Papers by Zhefeng Wang

18 papers
FISTAPruner: Layer-wise Post-training Pruning for Large Language Models (2025.emnlp-main)

Copied to clipboard

Challenge: Existing pruning methods require inefficient retraining for billion-scale LLMs or rely on heuristicically designed metrics to determine pruning masks, leading to performance degradation.
Approach: They propose a convex optimization model that induces sparsity in large language models by leveraging FISTA.
Outcome: The proposed method can remove 50% of model parameters while retaining 98.6% and 95.6% of the zero-shot performance.
An In-depth Study on Internal Structure of Chinese Words (2021.acl-long)

Copied to clipboard

Challenge: Unlike English letters, Chinese characters have rich and specific meanings.
Approach: They propose to model Chinese words' internal structures as dependency trees with 11 labels for distinguishing syntactic relationships.
Outcome: The proposed model of Chinese word-internal structures shows it can be used to parse sentences . it shows that the model can be applied to a sentence-level task with a competitive dependency parser.
Reference Matters: Benchmarking Factual Error Correction for Dialogue Summarization with Fine-grained Evaluation Framework (2023.acl-long)

Copied to clipboard

Challenge: Current evaluations of FEC models that depend on factuality metrics are not reliable and detailed enough.
Approach: They propose a fine-grained evaluation framework that automatically evaluates FEC models on different error categories.
Outcome: The proposed evaluation framework compares models on different error categories and finds the best training modes and significant differences in the performance of existing models.
HacRED: A Large-Scale Relation Extraction Dataset Toward Hard Cases in Practical Applications (2021.findings-acl)

Copied to clipboard

Challenge: Relation extraction (RE) is an essential topic in natural language processing and has attracted extensive attention.
Approach: They propose a case-oriented construction framework to build a hard case relation extraction dataset with 65,225 relational facts annotated from 9,231 documents.
Outcome: The proposed model achieves a high 96% F1 score on data quality and is far lower than humans.
Delving Deep into Regularity: A Simple but Effective Method for Chinese Named Entity Recognition (2022.findings-naacl)

Copied to clipboard

Challenge: Named entity recognition (NER) is a system for identifying text spans pertaining to specific entity types.
Approach: They propose a method to investigate the regularity of Chinese NER's entity mentions by a regularity-aware module and a periodicity-gnostic module.
Outcome: The proposed model significantly outperforms previous state-of-the-art methods on three benchmark datasets and a practical medical dataset.
Accurate KV Cache Quantization with Outlier Tokens Tracing (2025.acl-long)

Copied to clipboard

Challenge: Large Language Models (LLMs) require substantial computational resources during deployment.
Approach: They propose a method to identify outlier tokens and exclude them from quantization . they find that the method can deliver a 6.4 times reduction in memory usage and a 2.5 times increase in throughput .
Outcome: The proposed method delivers a 6.4 times reduction in memory usage and a 2.5 times increase in throughput under 2-bit quantization.
Adaptive Feature-based Low-Rank Compression of Large Language Models via Bayesian Optimization (2024.findings-emnlp)

Copied to clipboard

Challenge: Large language models require a balance between efficiency and performance.
Approach: They propose a low-rank compression technique that reduces non-essential parameters by decomposing weight matrices into products of two low-ranked matrici.
Outcome: The proposed method outperforms existing pruning and low-rank compression techniques in maintaining model performance at the same compression ratio.
Alignment-Augmented Speculative Decoding with Alignment Sampling and Conditional Verification (2025.emnlp-main)

Copied to clipboard

Challenge: Existing methods to accelerate autoregressive generation of large language models require training costs.
Approach: They propose a training-free alignment-augmented speculative decoding algorithm . it leverages the output distribution obtained in the prefilling phase to provide more aligned draft candidates .
Outcome: The proposed method increases the average generation score by 3.3 points for the LLaMA3 model.
Punctuation-Steered Representation Fine-Tuning (2026.acl-short)

Copied to clipboard

Challenge: Existing methods for parameter-efficient fine-tuning (PeFT) are limited due to their prohibitive size and computational demands.
Approach: They propose a method that fine-tunes punctuation representations to achieve performance improvements.
Outcome: The proposed method improves performance by altering the representation space alone . but it results in suboptimal performance due to the effects of the method on the output .
OPT-Tree: Speculative Decoding with Adaptive Draft Tree Structure (2025.tacl-1)

