EMNLP Findings2023
WikiChat: Stopping the Hallucination of Large Language Model Chatbots by Few-Shot Grounding on Wikipedia
Sina J. Semnani, Violet Z. Yao, Heidi C. Zhang, Monica S. Lam
摘要
This paper presents the first few-shot LLMbased chatbot that almost never hallucinates and has high conversationality and low latency. WikiChat is grounded on the English Wikipedia, the largest curated free-text corpus.
WikiChat generates a response from an LLM, retains only the grounded facts, and combines them with additional information it retrieves from the corpus to form factual and engaging responses. We distill WikiChat based on GPT-4 into a 7B-parameter LLaMA model with minimal loss of quality, to significantly improve its latency, cost and privacy, and facilitate research and deployment 1 .
Using a novel hybrid human-and-LLM evaluation methodology, we show that our best system achieves 97.3% factual accuracy in simulated conversations. It significantly outperforms all retrieval-based and LLM-based baselines, and by 3.9%, 38.6% and 51.0% on head, tail and recent knowledge compared to GPT-4. Compared to previous state-of-the-art retrieval-based chatbots, WikiChat is also significantly more informative and engaging, just like an LLM. WikiChat achieves 97.9% factual accuracy in conversations with human users about recent topics, 55.0% better than GPT-4, while receiving significantly higher user ratings and more favorable comments. User: Have you heard about the upcoming film Oppenheimer? WikiChat: Yes! Oppenheimer is a biopic … User: What do you think about the casting? Query: cast of Oppenheimer Christopher Nolan film Date: "recent" Title: Christopher Nolan Nolan's twelfth directorial venture will be "Oppenheimer" (2023), … Title: Oppenheimer (film) … Much of the main cast signed on between September 2021 and April 2022., … Title: Cillian Murphy … Murphy was confirmed to star as J. Robert Oppenheimer in Nolan's upcoming film Oppenheimer … Christopher Nolan is known for his meticulous casting choices. He often casts well-known actors in his films, and this time he has selected Tom Hanks and … Michael Caine.
• Christopher Nolan is known for his meticulous casting choices.
• Christopher Nolan often casts well-known actors in his films.
• Tom Hanks stars in the Oppenheimer (2023) movie.
• Michael Caine stars in the Oppenheimer (2023) movie.
• Much of the Oppenheimer (2023) movie's cast signed on between September 2021 and April 2022.
• Cilian Murphy is to play as Robert Oppenheimer in the Oppenherimer (2023) movie.
• Christopher Nolan is known for his meticulous casting choices.
• Christopher Nolan often casts well-known actors in his films.
Cillian Murphy has been cast to play Robert Oppenheimer in Christopher Nolan's "Oppenheimer" movie. Christopher Nolan is known for his meticulous casting choices and often casts well-known actors in his films. The movie's cast was signed on between September 2021 and April 2022.
Christopher Nolan has cast Cillian Murphy as the titular role of Oppenheimer, keeping with his tradition of meticulous casting choices. I'm sure it will be an excellent film!
问 Lune
尽管问,回答会引用原文。
Lune 会读完全文和相关的 Paper,每个回答都附上所依据的原文段落。
引用它的顶会 Paper15
- Language Models with Conformal Factuality GuaranteesChristopher Mohri, Tatsunori HashimotoICML 2024 · 107 次引用
- Unified Hallucination Detection for Multimodal Large Language ModelsXiang Chen, Chenxi Wang, Yida Xue, Ningyu Zhang 等ACL 2024 · 20 次引用
- ANAH-v2: Scaling Analytical Hallucination Annotation of Large Language ModelsYuzhe Gu, Ziwei Ji, Wenwei Zhang, Chengqi Lyu 等NeurIPS 2024 · 20 次引用
- WebANNS: Fast and Efficient Approximate Nearest Neighbor Search in Web BrowsersMugeng Liu, Siqi Zhong, Qi Yang, Yudong Han 等SIGIR 2025 · 4 次引用
- Enhancing LLM's Cognition via StructurizationKai Liu, Zhihang Fu, Chao Chen, Wei Zhang 等NeurIPS 2024 · 3 次引用
它引用的 Paper26
- Language Models are Few-Shot LearnersTom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah 等NeurIPS 2020 · 64,255 次引用
- Chain-of-Thought Prompting Elicits Reasoning in Large Language ModelsJason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma 等NeurIPS 2022 · 22,562 次引用
- Retrieval-Augmented Generation for Knowledge-Intensive NLP TasksPatrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni 等NeurIPS 2020 · 19,162 次引用
- FlashAttention: Fast and Memory-Efficient Exact Attention with IO-AwarenessTri Dao, Daniel Y. Fu, Stefano Ermon, Atri Rudra 等NeurIPS 2022 · 5,493 次引用
- Self-Refine: Iterative Refinement with Self-FeedbackAman Madaan, Niket Tandon, Prakhar Gupta, Skyler Hallinan 等NeurIPS 2023 · 4,972 次引用
相关 Paper
- FActScore: Fine-grained Atomic Evaluation of Factual Precision in Long Form Text GenerationSewon Min, Kalpesh Krishna, Xinxi Lyu, Mike Lewis 等EMNLP 2023 · 225 次引用
- Fine-tuned LLMs Know More, Hallucinate Less with Few-Shot Sequence-to-Sequence Semantic Parsing over WikidataSilei Xu, Shicheng Liu, Theo Culhane, Elizaveta Pertseva 等EMNLP 2023 · 10 次引用
- Instruction-tuned Language Models are Better Knowledge LearnersZhengbao Jiang, Zhiqing Sun, Weijia Shi, Pedro Rodríguez 等ACL 2024 · 11 次引用
- FreshLLMs: Refreshing Large Language Models with Search Engine AugmentationTu Vu, Mohit Iyyer, Xuezhi Wang, Noah Constant 等ACL Findings 2024 · 73 次引用
- LLaMA-Omni 2: LLM-based Real-time Spoken Chatbot with Autoregressive Streaming Speech SynthesisQingkai Fang, Yan Zhou, Shoutao Guo, Shaolei Zhang 等ACL 2025
