Typechat.py

A python implement to makes it easy to build natural language interfaces using types.

Install

pip install typechatpy

Usage

see simple and more in example.

from typing import List
from pydantic import BaseModel
from typechatpy.translator import Translator

class VenueData(BaseModel):
    venue: str
    description: str

class Response(BaseModel):
    data: List[VenueData]

prompt = "Provide 3 suggestions for specific places to go to in Seattle on a rainy day."

def main():
    t = Translator()

    # manual set
    res = t.generate(prompt, Response, VenueData)

    # auto analyse
    res = t.generate(prompt, auto=True)

    # set from globals
    res = t.generate(prompt, *globals().values())

    print(res)

if __name__ == "__main__":
    main()

Example

see simple for more detail (the response sample as bellow).

{
  "data": [
    {
      "venue": "Seattle Art Museum",
      "description": "Explore the extensive collection of art from around the world at the Seattle Art Museum. From contemporary art to ancient artifacts, there is something for everyone to enjoy."
    },
    {
      "venue": "Pike Place Market",
      "description": "Indulge in a unique shopping experience at Pike Place Market. Browse through local produce, crafts, and specialty shops, and enjoy a variety of delicious food options."
    },
    {
      "venue": "Chihuly Garden and Glass",
      "description": "Marvel at the stunning glass artworks created by Dale Chihuly at the Chihuly Garden and Glass exhibit. The vibrant colors and intricate designs are sure to captivate your senses."
    }
  ]
}

TODO

• translator
• validator
• llm interact

Ref

• pydantic
• typechat