grumbelbart

6 ms·

1.
by grumbelbart 17d ago
Get one with an inverter compressor, they are very quiet. They have a compressor that can be run at varying power levels, so instead of having one that goes of noisy at 100%, then sits off for a while, you have one that runs for example at
2.
by grumbelbart 3mo ago
It's bad. I believe them not to use it for training, but t means relevant data can and will be exfiltrated by US agencies or through court orders (see NY Times vs. OpenAI, where only traffic without any rentention was safe).
3.
by grumbelbart 4mo ago
Of course, should it be as cost efficient as claimed and if you don't use it but everybody else does, you might be pushed out of the market.
4.
by grumbelbart 5mo ago
Agreed. The solution will likely be some vision foundation model that directly sends controls to the robot ("move here, grab, move there"), trained by Amazon with RL to integrate collision avoidance, object detection, grasping poi
5.
by grumbelbart 5mo ago
> But electricity is under half (30%?40%?) and the rest of that energy isn't fossil-free. The trick of course is that if you electrify heating and transportation they'll need much less energy. Your average car with an ICE has a
6.
by grumbelbart 5mo ago
> If you were to design an entire ATC system from scratch (pilot interfaces, sensors everywhere in the airport and planes etc) it can be automated. Even then you'll probably run into the long-tail distribution issues, similar to sel
7.
by grumbelbart 6mo ago
Is this some kind of calibration then? I'd expect that the probabilities automatically adjust during training, such that in "lock" mode, for example, syntax-breaking tokens have a very low probability and would not be picked
8.
by grumbelbart 6mo ago
I'm reasonably sure Russia would take you.
9.
by grumbelbart 6mo ago
Would using an actors face and voice as training data be fair use? What it the model then creates a virtual actor that is very close to the real actor?
10.
by grumbelbart 6mo ago
The data behind the app is pretty solid, but lightningmaps.org has a much better visualization (based on the same data).
11.
by grumbelbart 7mo ago
I use them as cheap-man's VPN. A ssh server on a public IP but a non-obvious port brings you into the network, and port forwarding allows you to connect to relevant endpoints in your remote network via localhost:12345.
12.
by grumbelbart 7mo ago
> My Weird Hill is that we should be building things with GPT-4. Absolutely. I always advocate that our developers have to test on older / slower machines. That gives them direct (painful) feedback when things run slow. Optimizing w
13.
by grumbelbart 7mo ago
Putting sulfur into the right layers of the atmosphere seems to be the currently best viable options. It's not overly expensive, either. It acts fast and is reversible. https://en.wikipedia.org/wiki/Stratospheric_a
14.
by grumbelbart 7mo ago
The low prices of solar and batteries are a glimmer of hope. For many regions it's now the cheapest source of electricity.
15.
by grumbelbart 7mo ago
They would never be the same. It's just that everything still works the same if you switch out every i with -i (and thus every -i with i).
16.
by grumbelbart 7mo ago
> Is this the shadow of something natural that we just couldn't see, or just a convenience? They originally arose as tool, but complex numbers are fundamental to quantum physics. The wave function is complex, the Schrödinger equatio
17.
by grumbelbart 8mo ago
The thing is that your home's heatpump has an efficiency of 300%-500%. So even if your power plant and power delivery only has say 50% gas-to-electricity-at-home, you are still looking at 150%-250% gas-to-heat-your-house efficiency.
18.
by grumbelbart 8mo ago
Exactly. It is in general (much) more efficient to burn natural gas in a power plant and use the electricity for heatpumps compared to simply burning gas at home for heating.
19.
by grumbelbart 8mo ago
Would that help against a man in the middle that blocks the H3 traffic to snoop the URL when the client falls back to H2?
20.
by grumbelbart 9mo ago
> The "Elon process" relies specifically to the goal of getting rid of all dependencies. Musk has spoken extensively about building things from the ground up and not relying on other vendors (in this example complex software de
21.
by grumbelbart 9mo ago
Stupid question, but is 404 the real designator of that city, or a pun towards the HTTP error code? Edit: And what a great read, thank you!
22.
by grumbelbart 9mo ago
Can you say what hardware could do better? I.e. which kind of primitives do you miss, or would make it easier to develop safer software?
23.
by grumbelbart 9mo ago
Lightning detection. You have a couple of ground stations with known positions that wait for certain electromagnetic puses, and which record the timestamps of such pulses. With enough stations you can triangulate the location of the source
24.
by grumbelbart 9mo ago
Someone compiled a list of blocked domains (by probing different DNS servers): https://cuiiliste.de/ This is also how, for example, RT is blocked in Germany.
25.
by grumbelbart 10mo ago
Here: https://www.anthropic.com/engineering/claude-code-best-pract... claude.md seems to be important enough to be their very first point in that document.
26.
by grumbelbart 10mo ago
> In many European countries this can even reduce the usually robust protections you have as an employee. Huh, where?
27.
by grumbelbart 1y ago
> the federal workforce excluding the postal service (which has actually shrunk, as a semi-private employer) has grown by about 1% per year since 2000 That's less than it seems though, given that the US population has grown with ove
28.
by grumbelbart 1y ago
Mandatory: We should build it in space and beam the electricity back to earth using electromagnetic waves. We could collect those using solar cells. And then get rid of the plant and use the sun instead.
29.
by grumbelbart 1y ago
Long-term this would be done using LLMs. It would also solve LLMs' code quality issues - they could simply proof that the code works right.
30.
by grumbelbart 1y ago
That is correct. The permutations will likely break up into multiple cycles, and the cycle lengths follow a poisson distribution. https://dms.umontreal.ca/~andrew/PDF/CycleLengths1.pdf

More ›