FundamentalsFeaturedUpdated 2026-10-24
Hallucination
When a model produces plausible but factually wrong or unsupported content.
#风险#事实性#RAG
In plain language
Hallucinations arise because models generate by probability rather than verification. Common forms include invented citations, wrong numbers and nonexistent people or events. Mitigations include retrieval-augmented generation, requiring sources and human review of key claims.
Example
The model invented a paper title and authors that do not exist.
Recommended tools
Related terms
Fundamentals
Benchmark
A standardized dataset and metric for comparing model capabilities.
3 related terms
Fundamentals
Model Evaluation
Designing test sets and scoring rules to decide whether a model works for a specific task.
3 related terms
EngineeringFeatured
Retrieval-Augmented Generation
Retrieve relevant passages from a knowledge base, then have the model answer from them.
4 related terms
Applications
Agentic Workflow
Breaking a task into steps planned by AI, executed by tools and reviewed by humans.
3 related terms