Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/chandralegend/semantix
Non-Pydantic, Non-JSON Schema, efficient AutoPrompting and Structured Output Library
https://github.com/chandralegend/semantix
autoprompt llm meaning-typed python semantic-typing structured-output
Last synced: about 1 month ago
JSON representation
Non-Pydantic, Non-JSON Schema, efficient AutoPrompting and Structured Output Library
- Host: GitHub
- URL: https://github.com/chandralegend/semantix
- Owner: chandralegend
- License: mit
- Created: 2024-09-02T17:05:49.000Z (3 months ago)
- Default Branch: main
- Last Pushed: 2024-10-16T11:22:20.000Z (about 1 month ago)
- Last Synced: 2024-10-18T03:27:45.815Z (about 1 month ago)
- Topics: autoprompt, llm, meaning-typed, python, semantic-typing, structured-output
- Language: Jupyter Notebook
- Homepage: https://chandralegend.github.io/semantix/
- Size: 482 KB
- Stars: 21
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Contributing: docs/contributing.md
- License: LICENSE
- Roadmap: docs/roadmap.md
Awesome Lists containing this project
README
[](https://pypi.org/project/semantix/) [](https://colab.research.google.com/github/chandralegend/semantix/blob/main/try.ipynb) 
Semantix provides a simple but powerful way to infuse meaning into functions, variables and classes to leverage the power of Large Language models to generate structured typed outputs `without the need of JSON Schema or any other abstractions.`
## Key Features:
- **Semantic Type**: Add meaning to your variables. No need of additional abstractions like `InputField`, `OutputField` etc.
- **AutoPrompting**: Semantix Generate prompts using the `Meaning Typed Prompting` Technique.
- **Supercharged Functions**: Automatically augment functions enhance-powered capabilities. No Function body is needed.
- **Minimal Overhead**: Seamlessly integrate into existing Python codebases with minimal overhead.
## Minimal Example
```python
from enum import Enum
from dataclasses import dataclass
from semantix import Semantic, enhance
from semantix.llms import OpenAI
from semantix.types import Image
llm = OpenAI()
class Personality(Enum):
"""Personality of the Person"""
INTROVERT = "Introvert"
EXTROVERT = "Extrovert"
@dataclass
class Person:
full_name: str
yod: Semantic[int, "Year of Death"]
personality: Semantic[Personality, "Personality of the Person"]
@enhance("Get Person Informations use common knowledge", llm)
def get_person(name: Semantic[str, "Name of the Person"]) -> Person:
...
person_obj = get_person(name="Albert Einstein")
print(f"{person_obj.full_name} is an {person_obj.personality.value} who died in {person_obj.yod}")
# Albert Einstein is an Introvert who died in 1955
```
## Supports Vision
```python
from semantix.types import Image
@enhance("Get Person Informations use common knowledge", llm)
def get_person(img: Semantic[Image, "Image of the Person"]) -> Person:
...
person_obj = get_person(img=Image("mandela.jpg"))
print(f"{person_obj.full_name} is an {person_obj.personality.value} who died in {person_obj.yod}")
# Nelson Mandela is an Extrovert who died in 2013
```
## Installation
All you need is:
```bash
pip install semantix
```
To install the very latest from `main`:
```bash
pip install git+https://github.com/chandralegend/semantix.git
````
Or open our intro notebook in Google Colab: [
](https://colab.research.google.com/github/chandralegend/semantix/blob/main/try.ipynb)
By default, Semantix doesn't install any llm packages. You can install them separately:
```bash
pip install semantix[openai]
pip install semantix[anthropic]
pip install semantix[openai, anthropic] # Install both
```
If you want to use MultiModal capabilities, you can install the following:
```bash
pip install semantix[image]
pip install semantix[video]
```
## Citation
If you find Semantix helpful, give it a ⭐️ on [GitHub](https://github/chandralegend/semantix)!
and If you have used Semantix in your project, add the badge to your README.md file.
```markdown
[](https://github/chandralegend/semantix)
```
If you used Semantix in your research, please cite it as follows:
```bibtex
@misc{semantix,
author = {Chandra Irugalbandara},
title = {Semantix: Infusing Meaning into Code with Large Language Models},
year = {2024},
publisher = {GitHub},
journal = {GitHub Repository},
howpublished = {\url{https://github/chandralegend/semantix}}
}
```
## Contributing
Please read [CONTRIBUTING.md](docs/contributing.md) for a quick guide on how to contribute to Semantix.