiencies and limitations of step-wise and greedy search algorithms, enabling more efficient exploration of solution spaces.
New Terms, New Toxicity: Consensus-based Chinese Neologism Toxicity Detection via Search-Augmented LLMs (2026.acl-long)

Copied to clipboard

Challenge: Neologisms can foster new linguistic consensus by stabilizing shared meanings and usage in common communicative norms.
Approach: They propose a taxonomy that captures the origins and consensus-verification criteria of toxic neologisms . they propose 'SeTox' framework that integrates real-time web context for naeologim detection .
Outcome: The proposed framework outperforms large-scale models in detecting neologism toxicity.
INT: Establishing Information Transfer for Multilingual Intent Detection and Slot Filling (2025.findings-acl)

Copied to clipboard

Challenge: Existing studies struggle to achieve performance comparable to that on high-resource languages due to inherent linguistic diversity of multilingual SLU tasks.
Approach: They propose a multilingual information transfer network to solve these challenges . they propose to reformulate SF as a span prediction problem and introduce a slot-matching attention mechanism to achieve slot alignment across languages.
Outcome: The proposed model outperforms baseline models on the MASSIVE and MASSIV-UG datasets in overall accuracy across all languages.
Tasty Burgers, Soggy Fries: Probing Aspect Robustness in Aspect-Based Sentiment Analysis (2020.emnlp-main)

Copied to clipboard

Challenge: Existing ABSA test sets cannot be used to distinguish the sentiment of the target aspect from the non-target aspect.
Approach: They propose a simple but effective approach to enrich ABSA test sets by disentangle the confounding sentiments of non-target aspects from the target aspect’s sentiment.
Outcome: The proposed model can distinguish the sentiment of the non-target aspects from the target aspect’s sentiment by using the Aspect Robustness Test Set (ARTS).
Parrot: Enhancing Multi-Turn Instruction Following for Large Language Models (2024.acl-long)

Copied to clipboard

Challenge: Existing studies overlook the multi-turn instruction following ability of large language models (LLMs) Extensive experiments show that Parrot improves current LLMs by up to 7.2% in multi- turn instruction following.
Approach: They propose a method for collecting multi-turn instructions that feature human-like queries, such as anaphora and ellipsis, and a context-aware preference optimization strategy to further enhance LLMs for complex queries.
Outcome: The proposed method improves existing LLMs by up to 7.2% in multi-turn instruction following.
Rhombus: Incentivizing Coordination in Parallel Thinking through Reinforcement Learning (2026.findings-acl)

Copied to clipboard

Challenge: Parallel thinking is a promising avenue for scaling test-time compute in Large Language Models . however, coordinating the exploration and aggregation stages remains challenging .
Approach: They propose a parallel thinking framework that explicitly incentivizes coordination between components via end-to-end reinforcement learning.
Outcome: The proposed framework improves accuracy by 6.0% over long chain-of-thought baselines while reducing wall-clock latency by 39.4% under matched token budgets.

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