EngineeringUpdated 2026-10-24
Fine-tuning
Continue training a model on task data to better match a style or task.
#微调#训练#定制
In plain language
Fine-tuning suits stable formats, domain tone and specific tasks, not frequently changing knowledge. Full fine-tuning is expensive; parameter-efficient methods like LoRA are common. Validate with prompting and RAG first.
Example
Fine-tune on 5,000 historical tickets so replies match the brand voice.
Recommended tools
Related terms
Models & Architecture
Low-Rank Adaptation
Freezing the base model and training a small low-rank adapter for a new task.
3 related terms
Engineering
RAG vs Fine-tuning
RAG handles changing knowledge; fine-tuning handles behavior and style customization.
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