26 lines
1.0 KiB
Markdown
26 lines
1.0 KiB
Markdown
---
|
||
title: "Foundation Model"
|
||
type: concept
|
||
tags: [AI, generative-AI, ML]
|
||
sources: [public-cloud-learning-sessions-introduction-to-artificial-intelligence-ai-machine-learning-20240206]
|
||
last_updated: 2024-02-06
|
||
---
|
||
|
||
## Summary
|
||
基础模型是具有数十亿参数的大规模预训练模型，能够执行多种任务，是生成式 AI 的核心。
|
||
|
||
## Definition
|
||
Foundation Model（基础模型）是经过大规模预训练的大语言模型，具备零样本和少样本学习能力，可通过微调适应各种下游任务。
|
||
|
||
## Key Attributes
|
||
- **参数规模**：数十亿参数
|
||
- **训练方式**：大规模预训练
|
||
- **特点**：零样本学习、少样本学习、多任务能力
|
||
- **代表模型**：Amazon Titan、GPT-4、Claude、Llama
|
||
|
||
## Connections
|
||
- [[AI]] ← powers ← [[Foundation Model]]
|
||
- [[Generative AI]] ← powered_by ← [[Foundation Model]]
|
||
- [[Amazon Bedrock]] ← hosts ← [[Foundation Model]]
|
||
- [[Foundation Model]] ← supports ← [[Fine-Tuning]]
|
||
- [[Foundation Model]] ← supports ← [[RAG]] |