dmos62

18 ms·

421.
by dmos62 1y ago
That's a pretty cool approach!
422.
by dmos62 1y ago
I would really like a benchmark showing that AIs can use this. Just the possibility that AI can understand the compressed format similarly well as the original excites me. How did you come up with the format?
423.
by dmos62 1y ago
There's a sample for svelte in the repo.
424.
by dmos62 1y ago
It's a win, because people will not fear Google shuttering their experiments, and thus will be more likely to use them. It's also a win, in that it furthers a common good: if Google abandons a venture, why would they be upset if s
425.
by dmos62 1y ago
I want it. And, I'm hardpressed to imagine a multimedia consumer that doesn't care about color and brightness ranges, unless they don't understand what that is. Every movie is recorded and many phones now capture in a range t
426.
by dmos62 1y ago
Thanks for clarifying this!
427.
by dmos62 1y ago
Yes, but I don't think that their code is accessible. And the whitepaper is light on details.
428.
by dmos62 1y ago
It occurred to me that the cache could be indexed not only by environment state but also by intent. A second agent could subdivide trajectories into steps, upgrading trajectories into ordered lists of sub-trajectories. Each trajectory and l
429.
by dmos62 1y ago
I love the minimal approach and general-use focus. If I understand correctly, the engine caches trajectories in the simplest way possible, so if you have a cached trajectory a-b-c, and you encounter c-b-d, there's no way to get a &quot
430.
by dmos62 1y ago
No, the program and prompt databases use a genetic algorithm.
431.
by dmos62 1y ago
That's not inherent to HDR though. BFV (unless I'm confusing it with something else) has a HDR adjustment routine where you push a slider until the HDR white and the SDR white are identical. Same could be done for desktop environm
432.
by dmos62 1y ago
You're right. I guess this illustrates a downside of closed-source and walled-gardens.
433.
by dmos62 1y ago
I wish they'd open-source what they're shuttering. Would be a win-win as far as I can tell.
434.
by dmos62 1y ago
Thanks for the insightful counter-argument.
435.
by dmos62 1y ago
Bad software is not cheaper to make (or maintain) in the long-term.
436.
by dmos62 1y ago
You're right. I commented too hastily.
437.
by dmos62 1y ago
I don't think you read the article. > "They handcuffed a young boy and shot him," recalled one veteran who served with the SAS in Afghanistan. "He was clearly a child, not even close to fighting age." No Western
438.
by dmos62 1y ago
Really cool. "Other Key Findings" were worth the read too.
439.
by dmos62 1y ago
What could tweets be?
440.
by dmos62 1y ago
You need time/repetition for new information to sync in. The time needed is inversely proportional to how familiar you are with the subject. For example, I just said what I wanted to say, but now I'll repeat it in another way, bec
441.
by dmos62 1y ago
It irks me that I still don't know what an open model means. Or rather, I don't like calling a model that's trained on a closed dataset, with secret techniques as open, even if the weights are publicly available.
442.
by dmos62 1y ago
True, though it becomes more nuanced when it's your neighbour-N-times-removed that's getting invaded.
443.
by dmos62 1y ago
Impressive. I like ZeroTier, would you consider Mycoria a dropin replacement?
444.
by dmos62 1y ago
You seem insightful on this subject. What other parametric design tools, techniques or technologies do you consider noteworthy?
445.
by dmos62 1y ago
Presuming a higher bandwidth interface than what we currently have (voice/text chat).
446.
by dmos62 1y ago
In other news, bad software is bad. Heh, excuse the sarcasm. Tangential: I've come to think of the major software problems (like bloat, or closed-source, or lack of interop) as an effect of our generally chaotic and self-contradictory
447.
by dmos62 1y ago
> The decisions aren’t as hard as leaders are making them. Is there a statement that is more "armchair-policy-maker"? In all seriousness, I have a similar sentiment about support for Ukraine, but I'm saying that it's
448.
by dmos62 1y ago
There's as many theories about why Putin and Kremlin did what they did as there are war analyst books on it. Some of them are more right than others, but at the end of the day, if I'm honest, I'm as confused as I was in 2014.
449.
by dmos62 1y ago
We're misunderstanding each other. I take what you say to mean that Netflix could offer me unlimited bandwidth for 5 euros a month or whatever they're charging.
450.
by dmos62 1y ago
Is it self-evident to you that that's not cost-prohibitive with what people pay today?

More ›