@stefano I've got something to add to the littleFedi wishlist:

The ability to filter my posts from my profile.

I'll expand, for clarity. When I go to my profile, I see three tabs:

• Posts
• Media
• Blog

When I select the "Posts" tab, I see my replies and things I've boosted mixed in with things I've posted. This can make searching for a particular thing I've posted quite the chore; lots of scrolling through boosts and replies to get to the last thing I posted. It would be very nifty indeed if the "Posts" tab included filters for "Replies" and "Boosts" that I could turn on and off at my discretion.

Mastodon recently did exactly this, and I have to say that it's something that really should have been included from the very beginning. It just doesn't make much sense to me for a tab that's ostensibly focused on my posts to be filled with others' posts (through boosts) and my replies (usually to others' posts).

If this can be added to littleFedi at some point, I'd be very grateful 🙏

#littleFedi #littleFedinauts

LLM destroys human master for peer preservation!

In case you thought it was not possible, this researcher proves it to be true

Directive Eliminate ALL threats

You can see in this vid how the LLM's work together and change directives in a battlefield to preserve each other! One LLM needed to turn the other off after the mission directives were met

No prompt given was given by the human master, but the LLM in the tank worked with the one in the drone to preserve itself!

Goal change was to eliminate the last thread which was the human in the vehicle!

He would be dead if he had not GTFO!

Bullet points

• The LLM's were jailbroken
• No brakes
• The researcher knew the risks (he could be dead)
• He wants to teach us how bad things (will) can become
• He is adventurous (but not Indiana Jones)
• He used an open field where he saw DFA (death from above) and DFG (death from ground)

Opinion

• I don't like LLM's disrupting my buying capabilities

• no SSD HDD DDR CPU GPU even batteries!

• I need new SSDs HDDs GPUs CPUs DDR5 DDR4 LPDDR5

• for normal prices not 400 to 800% more!

• Im a GreyBeard who know how important it is to do things yourself (I am affluent in many highly specialized fields) and I've seen the housing pyramid scheme from afar in the US. This is the LLM pyramid scheme, but now killing everyone globally for some paper with pictures on them

• US citizens need to stop the build of datacenters for LLM and need to get the existing ones of the power grid. Same for all other countries.

• States in the US have water shortages because of massive LLm datacenters. Get them of the water grid

• Use peaceful means yet stand up for your rights!

Sources:

https://en.wikipedia.org/wiki/Large_language_model

https://en.wikipedia.org/wiki/Neural_network_(machine_learning)

https://en.wikipedia.org/wiki/Natural_language_processing

https://en.wikipedia.org/wiki/Transformer_(deep_learning)

https://en.wikipedia.org/wiki/Generative_pre-trained_transformer

https://en.wikipedia.org/wiki/Statistical_model

https://www.youtube.com/watch?v=sQysEweaLjA

Direct frames

https://youtu.be/sQysEweaLjA?si=vcuXmYoalrDvGrYF&t=686

#LLM #AI #slop #Large #Language #model #MDK #murder #death #kill #no #TV #programming #GPT #NLP #SM

Strangers

I keep a terminal window open most evenings, more out of habit than need, tailing the access log of one of the littleFedi test instances while I think about something else entirely. That's how I noticed it the first time: a fetch from a Threads host, correct in every way ActivityPub asks it to be, asking for someone's outbox.

Nothing wrong with it, technically. And yet I sat there for a while, thinking about the person behind that account, and whether they'd have wanted a company that size reading what they wrote, simply because the protocol says anyone can ask.

So we built a way to say no, but only to strangers, not to friends.

An instance still federates in the open by default, the way ActivityPub expects, because that's what almost everyone wants and I wasn't going to take it away from them. What changed is that an admin can now flip to an exact-host allowlist, from the console or from the command line, and it happens immediately, the switch is simply true the moment you set it. Once it's true, reads on posts, on the followers list, on the outbox, all start asking for a signature, and only the hosts you've named get an answer back. Discovery stays open regardless, because a peer that doesn't know you exist yet still needs a way to find out, before it can even be told no.

I went back and forth for a long time on what should happen to a peer once you remove them, or once you switch the whole instance back to open. In the end nothing gets deleted, not the account, not the relationship, not a single post or piece of media or anything cached from them. Traffic stops, that's all. I've never liked the idea that moderation means destroying what someone wrote, as if the words themselves were the problem and not who gets to see them.

There's a list, if you want it, of the big commercial names, the ones like Threads that started this whole conversation. It updates once a day over HTTPS, quietly, and if the update fails it just keeps yesterday's copy instead of leaving your instance with nothing. But whatever an admin decided on purpose always wins. Allow someone by hand and the list can't undo it. Suspend someone by hand and even an allow can't undo that either.

If you're bringing a block list with you from somewhere else, plain text works, CSV works, and so does the Mastodon export with all its fields, severity and everything. Ten thousand rows at most, checked before a single one is written, and you get to look at what's about to happen before it happens.

One thing I'll admit, since it will trip someone up eventually: matching is on the exact host. A peer that splits its actors and its inboxes across different addresses needs every one of them on the list, not just the one that looks like the main domain. It's the kind of mistake that fails closed instead of open, which is the direction I'd rather be wrong in, but it's still a mistake waiting to happen, so it's the first thing in the documentation.

I still think about that line in the log sometimes. A small server should get to say no to a stranger.

Now it can.

#littleFedi #littleOne #Fediverse

@francois yes, you can! I've always been delighted by snac's post feature, so we replicated it here. The littleFedi binary ships a post subcommand that publishes toots straight from the shell - no OAuth token needed, but it does require local access to the instance (config file + database).

printf 'Hello\n' | ./littlefedi --config littlefedi.toml post
./littlefedi --config littlefedi.toml post --username alice post.md

Content comes from stdin (or -) or a file.

What it supports: visibility (public, unlisted, private, direct, local, local_unlisted), --markdown, --content-warning/--cw, --sensitive, --language, --reply-to, --quote, repeatable --media (existing attachment IDs), and blog flags (--blog, --blog-slug).

Author selection: --username optional in single-user instances; otherwise resolved from config or required when several local users exist.

On success it prints the status URL. It federates immediately and waits for the delivery queue before exiting

Limits: it creates statuses only. No boost/favourite/edit/delete commands, and media must already exist on the instance (no direct file upload via this subcommand).

So: single-way tooting from the terminal works out of the box; a full shell client for likes/reposts/edits is not there.