Papers by Zhenwei Tang
SPIN: Sparsifying and Integrating Internal Neurons in Large Language Models for Text Classification (2024.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have revolutionized text classification, but current paradigms rely on output of final layer . implicit internal structures that contribute to LLMs' impressive performance are neglected, forgoing potential performance gains. |
| Approach: | They propose a model-agnostic framework that sparsifies internal neurons of intermediate layers of LLMs for text classification. |
| Outcome: | The proposed framework significantly improves text classification accuracy, efficiency and interpretability. |
Stepwise Perplexity-Guided Refinement for Efficient Chain-of-Thought Reasoning in Large Language Models (2025.findings-acl)
Copied to clipboard
Yingqian Cui, Pengfei He, Jingying Zeng, Hui Liu, Xianfeng Tang, Zhenwei Dai, Yan Han, Chen Luo, Jing Huang, Zhen Li, Suhang Wang, Yue Xing, Jiliang Tang, Qi He
| Challenge: | Chain-of-Thought (CoT) reasoning has improved the performance of large language models (LLMs) however, the detailed reasoning process in CoT often incurs long generation times and high computational costs due to the inclusion of unnecessary steps. |
| Approach: | They propose a method to identify critical reasoning steps using perplexity as a measure of their importance. |
| Outcome: | The proposed method achieves a better balance between reasoning accuracy and efficiency of CoT. |
DiffuDetox: A Mixed Diffusion Model for Text Detoxification (2023.findings-acl)
Copied to clipboard
Griffin Floto, Mohammad Mahdi Abdollah Pour, Parsa Farinneya, Zhenwei Tang, Ali Pesaranghader, Manasa Bharadwaj, Scott Sanner
| Challenge: | Existing text generation models that reduce toxicity of toxic text are inadequate for text detoxification tasks. |
| Approach: | They propose a conditional and unconditional diffusion model for text detoxification . conditional model takes toxic text as condition and reduces its toxicity . experimental results show the model achieves human-level fluency . |
| Outcome: | The proposed model reduces toxic text and produces diverse sentences . it can be used to train other models and ensure fluency . |
A General Framework to Enhance Fine-tuning-based LLM Unlearning (2025.findings-acl)
Copied to clipboard
Jie Ren, Zhenwei Dai, Xianfeng Tang, Hui Liu, Jingying Zeng, Zhen Li, Rahul Goutam, Suhang Wang, Yue Xing, Qi He, Hui Liu
| Challenge: | Existing approaches to remove copyrighted and privacy-sensitive data from Large Language Models (LLMs) have been proposed to remove specific data from LLMs without requiring full retraining. |
| Approach: | They propose a general framework that enhances the utility of fine-tuning-based methods by distinguishing target data and suppressing related generations. |
| Outcome: | The proposed framework improves the unlearning and utility of fine-tuning-based methods by distinguishing the target data and suppressing related generations. |
SimRAG: Self-Improving Retrieval-Augmented Generation for Adapting Large Language Models to Specialized Domains (2025.naacl-long)
Copied to clipboard
Ran Xu, Hui Liu, Sreyashi Nag, Zhenwei Dai, Yaochen Xie, Xianfeng Tang, Chen Luo, Yang Li, Joyce C. Ho, Carl Yang, Qi He
| Challenge: | Retrieval-augmented generation (RAG) enhances the question answering abilities of large language models (LLMs) however, adapting general-purpose RAG systems to specialized fields poses unique challenges due to distribution shifts and limited access to domain-specific data. |
| Approach: | They propose a method that equips large language models with joint capabilities of question answering and question generation for domain adaptation. |
| Outcome: | Experiments on 11 datasets across three different domains verify the efficacy of SimRAG over baselines by 1.2%–8.6%. |
LLM Safety From Within: Detecting Harmful Content with Internal Representations (2026.acl-long)
Copied to clipboard
| Challenge: | State-of-the-art guard models rely on terminal-layer representations and overlook safety-relevant features encoded across internal layers. |
| Approach: | They propose a lightweight guard model that harnesses safety neurons from LLM internals without modifying the underlying model. |
| Outcome: | The proposed model outperforms open-source guard models across multiple benchmarks while using 250 fewer trainable parameters. |
A Reward-Guided Dual-Phase Framework for Adaptive Inference-Time Reasoning (2026.findings-acl)
Copied to clipboard
Yingqian Cui, Zhenwei Dai, Pengfei He, Bing He, Hui Liu, Zhan Shi, Xianfeng Tang, Jingying Zeng, Suhang Wang, Yue Xing, Jiliang Tang, Benoit Dumoulin
| Challenge: | Large Language Models (LLMs) have made strong progress in reasoning. |
| Approach: | They propose a dual-phase test-time scaling framework that separates planning and execution and performs search over each phase independently. |
| Outcome: | Experiments on math reasoning and code generation benchmarks show that the proposed approach improves accuracy while reducing redundant computation. |