Nope. I don’t talk about myself like that.
No, not basically no.
https://mashable.com/article/openai-o3-o4-mini-hallucinate-higher-previous-models
By OpenAI’s own testing, its newest reasoning models, o3 and o4-mini, hallucinate significantly higher than o1.
Stop spreading misinformation. The company itself acknowledges that it hallucinates more than previous models.
I would say GitHub copilot ( that uses a gpt model ) uses more Wh than chatgpt, because it gets blasted more queries on average because the “AI” autocomplete just triggers almost every time you stop typing or on random occasions.
They did… You just refuse to acknowledge it. It’s no longer a discussion of simply 3Wh when GitHub copilot is making queries every time you pause typing. It could easily equate to hundreds or even thousands of queries a day (if not rate limited). That fully changes the scope of the argument.
If the premise of your argument is fundamentally flawed, then you’re not having a reasoned debate. You just a zealot.
if you can play an item back. you can enumerate it.
Most people I know that still play DDR just use a program called stepmania (https://www.stepmania.com/download/) and just grabs charts from the official games from https://zenius-i-vanisher.com/.
Stepmania hasn’t had a new official release in a long time… development has slowed down a lot on it. But it’s still good.
https://zenius-i-vanisher.com/v5.2/thread?threadid=11144&page=9#p449014 is a 2022 release against the main branch.
https://projectoutfox.com/ is another project if that’s still too old for you… but that last release looks like it’s mid 2023.
ITGmania is another project which has releases up to march of this year. https://www.itgmania.com/
I mean, that’s effectively the same boat I’m in. I run all my own stuff in my own cluster (recently posted some of it if you check my post history).
But putting up Jellyfin for any user that isn’t on your network is literally a security nightmare. I cannot run blatantly insecure software and leave it internet facing. It’s one thing if it was just found and they’re working on closing it… But this has been documented/known for 4 years. They’re not fixing it and have shown no interest in addressing it at all.
VPN is literally the only answer… and that breaks all TV-based access outright since none of them do VPN. Basic auth doesn’t work. Other forms of auths breaks all app access (leaving only browser). And each time any of these possible alternative answers come up, they’ve outright dismissed it.
If/When Plex finally gets hostile, I’ll simply turn it off. But I can’t let Jellyfin be what services my users, it just doesn’t work.
I’ve spoken out on this same issue before… and as a previous security instructor/researcher… it’s fucking scary how many people shit on Plex for a platform that has had known vulnerabilities in auth for 4 years now, that’s existed since the previous code-base… so at least 7 years old and those issues existed in the previous emby codebase going back over a decade.
Plex isn’t perfect… there’s risks involved there too… but at least when something is brought up as a significant risk it seems to get fixed outside of the implicit risks of the Plex org itself.
All I read in these threads is effectively “WAAAH I don’t WANNA pay!”… Without realizing that the payment gave them something significantly more secure.
h265
https://www.cnx-software.com/2017/10/30/h-265-hevc-license-pricing-updated-for-low-cost-devices/
Lot of companies don’t want to pay it themselves… and lots of people don’t see the point when there’s a list of perfectly capable codecs that are free… including AV1
https://en.wikipedia.org/wiki/List_of_open-source_codecs
Ultimately as a software developer making money… if you don’t license the codecs that you’re using properly (when using a non-FOSS codec) you are liable for damages at that point for violating the terms of the license for the codec. It IS a cost. And across millions of users that costs adds up.
to use your own GPU for hardware encoding was always a scumware tactic
It costs them money to distribute the codec. It’s not scumware. Otherwise they would have to make install/setup of plex a 2 step process… and updates would be annoying as shit.
They need you to pay so they can push codecs with their updates/install.
I’m fine with the one time payment. I donate to shit I use regardless.
It’s been enough to just seed behind a VPN at this scale?
I think that question really hinges on “what vpn provider you choose to use”. Honestly… People don’t hit me as hard as I wish they would.
~72 TB uploaded the past 30 days.
From earlier. ~64TB of that is on my VPN’d hosts. I know it’s a bottleneck by it’s very nature… the highest I’ve seen all 4 peak at the same time was just over 3gbps aggregate…
The VPN I use is a no log vpn… and I choose an exit point that’s outside of my country. So at the very least would require coordination between 2 countries, and a provider that has nothing to give up…
At that point it’s all about the trackers that you’re a part of.
