Classy-AI 0.0.15

A python module that utilizes AI to classify natural language into categories for processing with other models.

Classy

How to use

Install:

pip install Classy-AI

Setup: Only do this if you plan to run the model locally

import Classy
Classy.download()

Downloads "data.pth" to your working directory.

Usage (Complete version for a server):

import Classy
Classy.init('/path/to/data.pth','OpenAI api key')
primary, secondary, stemmed, organized = Classy.server("hello, how are you?")

Usage (Just the classifier)

import Classy
intent, certainty = Classy.classify("hello, how are you?",'/path/to/data.pth')

API

Usage

import Classy
intent, certainty = Classy.classify_api("hello, how are you?")

Go to http://app.chat314.com/api to learn about how to use the api yourself.

Other tools

GPT:

Classy.chat_gpt(input,model,api_key)

DALL-E:

Classy.dall_e(input,model,size,api_key)

Search:

Classy.search(input,model,api_key)

Personal:

Classy.personal(input)

Example:

import Classy
api_key='sk-proj-api-key'
while True:
    var=input(': ')
    output,prob= Classy.classify_api(var)
    final=''
    secondary=''
    if output == 'GPT' and prob >= 0.7:
        final=Classy.chat_gpt(var,"gpt-4o-mini",api_key)
    elif output == 'Dall-e' and prob >= 0.7:
        final=Classy.dall_e(var,"dall-e-3","1024x1024",api_key)
    elif output == 'Search' and prob >= 0.7:
        final=Classy.search(var,"gpt-4o-mini",api_key)
    elif output == 'Personal' and prob >= 0.7:
        final,secondary,x,y=Classy.personal(var)
    print(final)
    if secondary:
        print(secondary)

I need more training data

Open a pull request if you have some and put it in unorganized data. I will format it. If you have organized data, put it directly in the intents.json file.

An easy way to contribute is by scrolling through your ChatGPT conversations/Google searches/DALL-E inputs and pasting some of your own data into the unformatted data file.

Deployed here:

https://chat314.com

Uses models gpt-4o-mini and dall-e-3.

I can add a custom password for you on the website if you add training data. Message me at [e-mail]. Put in the description of the pull request that you want a custom password for your Openai key.

Thanks

Thanks to Patrick Loeber for teaching me PyTorch