Make Discord your LLM frontend

[tame path]

I created "llmcord.py" which aims to make Discord a universal LLM frontend.

Works with oobabooga! Over 200 300 GitHub stars and counting.

https://github.com/jakobdylanc/llmcord.py

[tame path]

Just added a new config option use_plain_responses (default: false). I personally keep it on false 🙂

Here's a comparison (false vs true):

Make Discord your LLM frontend

[tame path]

My latest commit reduces python dependencies down to just discord.py and openai while fully maintaining oobabooga support. A bit trickier than you'd expect, specifically with allowing custom API parameters.

https://github.com/jakobdylanc/discord-llm-chatbot

[ashen rapids]

This is cool, is it still being worked on?

[tame path]

Yes, still in active development 👍

It’s pretty perfected in its current state. I have further ambitions for features that will be realized eventually 😁

[ashen rapids]

Nice, I'll take a look at it then!

[tame path]

Cool, let me know any feedback!

[ashen rapids]

Will do

[ashen rapids]

There is an issue where if it cant find Assistant in ooba then everything breaks, i use custom named people... not sure what exactly ive done to break it minus getting rid of the Assistant character itself

[tame path]

are you talking about characters?

if I understand you correctly and you want to use a character, you can accomplish this by adding the following entries to "extra_api_parameters" in the config.json:

"mode": "chat"
"character": "Name"

(replace "Name" with your character's name)

[ashen rapids]

Yep, and will try this, thank you

[tame path]

New update dropped!

Now, any changes you make to your config.json settings (like swapping to a new model, changing the system prompt, etc.) are active INSTANTLY! No need to restart the bot.