rekaslop

rekaproxy

Set your frontend's source to OpenAI, proxy URL to https://adasdasdasdas-reka-proxy.hf.space/reka/v1, context size to 4k (yes, chat supports only up to 4k context) and max response length to something short and enjoy.

Supports both streaming and non-streaming.
Post any bugs/issues to thread, I will try to fix them.
Only meant for seeing its "capabilities", dont actually use it.
Temperature and other params doesnt effect anything.

Known issues:

• Sometimes the model ends responses with <sep, wont fix it, write regex or something.

localslop - How to setup

1. Install the rekaslop from catbox
2. Unzip it, open the CMD and type npm install
3. Go to reka.ai and create account.
4. Once you are in playground/chat open DevTools and go to network tab.
5. Set the filter to Fetch/XHR (optional) and send a dummy message.
6. Look for the https://chat.reka.ai/api/chat URL in the network tab, click on it and look for Authorization header, excluding the Bearer part, copy all of it.
7. Open the auth.txt file and paste your auth key.
8. Go back to CMD and type node index.js.
9. Open SillyTavern, set the source to OpenAI, proxy URL to http://localhost:3000/v1 and password to anything you want.
10. Set the context size to 4k (yes, chat supports only up to 4k context) and max response length to something short and enjoy.