The PC I’m using as a little NAS usually draws around 75 watt. My jellyfin and general home server draws about 50 watt while idle but can jump up to 150 watt. Most of the components are very old. I know I could get the power usage down significantly by using newer components, but not sure if the electricity use outweighs the cost of sending them to the landfill and creating demand for more newer components to be manufactured.

Cool. A few years ago I made similar sensors (but just PIR, LUX, esp8266 board, jumper wires, and some with ultrasonic proximity). Attached them to the wall with those 3m velcro things. ESPHome makes things so easy; didn’t even have to write code. Hardest thing was designing and iterating over the case to get the tolerances just right.

I’m loading up on vacuum tubes.

Last time I looked it up and calculated it, these large models are trained on something like only 7x the tokens as the number of parameters they have. If you thought of it like compression, a 1:7 ratio for lossless text compression is perfectly possible.

I think the models can still output a lot of stuff verbatim if you try to get them to, you just hit the guardrails they put in place. Seems to work fine for public domain stuff. E.g. “Give me the first 50 lines from Romeo and Juliette.” (albeit with a TOS warning, lol). “Give me the first few paragraphs of Dune.” seems to hit a guardrail, or maybe just forced through reinforcement learning.

A preprint paper was released recently that detailed how to get around RL by controlling the first few tokens of a model’s output, showing the “unsafe” data is still in there.

I think TikTok appeased the right by changing their algorithm. Charlie Kirk is apparently doing extremely well on the platform now.

You can also just install the libreelec os (Kodi), and install the Jellyfin Kodi addon. Haven’t tried that addon. I used to use Kodi when I had only one TV, and liked it. Now that I have 2 Android TVs, just installing Jellyfin on the TVs works fine. I might go back to rPIs and disconnect my TVs from the internet though.

I use GPT (4o, premium) a lot, and yes, I still sometimes experience source hallucinations. It also will sometimes hallucinate incorrect things not in the source. I get better results when I tell it not to browse. The large context of processing web pages seems to hurt its “performance.” I would never trust gen AI for a recipe. I usually just use Kagi to search for recipes and have it set to promote results from recipe sites I like.

Shaq’s best performance.

Tor for browsing is similar to a VPN. I2p and Tribbler for downloads is also similar. You could also just rent a cheap VPS and set up your own VPN. There’s a high chance people will be doing illegal shit through a VPN-like services, so I don’t think a p2p VPN-like service where everyone is like an exit node is viable.

Chevy Bolt and Nissan Leaf come to mind.

Hmm. I just assumed 14B was distilled from 72B, because that’s what I thought llama was doing, and that would just make sense. On further research it’s not clear if llama did the traditional teacher method or just trained the smaller models on synthetic data generated from a large model. I suppose training smaller models on a larger amount of data generated by larger models is similar though. It does seem like Qwen was also trained on synthetic data, because it sometimes thinks it’s Claude, lol.

Thanks for the tip on Medius. Just tried it out, and it does seem better than Qwen 14B.

Larger models train faster (need less compute), for reasons not fully understood. These large models can then be used as teachers to train smaller models more efficiently. I’ve used Qwen 14B (14 billion parameters, quantized to 6-bit integers), and it’s not too much worse than these very large models.

Lately, I’ve been thinking of LLMs as lossy text/idea compression with content-addressable memory. And 10.5GB is pretty good compression for all the “knowledge” they seem to retain.

My friend just hooks his laptop up to his TV, connects to his VPN, and plays popcorntime (streaming torrents). He used to use streaming sites, but those have been getting taken down left and right.

I think I’ve heard there are a lot of genetically male, but born female people in sports. I wonder if the same people are against those people playing in sports.

Idk how many transphobic people just care about specific issues. There’s a lot of “groomer” rhetoric, hate, and general disgust. It’s easy to get people to hate what they don’t understand; and a lot of media is trying their hardest to cultivate hate against trans people to create an out-group, so they can control the in-group.

Meh, I would’ve given 3/5 stars to U.S. democracy since the Voting Rights Act. Stars taken away for FPTP, gerrymandering, campaign finance, “lobbying,” and the electoral college. I believe we’re going to go to 0/5 stars with completely rigged elections rather than just manufacturing consent and lightly tipping the scales like they’ve been doing.

Yeah, I think this could be the end of free and fair elections in the U.S., and there’s no coming back from that without a revolution. Don’t get me wrong, I don’t think most of us will directly be killed by this change; our lives will just be shittier. It’ll be like living in Russia. Given how utterly incompetent the administration is looking, and the things they say they’re going to do (mass deportation of a significant part of our workforce, blanket tariffs, gutting social safety-nets), we may speed-run an economic and societal collapse. That could sow the seeds for a horrible and bloody revolution.

Or, maybe I’m wrong and the important institutions will somehow hold against a christo-fascist party controlling all branches of the federal government and a president with immunity. If there are still are free and fair elections, then congress could block a lot of things in 2026, and start repairing some of the damage in 2028.

Still, it does not bode well that the U.S. elected these people in the first place, and at best, the U.S. will slowly crumble for decades.

Do you remember when it was commonly advised to use fake names and birthdays on online forms, and when “spyware” was a term?

Find your local state delegate.

Would this be “State Democratic Party chairs?” I’m not really sure how the party is structured and who votes. And apparently, the chair for my state resigned.

I haven’t checked it out in years. From my understanding, IPFS aims to be a distributed filesystem that kinda works like Bittorent. If you access a file, you then seed it. Last time I checked it out, the project was jumping on the crypto bandwagon… Just checked out their website now, and don’t know WTF it is.