Okay? But they’re not holders… and their access to servers is changing and hinges on YOUR status. It’s not unreasonable to notify them about this change.
earthquakes,
I live in a geo-stable location… Lots of DCs here specifically because of that.
house fire,
Yep. Fire extinguishers are in house, and I check them pretty regularly… Blaze cut is in the junction box(now) and in the rack. Rack is in garage, so I can just unplug and push it the fuck out (assuming I have time after throwing my kids out the windows).
It was actually my setup that alerted me to the junction fire pre-emptively… as power usage was fluctuating wildly and I have Home Assistant alert on that sort of stuff. It only sparked and blew some insulation… The fire didn’t actually move anywhere meaningful and I cut power before it had a chance to get worse. My fire alarm itself is also tied into HASS at this point. So I get alerts on my phone for the rare occasion that I leave my house (most [99%] of my work is work from home, I go to a datacenter probably quarterly at this point for all of an hour).
flood,
We barely get rain :(. But because of the soil here there’s lots of washes and irrigation mitigations in place already.
volcano,
Yellowstone finally going up will just outright kill me… don’t care to mitigate this at that point. There’s no other active volcanos around.
tornado, hurricane,
Doesn’t happen here. Dust devils are about as bad as it gets.
crashing plane,
So there’s 2 ways to look at this one… I live near an AirForce base. Either they fuck up and I’m more at risk… or because the area around the base is no-fly zone, I’m less likely… I don’t know take your pick. I would hope “less likely” due to training… but I’ve seen stupid shit when I was in the military.
theft, nazis, shitty kids, anarchists,
Guns…I have guns. Lots of ammo. A good setup for cameras on my house. And no fucks to give as army training and a deployment has forced onto me. My state is a stand your ground state. If it’s my kids that are the shitty kids… I have a backyard and a shovel (/s).
or rain?
See “flood” above.
Other sources of water (water leak): no pipes above the server rack… Water heater isn’t that close, has a freezer in between to take the brunt of any initial impact, and is brand new (so unlikely to spontaneously explode). Rack is elevated on it’s coasters and garage is graded towards the street.
We can get some interesting thunder storms here. Rack is grounded, I have backup batteries in the rack and whole house battery from the solar. But luck be what it is, in theory that could nail me as unlikely as it is.
Everything is encrypted at rest… Backups are encrypted.
I don’t have a proepr offsite yet. But my cousin is finishing building his house. He also has a fat internet pipe, and I’ll just leave a 25TB node there to backup the important/unreplaceable stuff over there. He’s clear across the country in a pretty geo-stable location as well. And he has an interest in maintaining it as he uses some services that I offer anyway (Email, nextcloud, backups, etc…). Everything else is pretty replaceable and wouldn’t take all that much effort to rebuild otherwise.
Never claimed I was “perfect”… But I’m doing pretty well here and have been doing it for nearly a decade this way. With ~30 years worth of data and very little of it lost. (there was one event over a decade ago at this point… but wasn’t all that bad.)
The real risk is just me doing something stupid since I’m the sole owner and nobody else really knows how to access any of my stuff. My dad has emergency access to my password vault, but even though he’s been programming since the 80’s, a lot of my setup is likely over his head.
All this other stuff is pretty low risk/unlikely or has been relatively decently mitigated.
Edit: Oh… and some certain important items are burned to M-discs and put into my safe every quarter or so.
I don’t think this answers the question
They’re specifically showing you that in the use case you asked about the assertions must change. Your question is bad for the case that you’re specifically asking about.
So no, it doesn’t answer the question… But your question has a bunch more caveats that must be accounted for that you’re just straight up missing.
The two most common reasons I hear are 1) no trust in the companies hosting the tools to protect consumers and 2) rampant theft of IP to train LLM models.
My reason is that you can’t trust the answers regardless. Hallucinations are a rampant problem. Even if we managed to cut it down to 1/100 query will hallucinate, you can’t trust ANYTHING. We’ve seen well trained and targeted AIs that don’t directly take user input (so can’t be super manipulated) in google search results recommending that people put glue on their pizzas to make the cheese stick better… or that geologists recommend eating a rock a day.
