Papers by Kai Wei

37 papers
Adaptive Policy with Wait-k Model for Simultaneous Translation (2023.emnlp-main)

Copied to clipboard

Challenge: Existing approaches to simultaneous machine translation require a robust read/write policy . a standalone multi-path wait-k model performs competitively with adaptive policies .
Approach: They propose a more flexible approach by decoupling the adaptive policy model from the translation model.
Outcome: The proposed approach outperforms baseline approaches in translation tasks.
Layer-wise Importance Matters: Less Memory for Better Performance in Parameter-efficient Fine-tuning of Large Language Models (2024.findings-emnlp)

Copied to clipboard

Challenge: Parameter-Efficient Fine-Tuning (PEFT) methods have gained popularity for adapting pre-trained Large Language Models (LLMs) to downstream tasks.
Approach: They propose a method to optimize the importance of full layers with layer-wise importance scoring by leveraging the estimated importance scores.
Outcome: The proposed method is compatible with PEFT methods that operate on a per-layer basis and achieves better performance.
SDBench: A Survey-based Domain-specific LLM Benchmarking and Optimization Framework (2025.acl-long)

Copied to clipboard

Challenge: acquiring domain-specific knowledge often requires professional expert manpower.
Approach: They propose a generic framework for generating evaluation datasets for domain-specific LLMs.
Outcome: The proposed framework reduces the reliance on expert manpower while ensuring that the collected data is uniformly distributed.
Unified Demonstration Retriever for In-Context Learning (2023.acl-long)

Copied to clipboard

Challenge: In-context learning is a new learning paradigm where a language model conditions on a few input-output pairs (demonstrations) and a test input, and directly outputs the prediction.
Approach: They propose a single model to retrieve demonstrations for a wide range of tasks by combining training signals from various tasks into a unified list-wise ranking formulation by language model’s feedback.
Outcome: The proposed model outperforms baselines on 30+ tasks across 13 task families and multiple data domains.
A Multi-Agent Framework for High-Interaction Terminal Simulation (2026.acl-long)

Copied to clipboard

Challenge: Terminal simulation is a problem of symbolic language generation in dialogue and interactive systems.
Approach: They propose a terminal command-level Turing test framework that improves realism, consistency and robustness in command-language generation.
Outcome: The proposed framework outperforms state-of-the-art benchmarks by more than 9% on multi-turn terminal simulation.
Multi-Stage Pre-training for Automated Chinese Essay Scoring (2020.emnlp-main)

Copied to clipboard

Challenge: Existing methods for automatic essay scoring are based on hand-crafted surface-level features, but recent advances in representation learning have improved performance.
Approach: They propose a pre-training based automated Chinese essay scoring method with weakly supervised pre- training, supervised cross- prompt fine-tuning and supervised target- prompt refine-tuneing.
Outcome: The proposed method improves a state-of-the-art neural essay scorer in terms of effectiveness and domain adaptation ability, while in-depth analysis also reveals its limitations.
Low-Resource Generation of Multi-hop Reasoning Questions (2020.acl-main)

Copied to clipboard

Challenge: Existing methods to generate valid and fluent questions from text are limited and insufficient for training.
Approach: They propose to generate multi-hop reasoning questions from the raw text in a low resource circumstance by deducing over multiple relations on several sentences in the text.
Outcome: The proposed model can be applied to the task of machine reading comprehension and achieve significant performance improvements.
Better Simultaneous Translation with Monotonic Knowledge Distillation (2023.acl-long)

Copied to clipboard

Challenge: Existing methods to train offline MT models require generating target tokens before source sentence is fully consumed.
Approach: They propose a method that leverages traditional translation models as teachers to generate monotonic yet accurate reference translations for sequence-level knowledge distillation.
Outcome: The proposed approach improves on strong baselines and on a monotonic version of the WMT15 De-En test set.
Explore More Guidance: A Task-aware Instruction Network for Sign Language Translation Enhanced with Data Augmentation (2022.findings-naacl)

Copied to clipboard

Challenge: Existing studies focus on the recognition step, while paying less attention to sign language translation.
Approach: They propose a task-aware instruction network, namely TIN-SLT, for sign language translation, by introducing the isntruction module and the learning-based feature fuse strategy into a Transformer network.
Outcome: The proposed system outperforms existing solutions on two benchmark datasets, PHOENIX-2014-T and ASLG-PC12, and outperformed previous best solutions by 1.65 and 1.42 in terms of BLEU-4.
Memory-Guided Hard Data Augmentation for Multimodal Named Entity Recognition (2026.findings-acl)

