# Outlines — Structured Output from Any LLM

> Generate structured JSON, regex-constrained text, and type-safe outputs from any LLM. Outlines uses guided generation to guarantee valid structured responses.

## Install

Paste the prompt below into your AI tool:

## Quick Use

```bash
pip install outlines
```

```python
import outlines

model = outlines.models.transformers("mistralai/Mistral-7B-v0.1")

# Generate valid JSON matching a schema
generator = outlines.generate.json(model, {"type": "object", "properties": {"name": {"type": "string"}, "age": {"type": "integer"}}})
result = generator("Extract info: John is 30 years old")
print(result)  # {"name": "John", "age": 30}
```

## What is Outlines?

Outlines is a library for structured text generation from LLMs. Instead of hoping the model outputs valid JSON, Outlines guarantees it through guided generation — it constrains the model's token sampling to only produce tokens that match your schema. Works with any open-source model via transformers, llama.cpp, vLLM, or MLX.

**Answer-Ready**: Outlines guarantees structured LLM outputs through guided generation. Supports JSON schemas, regex patterns, Pydantic models, and grammars. Works with any open-source model. No retries needed — output is always valid. 10k+ GitHub stars.

**Best for**: AI engineers needing reliable structured extraction. **Works with**: HuggingFace transformers, vLLM, llama.cpp, MLX. **Setup time**: Under 2 minutes.

## Core Features

### 1. JSON Schema Generation

```python
from pydantic import BaseModel

class Character(BaseModel):
    name: str
    age: int
    weapon: str

generator = outlines.generate.json(model, Character)
character = generator("Create an RPG character named Aria")
# Always returns valid Character object
```

### 2. Regex-Constrained Generation

```python
# Generate valid email addresses
email_gen = outlines.generate.regex(model, r"[a-z]+@[a-z]+\.[a-z]{2,3}")
email = email_gen("Generate an email for support")

# Generate dates in specific format
date_gen = outlines.generate.regex(model, r"\d{4}-\d{2}-\d{2}")
date = date_gen("When was Python created?")
```

### 3. Choice (Classification)

```python
classifier = outlines.generate.choice(model, ["positive", "negative", "neutral"])
sentiment = classifier("This product is amazing!")
# Always one of the three options
```

### 4. Grammar-Based Generation

```python
# Generate valid SQL
sql_grammar = outlines.grammars.sql
sql_gen = outlines.generate.cfg(model, sql_grammar)
query = sql_gen("Write a query to find users older than 25")
```

### 5. Multiple Backend Support

```python
# HuggingFace Transformers
model = outlines.models.transformers("mistralai/Mistral-7B-v0.1")

# vLLM (fast serving)
model = outlines.models.vllm("mistralai/Mistral-7B-v0.1")

# llama.cpp (CPU/Metal)
model = outlines.models.llamacpp("model.gguf")

# MLX (Apple Silicon)
model = outlines.models.mlxlm("mlx-community/Mistral-7B-v0.1-4bit")
```

## Outlines vs Alternatives

| Feature | Outlines | Instructor | LMQL |
|---------|----------|------------|------|
| Guaranteed valid output | Yes | Retry-based | Yes |
| Works offline | Yes | API only | Partial |
| Regex constraints | Yes | No | No |
| Grammar support | Yes | No | Yes |
| Speed overhead | Minimal | Retry cost | Moderate |

## FAQ

**Q: Does it work with Claude or GPT-4?**
A: Outlines is designed for open-source models where you control token sampling. For API models, use Instructor (retry-based) instead.

**Q: How does guided generation work?**
A: Outlines builds a finite-state machine from your schema/regex and masks invalid tokens at each generation step, ensuring only valid tokens are sampled.

**Q: Does it slow down generation?**
A: Minimal overhead. The FSM is precompiled, and token masking is O(1) per step.

## Source & Thanks

> Created by [.txt](https://github.com/dottxt-ai). Licensed under Apache 2.0.
>
> [dottxt-ai/outlines](https://github.com/dottxt-ai/outlines) — 10k+ stars

<!-- ZH -->

## 快速使用

```bash
pip install outlines
```

保证 LLM 输出符合 JSON Schema、正则表达式或语法规则。

## 什么是 Outlines？

Outlines 通过引导生成保证 LLM 结构化输出。在 token 采样时约束模型只产生符合 schema 的 token，无需重试。

**一句话总结**：引导生成保证 LLM 结构化输出，支持 JSON Schema/正则/语法约束，适配所有开源模型，零重试零错误，10k+ stars。

**适合人群**：需要可靠结构化提取的 AI 工程师。

## 核心功能

### 1. JSON Schema 生成
Pydantic 模型或 JSON Schema 定义输出格式，保证有效。

### 2. 正则约束
用正则表达式约束输出格式（邮箱、日期、ID 等）。

### 3. 分类
限定输出为预定义选项之一。

### 4. 多后端
支持 transformers、vLLM、llama.cpp、MLX。

## 常见问题

**Q: 支持 Claude/GPT-4？**
A: Outlines 面向开源模型。API 模型用 Instructor（重试模式）。

**Q: 引导生成原理？**
A: 从 schema 构建有限状态机，每步采样时屏蔽无效 token。

## 来源与致谢

> [dottxt-ai/outlines](https://github.com/dottxt-ai/outlines) — 10k+ stars, Apache 2.0

---
Source: https://tokrepo.com/en/workflows/ff70f132-33a5-4daa-a974-a96c2fe28a0e
Author: Prompt Lab