If a custom tailored AI can’t cut it… the general ones are not going to be all that valuable without significant external validation/moderation.
I started on cluster (8) of rpi3b and a Synology NAS 10 years ago… prior to that was just storing it all on random hdd (harddrive toaster was alway present and loaded on my desk) and on my computer. these days you can get those little intel n100 or n150 boxes for pretty cheap too. There’s a lot of options, and a lot of mature software tech to make it all work well together.
Uh… Well, no? I mean you can go see for yourself. Click an article. Media is there. https://library.kiwix.org/viewer#wikipedia_en_all_maxi_2024-01/A/User:The_other_Kiwix_guy/Landing
I don’t know exactly what subsets of stuff this is, as I know that wikipedia is about 1/2 petabyte total data all in. But this specific zim export is all English articles with some media. I believe that ALL the English articles is about 50GB and the other 55-ish GB is just media. It’s definitely cutdown, but not devoid.
So about that… I actually got solar installed on my house and when the “electrician” (subbed out contractor with dubious credentials but was operating under a legit company) ran dedicated breakers for the servers… the junction box caught fire.
That was a fun 2am to wake up at.
But it’s pretty safe now. Have had several master electricians come in and evaluate it all at this point and all of them are happy with it now.
More or less exactly this. Most of my servers are company decommission (some companies have a strict upgrade cycle, and I even have like 3 more “spares” sitting around). I’ve found some government liquidation here and there as well which has been great for resale to recoup some of what I spend. I have something like 4TB of Ram sitting in a box somewhere that I should sell somewhere…
Some stuff was “decent” ebay buys. A few items I just had to pony up for. Battery backups was mostly paid at full msrp, but has been worth it. HDDs are mostly “refurb” and built up over time as well, which is why raidz2 and several spares in the zfs pool.
But yeah, about 10 years of building it up. I use it as my playground for professional development, and it’s helped prove a lot of what I say on my resume.
When you have a functional setup like this, and can show that to potential employer. It’s been the best sole investment I’ve ever made. It’s netted me more money in contracts by many many times what I put into it.
Edit: and saved me from countless “cloud” privacy violations, data breaches, etc…
Edit2: Recurring costs is basically energy. The rack itself uses about 90kWh a day (about $5/day for me), cooling and all. My solar install creates just about that much per day as well so that’s all offset (and if I ever move my equipment into a colo, my house will basically have no electric bill at all). Internet is $165 a month, which is fucking great IMO… VPN is paid every 3 years or whatever that cycle is on. If I was to just take the non-replaceable stuff (about 20tb worth of data at last check, and a bunch of lxc containers) and put that on a VPS somewhere I’d be paying at least 10x what I do now in someone else’s datacenter, forget that that’s a recurring monthly cost.
Edit3: Just because it is likely useful information for SOMEONE out there… If you know that your setup takes 10kWh a day, you need to account for another %50 for cooling, so you should actually expect 15kWh in that usecase. My actual rack uses ~60… 30 more is cooling. I actually have all my power usages broken out in a sankey graph… Even goes further to break it down on a per server usage as well, but I can’t take a reasonable screenshot that doesn’t show personal information. This is 24 hours of usage (specifically 04/30, yesterday) and “garage dedicated” is what the A/C unit I have in the bottom of my rack is plugged into.
Fiber, 8gbps through Quantum Fiber (part of the Lumen/Centurylink family). No throttling here…
My bottleneck is the vpn that I have to have between me and the world… unfortunately. I only trust specific private trackers to not use it. I actually setup 4 seed boxes (VMs) in my garage to push more data out.
Proxmox cluster… and big storage truenas node.
And a dirty amount of networking… big cables on the right is QSFP, 40gbps 2x in lagg per server.
These pictures were just after a transplant to a new server rack. So everything is off… but the blinkin’ lights are real.
Edit: Didn’t turn off anything before I ran these…
Yet again… You fundamentally have the wrong answer…
https://en.wikipedia.org/wiki/GitHub_Copilot
https://github.com/features/copilot
GitHub copilot was literally developed WITH OpenAI the creators of ChatGPT… and you can run o1, o3, o4 directly in there.
https://docs.github.com/en/copilot/using-github-copilot/ai-models/changing-the-ai-model-for-copilot-code-completion
It defaults to 4o mini.