Copied to clipboard

Challenge: Existing methods for Named Entity Recognition (NER) ignore the internal state of the target model.
Approach: They propose a framework to repair model-specific errors by using a model-based approach . they employ cross-validation to identify model- specific Hard Data and a memory tree to induce macro-level error patterns from micro-level failures.
Outcome: The proposed framework yields significant performance gains on Twitter and other platforms.
BotChat: Evaluating LLMs’ Capabilities of Having Multi-Turn Dialogues (2024.findings-naacl)

Copied to clipboard

Challenge: Modern Large Language Models (LLMs) facilitate high-quality, multi-turn dialogues with humans, but human-based evaluation of such a capability requires substantial manual effort.
Approach: They propose to evaluate LLMs' ability to emulate human-like, multi-turn conversations using an LLM-centric approach.
Outcome: The proposed model emulates human-like, multi-turn conversations using an LLM-centric approach.
GradOT: Training-free Gradient-preserving Offsite-tuning for Large Language Models (2025.acl-long)

Copied to clipboard

Challenge: Existing methods for offsite-tuning of large language models require high computational costs and lack theoretical analysis.
Approach: They propose an offsite-tuning approach that selectively applies compression techniques such as rank compression and channel pruning to preserve the gradients of fine-tuned adapters while ensuring privacy.
Outcome: The proposed method surpasses existing OT methods in privacy protection and model performance.
How do LLMs’ Preferences Affect Event Argument Extraction? CAT: Addressing Preference Traps in Unsupervised EAE (2025.findings-acl)

Copied to clipboard

Challenge: Existing approaches to supervised EAE suffer from preference traps due to misalignments between prior knowledge, instructions, or output constraints and LLMs’ preferences.
Approach: They propose an unsupervised EAE framework that handles LLMs' preference traps by targeting their prior knowledge and instructions.
Outcome: The proposed framework matches the best DeepSeek-R1 API model with a significantly lower time cost.
MARIO: MAth Reasoning with code Interpreter Output - A Reproducible Pipeline (2024.findings-acl)

Copied to clipboard

Challenge: Large language models lack mathematical reasoning, a hurdle on the path to true artificial general intelligence.
Approach: They propose a protocol for fine-tuning large language models with a Python code interpreter to enhance the text analysis of the LLMs.
Outcome: The proposed protocol improves the performance of a 7B-parameter LLM on the GSM8K and MATH datasets while allowing for an outlier-free value model-based inference method.
CLOWER: A Pre-trained Language Model with Contrastive Learning over Word and Character Representations (2022.coling-1)

Copied to clipboard

Challenge: Pre-trained language models (PLMs) have achieved remarkable performance gains across numerous downstream tasks in natural language understanding.
Approach: They propose a Chinese pre-trained language model that implicitly encodes words into characters . they propose 'contrastive learning over word' and 'character' representations to improve learning .
Outcome: The proposed model can encode words into fine-grained representations without modification of production pipelines.
EDA: Easy Data Augmentation Techniques for Boosting Performance on Text Classification Tasks (D19-1)

Copied to clipboard

Challenge: Existing data augmentation techniques for text classification are difficult to implement and cost a high amount of money.
Approach: They propose to use four simple but powerful operations to boost performance on text classification tasks to improve synonym replacement, random insertion, random swap, and random deletion.
Outcome: The proposed techniques improve performance on five classification tasks and are particularly useful for smaller datasets.
Improve Speech Translation Through Text Rewrite (2025.coling-industry)

Copied to clipboard

Challenge: Recent advances in speech translation (ST) research have focused on the unique characteristics of spontaneous speech, including accents and presentation quality.
Approach: They propose to transform transcribed speech into a cleaner style more in line with the expectations of translation models built from written text.
Outcome: Experiments on public and in-house translation models show that the proposed model can be effectively distilled into a standalone translation model.
Structure-aware Domain Knowledge Injection for Large Language Models (2025.acl-long)

Copied to clipboard

Challenge: Structure-aware Continual Pre-Training (SCPT) and Structure-Aware Supervised Fine-Tuning (SSFT) are two-stage strategies for knowledge injection and alignment that reduces the training corpus needs to 5% while achieving 100% of traditional knowledge injection performance.
Approach: They propose a method to efficiently transform foundation Large Language Models into domain specialists by using two-stage strategies: Structure-aware Continual Pre-Training and Structure-Aware Supervised Fine-Tuning.
Outcome: The proposed method significantly reduces the training corpus needs to a mere 5% while achieving 100% of traditional knowledge injection performance.
Knowledge as A Bridge: Improving Cross-domain Answer Selection with External Knowledge (C18-1)

