Papers by Ze Li

12 papers
ChunkAttention: Efficient Self-Attention with Prefix-Aware KV Cache and Two-Phase Partition (2024.acl-long)

Copied to clipboard

Challenge: Experiments show that ChunkAttention can speed up the self-attention kernel by 3.2-4.8 compared to the start-of-the-art implementation.
Approach: They propose a prefix-aware self-attention module that can detect matching prompt prefixes across multiple requests and share their key/value tensors in memory at runtime.
Outcome: The proposed module can speed up the self-attention kernel by 3.2-4.8 compared to the start-of-the-art implementation, with the length of the system prompt ranging from 1024 to 4096.
HILL: Hierarchy-aware Information Lossless Contrastive Learning for Hierarchical Text Classification (2024.naacl-long)

Copied to clipboard

Challenge: Existing self-supervised methods in natural language processing rely on augmentation rules to generate contrastive samples.
Approach: They propose a hierarchy-aware information lossless contrastive learning scheme that uses syntactic information reserved in the input sample and fused during the learning process.
Outcome: The proposed learning scheme is superior to existing methods in hierarchical text classification . the proposed learning system is based on a structure encoder and a text encoder .
Do VLMs Have a Moral Backbone? A Study on the Fragile Morality of Vision-Language Models (2026.findings-acl)

Copied to clipboard

Challenge: Vision-Language Models (VLMs) have advanced multimodal learning, driving progress in cross-modal reasoning.
Approach: They propose to examine moral robustness of vision-language models by analyzing their moral stances under multimodal perturbations.
Outcome: The proposed model-agnostic multimodal perturbations expose VLMs to a variety of moral vulnerabilities, including a sycophancy trade-off where stronger instruction-following models are more susceptible to persuasion.
Read, Attend and Comment: A Deep Architecture for Automatic News Comment Generation (D19-1)

Copied to clipboard

Challenge: Existing methods for news comment generation have not been well studied.
Approach: They propose a “read-attend-comment” procedure for automatic news comment generation and formalize it with a reading network and a generation network.
Outcome: The proposed procedure outperforms existing methods in terms of automatic evaluation and human judgment on two public datasets.
FlexGuard: Continuous Risk Scoring for Strictness-Adaptive LLM Content Moderation (2026.acl-long)

Copied to clipboard

Challenge: Existing guardrail models for content moderation assume a fixed definition of harmfulness, but enforced strictness varies across platforms and evolves over time, resulting in brittle moderators.
Approach: They propose a strictness-adaptive LLM moderation benchmark that enables controlled evaluation under multiple strictness regimes.
Outcome: The proposed moderator performs better under one regime and under another, and is more robust under varying strictness.
How do Language Models Reshape Entity Alignment? A Survey of LM-Driven EA Methods: Advances, Benchmarks, and Future (2025.emnlp-main)

Copied to clipboard

Challenge: Entity alignment (EA) is critical for knowledge graph (KG) integration.
Approach: They propose a taxonomy that categorizes methods in three stages: data preparation, feature embedding, and alignment.
Outcome: The proposed taxonomy categorizes methods in three key stages: data preparation, feature embedding, and alignment.
Let Retrievers Think Before Action: Thought-Augmented Embedding for Dense Retrieval (2026.findings-acl)

Copied to clipboard

Challenge: Large language models have demonstrated that explicit step-by-step thinking can substantially improve performance on complex tasks.
Approach: They propose a model that generates preliminary thoughts for input queries before document retrieval.
Outcome: The proposed model generates preliminary thoughts for input queries before document retrieval.
Generating Domain-Specific Knowledge Graphs from Large Language Models (2025.findings-acl)

Copied to clipboard

Challenge: Large language models (LLMs) have shown impressive world knowledge across different benchmarks and domains but their knowledge is inconveniently scattered across their billions of parameters.
Approach: They propose a prompt-based method to extract knowledge solely from LLMs’ parameters to construct domain-specific KGs by a schema-based process.
Outcome: The proposed method generates large domain-specific KGs containing tens of thousands of entities and relations, and then evaluates against Wikidata, an open-source human-created KG.
Low-Resource Response Generation with Template Prior (D19-1)

Copied to clipboard

Challenge: Existing open domain response generation models are limited to paired data, but are less explored in real-world applications.
Approach: They propose to train a neural response generation model with unpaired data and paired data as prior.
Outcome: The proposed model outperforms state-of-the-art models in both automatic and human evaluation when only a few pairs are available.
StyleDGPT: Stylized Response Generation with Pre-trained Language Models (2020.findings-emnlp)

Copied to clipboard

Challenge: Existing methods for generating responses following a desired style are lacking of parallel data for training.
Approach: They propose a KL loss and a style classifier to fine-tune response generation . they show that their model can significantly outperform state-of-the-art methods .
Outcome: The proposed model outperforms state-of-the-art models in style consistency and contextual coherence with two public datasets.
MMAPS: End-to-End Multi-Grained Multi-Modal Attribute-Aware Product Summarization (2024.lrec-main)

Copied to clipboard

Challenge: Existing product summarization methods lack end-to-end product summaries and multi-grained multi-modal modeling.
Approach: They propose an end-to-end multi-grained multi-modal attribute-aware product summarization method that jointly models product attributes and generates product summaries.
Outcome: The proposed method outperforms state-of-the-art product summarization methods on a large-scale Chinese e-commence dataset.
TANet: Thread-Aware Pretraining for Abstractive Conversational Summarization (2022.findings-naacl)

Copied to clipboard

Challenge: Existing pre-trained language models are difficult to apply to abstractive conversational summarization tasks.
Approach: They propose a thread-aware Transformer-based network that incorporates contextual dependency into the conversational summarization model.
Outcome: The proposed model can be applied to real conversations using a large-scale pretraining dataset.

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