Post

would be cool if you could deputize your mod service to hide replies in your threads for everyone, even those not subscribed. similar to a threadgate but you select a mod service to deal with it instead.

like if it's a bad faith reply, just hide for everyone? that'd be neat

yeah, and the person in question sees the label (even if they're not subscribed), sees that the OP was the one to choose that moderation service and those labels, and they can appeal the label to the mod service (but not to the OP). a protective barrier, or a community even?

Would it be less wasteful to use ibm granite docling huggingface.co/ibm-granite/...

What @ do you have it running under right now 👀

What’re you running it on?

framework desktop

good stuff. does this just use the completions api, i.e. can i point it at any completions (openai) compatible inference endpoint?

yup. should probably just make the `model` param an environment variable so you can use whatever

I don't have the expertise to do it, but I really want to make a feed that ingests all posts, looks for a bunch of keywords of interest to me from follows, all of my lists, and follows of follows (or similar), and then uses an LLM to read through each one and determine whether it fits my interests..

bad faith is one of the biggest problems in society right now, so this could be the start of something huge. I would love to see all my social feeds and group chats scrubbed of bad faith garbage

i agree. the bad faith stuff on this website is just getting out of control honestly.