Files
nexus/wiki/concepts/数据蒸馏.md
2026-04-18 12:03:16 +08:00

23 lines
821 B
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
---
title: "数据蒸馏"
type: concept
tags: [ai, 数据蒸馏, 模型压缩]
date: 2025-12-20
---
## Definition
数据蒸馏Data Distillation利用一个高性能的大模型生成精简但有价值的数据使得一个小模型可以从中学习并逼近大模型的效果。
## Core Mechanism
1. **教师模型Teacher Model**:高性能的大模型,负责生成高质量数据
2. **数据生成**:教师模型产出精简但有价值的数据
3. **学生模型Student Model**:小模型从蒸馏数据中学习
## Use Cases
- 模型压缩:将大模型知识迁移到小模型
- 领域适应:在特定领域数据上微调
- 成本优化:推理效率与效果的平衡
## Related Concepts
- [[LLM]]:教师模型和学生模型的源头
- [[vLLM]]:高效的模型推理引擎