Papers by Weiwei Shi
Generate-then-Ground in Retrieval-Augmented Generation for Multi-hop Question Answering (2024.acl-long)
Copied to clipboard
| Challenge: | Existing approaches to solve multi-hop question are constrained by the retriever and the noise in the retrieved documents. |
| Approach: | They propose a framework that integrates parametric knowledge of large language models with external documents to solve a multi-hop question. |
| Outcome: | The proposed framework is based on the parametric knowledge of LLMs and external documents to solve a multi-hop question. |
RADE: Reference-Assisted Dialogue Evaluation for Open-Domain Dialogue (2023.acl-long)
Copied to clipboard
| Challenge: | Evaluating open-domain dialogue systems is challenging because of the one-to-many problem. |
| Approach: | They propose a reference-based dialogue evaluation approach that leverages the pre-created utterance as reference other than the gold response to relieve the one-to-many problem. |
| Outcome: | The proposed method outperforms state-of-the-art evaluation methods on three datasets and two existing benchmarks. |
CT-FineBench: A Diagnostic Fidelity Benchmark for Fine-Grained Evaluation of CT Report Generation (2026.acl-long)
Copied to clipboard
| Challenge: | Existing evaluation metrics for radiology report generation focus on lexical overlap and entity matching. |
| Approach: | They propose a benchmark to evaluate the fine-grained factual consistency of CT reports . they use a question-answering process to query a machine-generated report . |
| Outcome: | The proposed benchmark evaluates the fine-grained factual consistency of CT reports . it correlates better with expert clinical assessment and is more sensitive to errors . |
RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining (2022.acl-long)
Copied to clipboard
| Challenge: | Large-scale pretrained language models have achieved SOTA results on NLP tasks but are vulnerable to adversarial attacks especially for logographic languages like Chinese. |
| Approach: | They propose a pretrained Chinese Bert that is robust to various forms of adversarial attacks like word perturbation, synonyms, typos, etc. |
| Outcome: | The proposed model outperforms baselines on 5 Chinese NLU tasks without sacrificing performance on clean testsets. |
ICL-Bandit: Relevance Labeling in Advertisement Recommendation Systems via LLM (2025.findings-emnlp)
Copied to clipboard
Lu Wang, Chiming Duan, Pu Zhao, Fangkai Yang, Yong Shi, Xuefeng Luo, Bingjing Xu, Weiwei Deng, Qingwei Lin, Dongmei Zhang
| Challenge: | In-context learning (ICL) is a common practice to enhance LLM performance on domain-specific tasks. |
| Approach: | They propose a method that leverages large language models to enhance query-ad relevance labeling . they identify and provide superior demonstrations for ICL, thereby improving labeling performance . |
| Outcome: | The proposed method improves query-ad relevance labeling performance by providing demonstrations. |
ResLoRA: Identity Residual Mapping in Low-Rank Adaption (2024.findings-acl)
Copied to clipboard
Shuhua Shi, Shaohan Huang, Minghui Song, Zhoujun Li, Zihan Zhang, Haizhen Huang, Furu Wei, Weiwei Deng, Feng Sun, Qi Zhang
| Challenge: | Low-rank adaptation (LoRA) is one of the most popular parameter-efficient fine-tuning methods. |
| Approach: | They propose a low-rank adaptation method that adds residual paths during training and merges them together during inference to achieve better results. |
| Outcome: | The proposed method achieves 2.5x faster convergence speed and improves performance by 14.3% on NLG, NLU, and text-to-image tasks. |
MAIR: A Massive Benchmark for Evaluating Instructed Retrieval (2024.emnlp-main)
Copied to clipboard
Weiwei Sun, Zhengliang Shi, Wu Long, Lingyong Yan, Xinyu Ma, Yiding Liu, Min Cao, Dawei Yin, Zhaochun Ren
| Challenge: | Existing IR benchmarks focus on a limited scope of tasks, making them insufficient for evaluating the latest IR models. |
| Approach: | They propose a multi-task instruction-tuned IR benchmark that includes 126 distinct IR tasks across 6 domains. |
| Outcome: | The proposed model performs better on instruction-tuned models than non-instruction-tunned models on MAIR. |