rebelai

free access to multiple proprietary ai models in python

mission

exploit freely available apis to gain access to proprietary ai models because proprietary ai companies took advantage of our open source resources to develop their proprietary money machines unfairly

( fuck you """""""""""""""""""""""""""""open"""""""""""""""""""""""""""""ai )

example usage

https://ari-web.xyz/gh/124 -- my discord bot for my friend group server ( https://discord.gg/PqcseUzDGF ), this is the main project i wanted to use this library in :3

installation

$ pip install rebelai

usage

module structure :

rebelai -- top level package
- const -- all constants, defaults, etc
- enums -- all enums for convenience sake
  - ProdiaModel -- enum for prodia image generation ai models
  - ProdiaSampler -- enum for prodia image generation samplers
  - DeepAIModel -- enum for deepai models
- ai -- all ai wrappers
  - alpaca -- access to alpaca models
    - alpaca_7b -- the 7 billion neuron alpaca model
  - deepai -- access to all deepai models
    - deepai -- generic function for deepai access for free, takes DeepAIModel, see doc string for more info
  - gpt -- access to "open"ai models ( kinda )
    - gpt3 -- access to gpt3 model
    - gpt4 -- access to gpt3 model ( youchat )
  - pollinations -- access to pollinations image generation api
    - pollinations -- access to the generation model
  - prodia -- access to prodia image generation api ( limited access and proxies dont seem to work :( )
    - prodia -- generic function to access prodia api, takes ProdiaModel and ProdiaSampler, see doc string
  - inferkit -- access inferkit text completion model
    - standard -- standard model access
  - deepl -- access to deepl translation
    - deepl -- generic deepl function, see doctring for help
  - h2o -- access to h2o models
    - falcon_40b -- access to the falcon model with 40 b neurons ( open source )

tips

use proxies ( for example gimmeproxy api,, https://gimmeproxy.com/api/getProxy?post=true&get=true&user-agent=true&supportsHttps=true&protocol=http&minSpeed=20&curl=true )
- dont forget to test if theyre responsive by making a proxies request to for example https://example.com/
check docstrings, they have info about arguments and proxies