Copied to clipboard

Challenge: Existing approaches to answer selection are limited in domains with limited labeled data.
Approach: They propose a Knowledge-aware Attentive Network framework for cross-domain answer selection that uses the knowledge base as a bridge to enable knowledge transfer from the source domain to the target domain.
Outcome: The proposed model outperforms strong competitors by a noticeable margin in cross-domain answer selection.
MPPO: Multi Pair-wise Preference Optimization for LLMs with Arbitrary Negative Samples (2025.coling-main)

Copied to clipboard

Challenge: Existing preference optimization methods such as DPO and KTO are inherently derived from PPO, requiring a reference model that adds GPU memory resources and relies heavily on abundant preference data.
Approach: They propose an algorithm that leverages the average likelihood of model responses to fit the reward function and maximizes the utilization of preference data.
Outcome: The proposed algorithm outperforms DPO, ORPO, and SimPO on MT-Bench and Arena-Hard.
Incorporating Dynamic Semantics into Pre-Trained Language Model for Aspect-based Sentiment Analysis (2022.findings-acl)

Copied to clipboard

Challenge: Aspect-based sentiment analysis (ABSA) predicts sentiment polarity towards a specific aspect in a sentence.
Approach: They propose to use a dynamic aspect-oriented semantics-based method to learn ABSA.
Outcome: The proposed method can learn dynamic aspect-oriented semantics for ABSA on three benchmark datasets.
RankCSE: Unsupervised Sentence Representations Learning via Learning to Rank (2023.acl-long)

Copied to clipboard

Challenge: Unsupervised sentence representation learning is one of the fundamental problems in natural language processing . contrastive learning methods fail to capture fine-grained ranking information among the sentences .
Approach: They propose a novel approach for unsupervised sentence representation learning that integrates ranking consistency and ranking distillation with contrastive learning into a unified framework.
Outcome: The proposed approach performs better over state-of-the-art models on STS and TR tasks.
LegalAgentBench: Evaluating LLM Agents in Legal Domain (2025.acl-long)

Copied to clipboard

Challenge: Existing general-domain benchmarks do not capture complexity of real-world judicial cognition and decision-making.
Approach: They propose a benchmark specifically designed to evaluate LLM Agents in the legal domain.
Outcome: The proposed benchmark includes 17 corpora from real-world legal scenarios and provides 37 tools for interacting with external knowledge.
Decoupling Memories, Muting Neurons: Towards Practical Machine Unlearning for Large Language Models (2025.findings-acl)

Copied to clipboard

Challenge: Existing methods for MU degrade model utility, especially when accessing the original training data.
Approach: They propose a method that eliminates the influence of unlearned data by modulating the outputs of merely 1% of the neurons in the feed-forward network modules within the Transformer blocks.
Outcome: The proposed method eliminates the influence of unlearned data from Large Language Models by modulating the outputs of 1% of the neurons in the feed-forward network modules within the Transformer blocks, minimizing disruption to the model’s performance.
MoQAE: Mixed-Precision Quantization for Long-Context LLM Inference via Mixture of Quantization-Aware Experts (2025.acl-long)

Copied to clipboard

Challenge: Existing approaches to optimize large language models for long-context inference are inefficient and consume memory.
Approach: They propose a mixed-precision quantization method via mixture of experts that inputs tokens into router chunk by chunk to reduce inference overhead.
Outcome: The proposed method outperforms state-of-the-art KV cache quantization methods on multiple benchmark datasets.
Cooperative Denoising for Distantly Supervised Relation Extraction (C18-1)

Copied to clipboard

Challenge: Existing methods for distantly supervised relation extraction suffer from noisy labeling problem, which can severely degrade its performance.
Approach: They propose a framework for distantly supervised relation extraction that leverages text corpus and knowledge graph and a cooperative module involving their mutual learning.
Outcome: The proposed method reduces the noisy labels and achieves substantial improvement over the state-of-the-art methods.
Multi-Passage Machine Reading Comprehension with Cross-Passage Answer Verification (P18-1)

Copied to clipboard

Challenge: Recent years have seen rapid growth in the MRC community . MRC is believed to be a crucial step in building a general intelligent agent .
Approach: They propose an end-to-end neural model that enables multiple passages to verify each other based on their content representations.
Outcome: The proposed model outperforms the baseline on the English MS-MARCO dataset and the Chinese DuReader dataset, and achieves state-of-the-art performance on both datasets.
League of LLMs: A Benchmark-Free Paradigm for Mutual Evaluation of Large Language Models (2026.acl-long)

