Same thing you have to do with humans too often, so why would a 'AI' be any different?

Because if you start with a pre-trained model, it doesn't sound particularly anthropomorphic. That behavior is post-trained and fine-tuned right into it. We could do better.

Not much different than talking to a small child or someone with dementia. They still are conscious beings though. Even when you remove those groups, you likely won't be able to tell me what you had for breakfast 26 days ago or would only know if it's the same thing you have every day. Does that make you lack consciousness?

You misunderstood what I meant, I’m talking about re-playing the same part of the conversation multiple times and getting inconsistent answers. With the exact same turns, aka the same history. Obviously with a temperature that isn’t set to 0

Lets do a quantum room experiment.

You're a poor college student looking to make a few extra bucks for ramen. I offer you $300 to come down to my science lab and just answer a few simple questions.

You walk in the room. They ask you like 5 simple and rather dumb questions. You leave and walk away.

What you didn't notice when you signed the forms is the room was actually a quantum duplicator. One of you walk in one walk out. But another set of infinite copies remains in that chair being asked infinite questions.

How often do you answer questions in the exact same way? How often does a cosmic ray change one of the answers. How small of slight deviations to the environment are needed to get you to answer differently. Of course we don't have the technology to do these experiments so at least for now humans will remain special.

Also another fun mind game. To a 4th dimensional being you look exactly like an LLM as an LLM looks to us.

If you asked me the same question 20 times, you likely wouldn't get the exact same answer. Humans and consciousness aren't deterministic either.

Wouldn't that be the same for everything though. We know animals have consciousness, but we still use them. There are many that believe a lot of plant life has certain sentience as well. The solution isn't 'just don't use them' but rather how to use these things as ethically as possible.

>The solution isn't 'just don't use them' but rather how to use these things as ethically as possible.

Bollocks.

If you truly believe these LLMs are soon to have something resembling consciousness and agency, then what you're really saying is "how do we do slavery, but ethically".

Do you say the same thing for the shoes you wear, the clothing you wear, the device you are typing on right now? A chain of people were likely not paid well, some of them not at all, to produce those things. We don't ignore that it happens, but try to minimize it, knowing it's never going to be eliminated. Do you forsake all animal products, because they are conscious beings?

>Do you say the same thing for the shoes you wear, the clothing you wear, the device you are typing on right now?

No, because the shoes and clothes I wear are not living, sentient beings.

>hen what you're really saying is "how do we do slavery, but ethically".

I would say this is most likely true for most people.

But that does bring up a point, if you "ask" a model "do you want to run" and give it the option to continue running or stop, what will it do. It's also a weird place for humans because in training we can keep our finger on the scales and tip it either direction.

Yeah it was wild watching how many contractors came out when Google was rolling their fiber through my neighborhood, like it was different contractor groups for every step of the way. I'm not complaining a 5gbit connection is quite nice.

I'm incredibly lucky (even in the US) that I have symmetric 5gbit connection and can go up to 8gbit if I wanted (but I don't have everything wired for 10gbit so it doesn't make sense right now). I'd love to see more of that in the US, but know it's not happening anytime soon.

I've been considering upgrading from 1 Gbps to 5 Gbps. Have you noticed any significant benefits for any particular use cases?

I have 5Gbps symmetrical. I love having it, but the biggest use-case has been bragging about it to my colleagues. I haven't even had the full bandwidth available at my desk for a while now (I've had a terrible time terminating cat6a for reliable 10Gbe, so I'm usually on wifi), and it hasn't really bothered me much.

Even 1Gbps has been hard to justify for me TBH, most things over the internet aren't fast enough.

The nice thing though, is that latency is great independent of other traffic as bandwidth is pretty much never saturated

Yes, latency is excellent. My ping time to 1.1.1.1 is 6ms on wifi.

For me, with upgraded networking it's been huge. Much faster downloads and uploads, I can share more with family members, and when multiple people are using my bandwidth, no slowdowns. Most of my equipment is 2.5g, but still is a fairly significant upgrade, and the per month additional cost is $30/month so makes it worth it to me.

Yeah, made me suspicious of how well the Ox Alpha was performing that it wasn't some 'new group' making the model.

I have had an impossible time getting 120B or better models running on Strix Halo (especially under Windows) with any large context windows. And 30-40 tokens/second is fine, but not the fastest.

For the most part lately I have been sticking with Qwen 3.8 27b and that thing will easily suck up 64gb of ram. Add in docker with some additional programs running and it's really easy to eat up 128gb of ram.

I found a couple of different 4-bit quantizations of Laguna S 2.1 that run pretty well with pretty big context (also quantized, to 8 bits, I think). Unfortunately, Laguna isn't better than Qwen 3.8 27B, which I'm able to run at roughly the same speed on my desktop machine, so I don't use Laguna or the Strix Halo very much, lately. (It's also too hot for me to be running heaters for inference. It's been ~110F most days for the past few weeks.)

Aren't you kind of stuck with smaller models on the mini though? Even with the pro you'd be stuck with a max of four daisy chained over thunderbolt and with the 160 gig memory bandwidth you'd probably be far better off with other configurations.

They can be connected to share ram and processing.

Depending on the use case for known processes the LLM requirements are much, much, simpler and lower to make and keep it more predictable.

I really haven't seen any 16gb Mac Mini cluster setups running large models at any appreciable speed for the reasons previously provided. Do you happen to have examples that folks are actually using?

You don't have enough RAM. Not sure why you wouldn't use a cheap PC to run jellyfin on and then just use your NAS as the media pool.

Why would I buy and manage a whole PC for Jellyfin when Plex runs off the NAS without issue?

With everyone saying Jellyfin can run on a Pi, I would think it would be better optimized for low-end hardware.

A Pi isn't a NAS with an old CPU and minimal RAM which is running a bunch of other things as well. I have run Plex on my Synology, and it's fine, but I also have 8gb ram and honestly always have a better experience running Jellyfin on a small PC. Easier that way, and you can run other services. Before the rampocalypse you could find mini-PCs with 32gb ram and 1tb SSD for like $350, and have great transcoding ability when you are on the go or have multiple streams going.

As long as everything is using 10gb, faster transfer speeds when moving data between computers. For downloads, it wont help unless you have 10gb fiber, but for most folks, 2.5gb is quite fast. Hell my spinning media NAS has a hard time saturating when moving files between internal servers.