FreshLLMs: Refreshing Large Language Models with Search Engine Augmentation

2023/10/05 by Tu Vu, Mohit Iyyer, Vu, Tu +19 · 3 voices · 40 citations
#cs.CL

paper · pdf · doi:10.48550/arxiv.2310.03214

Abstract

Most large language models (LLMs) are trained once and never updated; thus, they lack the ability to dynamically adapt to our ever-changing world. In this work, we perform a detailed study of the factuality of LLM-generated text in the context of answering questions that test current world knowledge. Specifically, we introduce FreshQA, a novel dynamic QA benchmark encompassing a diverse range of question and answer types, including questions that require fast-changing world knowledge as well as questions with false premises that need to be debunked. We benchmark a diverse array of both closed and open-source LLMs under a two-mode evaluation procedure that allows us to measure both correctness and hallucination. Through human evaluations involving more than 50K judgments, we shed light on limitations of these models and demonstrate significant room for improvement: for instance, all models (regardless of model size) struggle on questions that involve fast-changing knowledge and false premises. Motivated by these results, we present FreshPrompt, a simple few-shot prompting method that substantially boosts the performance of an LLM on FreshQA by incorporating relevant and up-to-date information retrieved from a search engine into the prompt. Our experiments show that FreshPrompt outperforms both competing search engine-augmented prompting methods such as Self-Ask (Press et al., 2022) as well as commercial systems such as Perplexity.AI. Further analysis of FreshPrompt reveals that both the number of retrieved evidences and their order play a key role in influencing the correctness of LLM-generated answers. Additionally, instructing the LLM to generate concise and direct answers helps reduce hallucination compared to encouraging more verbose answers. To facilitate future work, we release FreshQA at github.com/freshllms/freshqa and commit to updating it at regular intervals.

Cited by

• AdaSearch: Balancing Parametric Knowledge and Search in Large Language Models via Reinforcement Learning
• Beyond Fast and Slow: Cognitive-Inspired Elastic Reasoning for Large Language Models
• The FACTS Leaderboard: A Comprehensive Benchmark for Large Language Model Factuality
• Cooperative Retrieval-Augmented Generation for Question Answering: Mutual Information Exchange and Ranking by Contrasting Layers
• CryptoBench: A Dynamic Benchmark for Expert-Level Evaluation of LLM Agents in Cryptocurrency
• "AGI" team at SHROOM-CAP: Data-Centric Approach to Multilingual Hallucination Detection using XLM-RoBERTa
• "To Survive, I Must Defect": Jailbreaking LLMs via the Game-Theory Scenarios
• HEDGE: Hallucination Estimation via Dense Geometric Entropy for VQA with Vision-Language Models
• Honesty over Accuracy: Trustworthy Language Models through Reinforced Hesitation
• Towards Understanding, Analyzing, and Optimizing Agentic AI Execution: A CPU-Centric Perspective
• InfoFlow: Reinforcing Search Agent Via Reward Density Optimization
• Argument Collapse: LLMs Flatten Long-Form Public Debate
• Prior Knowledge Makes It Possible: From Sublinear Graph Algorithms to LLM Test-Time Methods
• Judging Against the Reference: Uncovering Knowledge-Driven Failures in LLM-Judges on QA Evaluation
• Mitigating Hallucination in Large Language Models (LLMs): An Application-Oriented Survey on RAG, Reasoning, and Agentic Systems
• SynthWorlds: Controlled Parallel Worlds for Disentangling Reasoning and Knowledge in Language Models
• MAD-Fact: A Multi-Agent Debate Framework for Long-Form Factuality Evaluation in LLMs
• A Comprehensive Survey on Reinforcement Learning-based Agentic Search: Foundations, Roles, Optimizations, Evaluations, and Applications
• FaStfact: Faster, Stronger Long-Form Factuality Evaluations in LLMs
• RefusalBench: Generative Evaluation of Selective Refusal in Grounded Language Models
• When Benchmarks Age: Temporal Misalignment through Large Language Model Factuality Evaluation
• Large Language Models Hallucination: A Comprehensive Survey
• SafeSearch: Automated Red-Teaming for the Safety of LLM-Based Search Agents
• The Artificial Intelligence Cognitive Examination: A Survey on the Evolution of Multimodal Evaluation From Recognition to Reasoning
• Speculating LLMs' Chinese Training Data Pollution from Their Tokens
• Improving Context Fidelity via Native Retrieval-Augmented Reasoning
• SimpleQA Verified: A Reliable Factuality Benchmark to Measure Parametric Knowledge
• Facts Fade Fast: Evaluating Memorization of Outdated Medical Knowledge in Large Language Models
• AraHalluEval: A Fine-grained Hallucination Evaluation Framework for Arabic LLMs
• Re3: Learning to Balance Relevance & Recency for Temporal Information Retrieval
• Beyond Memorization: Reasoning-Driven Synthesis as a Mitigation Strategy Against Benchmark Contamination
• Beyond Benchmark: LLMs Evaluation with an Anthropomorphic and Value-oriented Roadmap
• Identifying and Answering Questions with False Assumptions: An Interpretable Approach
• A Question Answering Dataset for Temporal-Sensitive Retrieval-Augmented Generation
• MoNaCo: More Natural and Complex Questions for Reasoning Across Dozens of Documents
• Never compromise with vulnerabilities: a comprehensive survey on AI governance
• TiMoE: Time-Aware Mixture of Language Experts
• Beyond Prompt-Induced Lies: Investigating LLM Deception on Benign Prompts
• Harnessing Temporal Databases for Systematic Evaluation of Factual Time-Sensitive Question-Answering in Large Language Models
• FACTORY: A Challenging Human-Verified Prompt Set for Long-Form Factuality

Discussions

• FreshLLMs: Refreshing Large Language Models with Search Engine Augmentation https://arxiv.org/abs/2310.03214v1 (A new paper by Google found that Perplexity and GPT-4 w/ prompting outperformed Google S [bsky, 9 points, 1 comments]
• "we present FreshPrompt, a simple few-shot prompting method that substantially boosts the performance of an LLM on FreshQA by incorporating relevant and up-to-date information retrieved from a search [bsky, 2 points, 0 comments]
• 📝 FreshLLMs: Refreshing Large Language Models with Search Engine Augmentation 📚 "Benchmarks a diverse array of both closed and open-source large language models under a two-mode evaluation procedure [mastodon, 0 points, 0 comments]

Related

• Evaluating Open-Domain Question Answering in the Era of Large Language Models
• LiveSearchBench: An Automatically Constructed Benchmark for Retrieval and Reasoning over Dynamic Knowledge
• Search Engines in an AI Era: The False Promise of Factual and Verifiable Source-Cited Responses
• Internet-augmented language models through few-shot prompting for open-domain question answering
• Winning Big with Small Models: Knowledge Distillation vs. Self-Training for Reducing Hallucination in QA Agents