mungoman2

8 ms·

61.
by mungoman2 2y ago
This is getting ridiculous! Curious, since this is a strong loop old frame + input -> new frame, What happens if a non-CS image is used to start it off? Or a map the model has never seen. Will the model play ball, or will it drift back
62.
by mungoman2 2y ago
Not knowing whether a tap shows more text or moves to the next slide makes me super anxious. Would love an indicator whether there's more text for the slide in the next tap.
63.
by mungoman2 2y ago
Not buying it. Japanese has also gone through the same tech tree.
64.
by mungoman2 2y ago
I'm looking forward to this expanding to other languages. There is so much great information out there, but unfortunately mostly in English. LLMs are really great at translating, and together with recent text-to-speech, seems straight-
65.
by mungoman2 2y ago
Surely that's only their interpretation of privacy laws, and not something tested in courts. It seems unlikely to actually break regulations if it's clear that the data has been fished out of the entropy well.
66.
by mungoman2 2y ago
This is very cool and a very impressive effort. I want to discuss though if "real" 3D is the right solution for such constrained platforms. The aliasing is quite extreme due to large pixels and lack of filtering. A more constraine
67.
by mungoman2 2y ago
Nice, but kerning seems off. See for example word "Tails" in the article. a and i should be closer.
68.
by mungoman2 2y ago
Image showing the star on the statue https://commons.m.wikimedia.org/wiki/File:Braunschweig_Gauss...
69.
by mungoman2 2y ago
To avoid changing the overall brightness like in the examples it is important to work on linearized values or to adjust the running error accordingly. It's not correct to work on the encoded sRGB values of the image directly. This is a
70.
by mungoman2 2y ago
Yeah this is awesome. Keep reducing that latency, that's the path to the killer assistant.
71.
by mungoman2 2y ago
Hm I don't follow. Could you please define "expected value in money terms" vs "expected value in utility of money terms"
72.
by mungoman2 2y ago
What is the source for the logos? They are sometimes jpg with bad compression artifacts.
73.
by mungoman2 2y ago
No blunder here. They're saying that with a 64-bit hash you will likely see hash collisions with 2^32 items. It's easy to generate 2^32 items, so it follows that it is easy to generate a collision. The high likelihood of collision
74.
by mungoman2 2y ago
Super curious what the distortion was! Unfortunately no GitHub link
75.
by mungoman2 2y ago
How many cycles could this actually save? I would assume the latency to actually get data from DDR is only a small part of the whole round-trip in a L1 miss. Actual savings much smaller than 30%.
76.
by mungoman2 3y ago
11.4 g per 100 g, yes. OPs example is also 11%.
77.
by mungoman2 3y ago
> The fix was easy: upgrade our hardware. This made me smile. My brain autocompleted the fix to something like "help the neighbors trim their tree", but of course the fix is new hardware.
78.
by mungoman2 3y ago
Very interesting. So if you want an intelligent model, you should train it on data that gives a high IQ score. You can then test a text to see if it increases or decreases the IQ of a model, if included in the training set. This makes it po
79.
by mungoman2 3y ago
Really love this idea and execution. Truly fantastic!
80.
by mungoman2 3y ago
Check out the game Return of the Obra Dinn for exactly this.
81.
by mungoman2 3y ago
This is awesome. I dream of a mechanical computer that can be 3D-printed in one go, fully assembled. A subset of all linkages and joints can be printed this way. Is it possible to build a computer out of them?
82.
by mungoman2 3y ago
Cool! What laser cutter can be used for this kind of project?
83.
by mungoman2 3y ago
Well tbh, failing to deliver after over committing is also a good experience.
84.
by mungoman2 3y ago
Happy to see this! Spent too much time in Ultima Online in my early teens. This lead to hosting free shards, working on C# in a large code base (RunUO), and dealing with real players/users. Very good experience and set me on the path t
85.
by mungoman2 3y ago
Fun! But how to read the Y axis? Why does it start from 1 twice?
86.
by mungoman2 3y ago
Come on, this is the wrong attitude. What they asked for could be prototyped in a day to see if it makes sense. Have the left image in the demo be fed from a rendering from a simple 3d scene of a camera and simple colored primitives.
87.
by mungoman2 3y ago
Artificial Intelligence is mentioned in the intro, but was it actually applied?
88.
by mungoman2 3y ago
Everything? I don't understand from this article or it's links what this is actually about.
89.
by mungoman2 3y ago
Yes, completely agree about the concept. I've had success discussing floats with compute engineers by framing it like a really large fixed point number with all zeroes, except a small bit range. The location and value of the bit range
90.
by mungoman2 3y ago
Yeah this doesn't hold water. If you need the precision, float isn't suitable since he precision drops of. If you don't need the precision, no need to use float. Just use int64.

More ›