Copied to clipboard

Challenge: Large language models (LLMs) have shown exceptional capabilities across a wide range of tasks, but reliable evaluation remains a challenge due to data contamination, opaque operation, and subjective preferences.
Approach: They propose a benchmark-free evaluation paradigm that organizes multiple LLMs into a self-governed league for multi-round mutual evaluation.
Outcome: Experiments on eight mainstream LLMs in mathematics and programming show that the proposed model can distinguish capabilities while maintaining high internal ranking stability.
LLM Factoscope: Uncovering LLMs’ Factual Discernment through Measuring Inner States (2024.findings-acl)

Copied to clipboard

Challenge: Large Language Models (LLMs) produce outputs that deviate from factual reality, especially in sensitive applications such as medical consultation and legal advice.
Approach: They propose a Siamese network-based model that leverages LLMs’ inner states for factual detection.
Outcome: The proposed model achieves over 96% accuracy on a custom-collected factual detection dataset.
SoftDedup: an Efficient Data Reweighting Method for Speeding Up Language Model Pre-training (2024.acl-long)

Copied to clipboard

Challenge: Current methods focus on detecting and removing duplicates, which risks the loss of valuable information and neglects the varying degrees of duplication.
Approach: They propose a method that maintains dataset integrity while selectively reducing the sampling weight of data with high commonness.
Outcome: The proposed method significantly improves training efficiency on deduplicated datasets and improves downstream accuracy by 1.77%.
pFedRAG: A Personalized Federated Retrieval-Augmented Generation System with Depth-Adaptive Tiered Embedding Tuning (2025.findings-emnlp)

Copied to clipboard

Challenge: Personalized Federated RAG framework enables efficient collaborative fine-tuning of embedding models . depth-adaptive tieered Embedding (DATE) architecture is tailored for local data and training results of each client.
Approach: a new Personalized Federated RAG framework is proposed for large language models . the framework enables efficient collaborative fine-tuning of embedding models based on common knowledge .
Outcome: a novel Personalized Federated RAG framework is proposed for large language models . the framework enables efficient collaborative fine-tuning of embedding models based on common knowledge .
Competency-Aware Neural Machine Translation: Can Machine Translation Know its Own Translation Quality? (2022.emnlp-main)

Copied to clipboard

Challenge: Neural machine translation models are often criticized for failures that happen without competency awareness.
Approach: They propose a method that extends conventional NMT with a self-estimator to translate a source sentence and estimate its competency.
Outcome: The proposed method performs on translation tasks intact and on quality estimation tasks better than existing methods.
DenseSSM: State Space Models with Dense Hidden Connection for Efficient Large Language Models (2025.naacl-long)

Copied to clipboard

Challenge: Large language models (LLMs) face excessive computational and memory requirements due to the commonly used Transformer architecture.
Approach: They propose a method to enhance the flow of hidden information between layers in large language models by selectively integrating shallow-layer hidden states into deeper layers.
Outcome: The proposed method maintains parallelizability and inference efficiency of SSMs while significantly boosting performance on public benchmarks.
Agent-based Substructure Counting under Local Differential Privacy (2026.acl-long)

Copied to clipboard

Challenge: Recent studies have demonstrated the ability of Large Language Models (LLMs) to process graph problems.
Approach: They propose to decompose substructure counting into node-level tasks distributed among node agents and embed the knowledge of distributed algorithms and DP frameworks in the curator agent and privacy controller.
Outcome: Extensive experiments on 6 real-world datasets validate the effectiveness of the proposed framework for substructure counting tasks under edge local differential privacy (LDP).
ItiNera: Integrating Spatial Optimization with Large Language Models for Open-domain Urban Itinerary Planning (2024.emnlp-industry)

Copied to clipboard

Challenge: Existing urban itinerary planning studies focus on traditional tourism, but they lack the precision and accuracy needed to create a personalized itinerary.
Approach: They propose an open-domain urban itinerary planning system that integrates spatial optimization with large language models to provide customized urban itineraries based on user needs.
Outcome: The proposed system can generate personalized urban itineraries based on user needs and scale with existing methods.
TeCES: Collaborative Geometric Knowledge Representation Framework under Evolving Fact Snapshots (2026.acl-long)

Copied to clipboard

Challenge: Existing knowledge graphs represent static facts but lack collaborative modeling of both . e.g., existing knowledge graph models lack a framework for integrating snapshots into knowledge graph.
Approach: They propose a framework for high-fidelity modeling of evolving snapshots using concept of snapshots.
Outcome: The proposed framework outperforms existing models on six benchmarks.

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