mds

After getting ragebaited by LangChain for a while, I’m happy to announce that I have finally completed the LLM moderation feature. Now you can finally rest assured that the memes being sent are appropriate! It is off by default, since you obviously need an llm endpoint to use it, but it is nice to have! Later I may add a setting to control how strict the moderation is. I additionally improved the overall reliability of the app, and made it easier to use as well as easier to understand what’s going on!

LLMs, the cause of the ram shortage, are now a vital (kinda) part of my app. So basically in the about 2 hours ish of this devlog, I managed to implement the first half of the LLM moderation & ranking feature. I first using the OpenAI API, but then switched to langchain to allow support for other API specifications! But yeah, this feature is pretty time consuming to implement… Also, the screenshot this time is just some code as I don’t wanna add yet another screenshot of the app, with the LLM feature not even implemented yet.

Waiter waiter, more upgrades! Alright, so before I jump to the complicated stuff, I just improved the overall code structure by making it more readable and spread apart, added better logging capabilities, as well as added the core feature: scheduled meme posting, and finally, improved the source links button & overall reliability.

Finished the basic structure of the Slack bot; I got it working and posting memes along with a basic scraper, and a sources button. This is a very basic demo for now, there is so much more to come!