halyconWays

7 ms·

61.
by halyconWays 1y ago
Thanks. If you have experience with the Quest Pro would you say the text clarity is a step up with the Ones? Supposedly the One Pros will be even better, and are coming out soon.
62.
by halyconWays 1y ago
I tried it for a while with the best AR glasses I could find at the time, XReal Air 2 Pros with an Xreal Beam, and although I could see the potential, it wasn't good enough to get work done. The screen size was too small, the resolutio
63.
by halyconWays 1y ago
There's massive evolutionary pressure for maximizing energy efficiency in brains. I'd like to see LLMs procreate and select for energy efficiency while, ideally, minimizing insanity and maintaining g-factor.
64.
by halyconWays 1y ago
They drove Aaron Swartz to suicide using the power of the State because he downloaded public domain documents. Meanwhile these AI training companies scrape the entirety of copyrighted human content and then have the audacity to claim people
65.
by halyconWays 1y ago
Oh no, not the sanctity of the license for weights trained by literally scraping all material and the works of countless artists and hobbyists.
66.
by halyconWays 2y ago
I'm not dismissing shit. I'm offering you an observation about your rude behavior.
67.
by halyconWays 2y ago
You should put more effort into addressing the very detailed and thoughtful reply you got (at your request) and which you're currently ignoring with just another challenge (with a grammatical mistake). You're currently a troll in
68.
by halyconWays 2y ago
Normal reddit died when Carmen Ortiz drove Aaron Swartz to suicide over persecuting him like a terrorist for downloading a few thousand public domain scholarly articles. (Which Meta and others now do with proprietary, copyrighted material a
69.
by halyconWays 2y ago
The major subreddits are almost entirely bots as well. Try posting anything authentic and you'll be banned, regardless of how old your account is or how human you appear to be. It's entirely controlled, fake discussion pretending
70.
by halyconWays 2y ago
Sounds like plenty of time for me to get my bag, pull out my phone, and start texting
71.
by halyconWays 2y ago
>passive safety is the only safety Cybertruck has Not even remotely true
72.
by halyconWays 2y ago
While we're sharing anecdotes, I have FSD (13 now) on my model Y and love it. I was anxious at first and remain very guarded while using it (which you're obligated to do anyway) but it's taken away a lot of the tedium and fat
73.
by halyconWays 2y ago
More impressive than the "digital thermometer runs doom!" where really they just connected the OLED display, but IMO if it's not being processed on the device then it's not running Doom.
74.
by halyconWays 2y ago
The author's example of regime-beneficial censorship is poorly chosen. He says that Deepseek refuses to answer questions about the Tienanmen Square massacre but will list US-backed human rights' violations, and that's true. B
75.
by halyconWays 2y ago
She actually stole that code from a user named slaren and was personally banned by Gerg from the llama.cpp repo for about a year because of it. Also it was just lazy loading the weights, it wasn't actually a 50% reduction. https:&#x2
76.
by halyconWays 2y ago
Maybe I'm an outlier but I don't see much value in running tiny local models vs. using a more powerful desktop in my house to host a larger and far more usable model. I run Open WebUI and connect it to my own llama.cpp/kobold
77.
by halyconWays 2y ago
Oh no, so sad. The Open non-profit that steals 100% of all copyrighted content and makes multiple billion-dollar for-profit deals while releasing no weights is crying. This is going to ruin my sleep. :(
78.
by halyconWays 2y ago
Not sure if you'll see this, but I ended up making them and they're awesome! I had to buy a pickkit 3 which wasn't ideal, but it worked out in the end. I still have mine in random pattern mode, as I should've realized th
79.
by halyconWays 2y ago
Why do you consider it "putting down others" to say that they're wrappers of other people's work, which they are?
80.
by halyconWays 2y ago
It became that, due to the people that Reddit banned first. For quite a while (I forgot it was called Whoaverse) it was far more how Reddit used to be, with open discussions.
81.
by halyconWays 2y ago
Please make one for Jetbrains' IDEs!
82.
by halyconWays 2y ago
I think Bluesky's content is largely defined by the crowd of people that left Twitter. Like Voat when Reddit had its embrace of censorship, only the other way around: opposite politics, and it's more based around being upset that
83.
by halyconWays 2y ago
I had better luck going to jlcpcb.com and uploading the files. I did modify the BOM for U5 but they didn't have any in stock, so I ended up getting them from mouser and will hand-solder those. I'm super excited for the project, I&
84.
by halyconWays 2y ago
I'd love to see a design that uses a dirt-cheap uC, because the PIC16F1782 chips he had laying around are $2.50 each on Mouser! Edit: Tried to buy these (either PCBs, the BOM, or ideally fully assembled boards) and holy shit I've
85.
by halyconWays 2y ago
PSA: it's "fib," not "phib"
86.
by halyconWays 2y ago
I was channeling /g/ more than Reddit, but sometimes something is so stupid it has to be mocked with matching stupidity. My point is similar to other high-ranked points on this thread, I just didn't feel like gussying it up.
87.
by halyconWays 2y ago
Imagine the dread when he realizes he should've taken short videos of each object from different angles, so that an AI could create 3D models to go along with a photogrammetry/gaussian splat/NeRF recreation of his home.
88.
by halyconWays 2y ago
That's fair. I did run into that issue when trying to speed up Hunyuan
89.
by halyconWays 2y ago
Super basic intro but perhaps useful. Doesn't mention quant sizes, which is important when you're GPU poor. Lots of other client-side things you can do too, like KoboldAI, TavernAI, Jan, LangFuse for observability, CogVLM2 for a v
90.
by halyconWays 2y ago
Why not 3090s? Same VRAM and cheaper. With both setups you'd be limited to 1B. By contrast, you can run 4-bit quants of Llama 70B on two {3,4}090s, and it's still pretty lobotomized by modern standards. You can also train your own

More ›