Copied to clipboard

Challenge: Autoregressive language models generate one token in one step, limiting inference efficiency . Existing methods do not adapt to different situations to maximize acceptance length . speculative decoding has shown great potential for lossless acceleration .
Approach: They propose an algorithm to construct adaptive and scalable draft trees for autoregressive language models.
Outcome: Experimental results show that OPT-Tree outperforms existing draft trees and achieves speed-up ratio of up to 3.2 compared with autoregressive decoding.
Revisiting Pre-trained Language Models and their Evaluation for Arabic Natural Language Processing (2022.emnlp-main)

Copied to clipboard

Challenge: Existing pre-trained language models are not well-explored and are not reproducible in the literature.
Approach: They propose to improve existing Arabic language pre-trained language models using a more methodical approach.
Outcome: The proposed models outperform existing models on ALUE, a leaderboard-powered benchmark for Arabic NLU and NLG tasks.
Mirror: A Universal Framework for Various Information Extraction Tasks (2023.emnlp-main)

Copied to clipboard

Challenge: Recent studies often formulate IE tasks as a triplet extraction problem, but this paradigm does not support multi-span and n-ary extraction, leading to weak versatility.
Approach: They propose a multi-span cyclic graph extraction problem and a non-autoregressive graph decoding algorithm to extract all spans in a single step.
Outcome: The proposed model outperforms or reaches competitive performance with SOTA systems under few-shot and zero-shot settings and it is compatible with 57 datasets.
High-order Joint Constituency and Dependency Parsing (2024.lrec-main)

Copied to clipboard

Challenge: Syntactic parsing aims to reveal how sentences are syntactically structured.
Approach: They propose to produce compatible constituency and dependency trees simultaneously for input sentences . they adopt a much more efficient decoding algorithm and explore joint modeling at training phase .
Outcome: The proposed model significantly improves matching ratio of whole trees compared to separate models . the proposed model adopts a much more efficient decoding algorithm .
Improving Chinese Named Entity Recognition with Multi-grained Words and Part-of-Speech Tags via Joint Modeling (2024.lrec-main)

Copied to clipboard

Challenge: Named entity recognition (CNER) is a fundamental task in natural language processing (NLP).
Approach: They propose a tree parsing approach for jointly modeling Chinese named entity recognition (CNER) with multi-grained word segmentation (MWS) and POS tagging tasks.
Outcome: The proposed approach achieves better or comparable performance with current methods.
APGN: Adversarial and Parameter Generation Networks for Multi-Source Cross-Domain Dependency Parsing (2021.findings-emnlp)

Copied to clipboard

Challenge: Existing models for dependency parsing use labeled training data for several fixed domains, but performance drops when labeles only exist for several out-domains.
Approach: They propose a model for multi-source cross-domain dependency parsing that uses a parameter generation network and adversarial network for learning domain-invariant representations.
Outcome: The proposed model improves cross-domain parsing performance by about 2 points over strong BERT-enhanced baselines over a recently released dataset for multi-domain dependency parse.
CopyNE: Better Contextual ASR by Copying Named Entities (2024.acl-long)

Copied to clipboard

Challenge: Existing approaches to transcribe contextual named entities (NEs) treat entities as tokens and generate them token-by-token, which may result in incomplete transcriptions of entities.
Approach: They propose a mechanism that can copy entities from the NE dictionary and reduce errors during entity transcription.
Outcome: The proposed mechanism can copy entities from the NE dictionary, reducing errors during entity transcription, ensuring the completeness of the entity.
AraMUS: Pushing the Limits of Data and Model Scale for Arabic Natural Language Processing (2023.findings-acl)

Copied to clipboard

Challenge: Developing monolingual large Pre-trained Language Models (PLMs) is shown to be very successful in handling different tasks in Natural Language Processing (NLP).
Approach: They present AraMUS, the largest Arabic PLM with 11B parameters trained on 529GB of high-quality Arabic textual data.
Outcome: The proposed model achieves state-of-the-art performance on a diverse set of Arabic classification and generative tasks.
A High Precision Pipeline for Financial Knowledge Graph Construction (2020.coling-main)

Copied to clipboard

Challenge: Knowledge graphs are a standard for structured knowledge representation in the Semantic Web.
Approach: They propose to extract financial news articles into a knowledge graph by using a financial dictionary.
Outcome: The proposed pipeline extracts 342,000 financial news articles with a precision of 78% at the top-100 extractions.

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