Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
bArray
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
17 ms
·
391.
▲
by
bArray
2y ago
> It is very strange to see a grayed out comment on top of a thread that has about 160 non-grayed out comments. I thought that highest rated comments go to the top and grayed out comments are rated below zero. I don't mind, I have i
392.
▲
by
bArray
2y ago
I think that combustion engines will make a come back, and we're at a false peak. I believe that due to a poor economic outlook, we see more people purchasing from the used-car market than new vehicles. I personally have seen a recent
393.
▲
by
bArray
2y ago
Writing docs for software is something that in theory LLMs should be great at. It's clear they are useful, and that developers generally do not like writing them, and that LLMs are good at condensing complex information into easier to
394.
▲
by
bArray
2y ago
To mark the parsing of the day and to note that any incomplete tasks that I intend to keep have been re-planned to elsewhere. Any marked days are then there for historic purposes only.
395.
▲
by
bArray
2y ago
I think the format I suggested would parse in markdown, but I created it pre-markdown. For me the best parts are: 1. You could write a parser very quickly. 2. It's very easy to edit and add to as a human. 3. It can easily be tracked in
396.
▲
by
bArray
2y ago
I think it's somewhat passable with line-based tools, i.e. you could easily use sed to match a range [1]. I think that it's far more important that it is human readable, if it becomes too tricky to edit manually, then just go with
397.
▲
by
bArray
2y ago
My format used to be as follows: * [ ] YYYY-MM-DD * [ ] 1800: Dinner with parents * [x] Walk the dog * [ ] YYYY-MM-DD * [ ] 0900: Interview new recruit * [ ] Prepare interview questions * [ ] Uplo
398.
▲
by
bArray
2y ago
It's times like these I look forward to the likes of the Ladybird browser [1] making progress. [1] https://ladybird.org/
399.
▲
by
bArray
2y ago
> ... I mean, this is what LLMs are. I suspect it's not an LLM for dictation? It's probably a network that converts sound to some phonetic tokens, then some kind of probabilistic matching like a Markov chain (which would be com
400.
▲
by
bArray
2y ago
Firstly, it's pretty crazy they allow words to be associated like this. You would think that such a large company would check for this. You could end up with some extremely awkward suggestions. I would probably have some process for do
401.
▲
by
bArray
2y ago
Exactly this. In the pages and pages of terms and conditions, this will likely be covered, plus a lot of these services are opt-out. By not unticking "train on my workspace data" hidden several matrix layers deep, you are probably
402.
▲
by
bArray
2y ago
Google openly train stuff based on your email, they used is specifically to train Smart Compose, but maybe other stuff too. He likely uses multiple Google products. Draft papers in Google Drive perhaps? These LLM models are essentially tryi
403.
▲
by
bArray
2y ago
Exactly my thought, probably the least likely part of the whole thing. He emailed Google and they replied. Not only that, he asked a question they would really rather not answer.
404.
▲
by
bArray
2y ago
In theory you have the likes of the PinePhone where you can run a full Linux kernel [1]. You could then use something like Waydroid to run Android apps [2]. I think the biggest concern is that many of the important apps are anti-emulation,
405.
▲
by
bArray
2y ago
> No, it's much broader than that. The UK is asking for a backdoor to your data and backups in the cloud, not on your device. Why bother with searching physical devices when they can just issue a secret subpoena to any account they
406.
▲
by
bArray
2y ago
If you go too far right or left, both types of authoritarianism are difficult to distinguish. I think this just makes the case that every election you need to be a swing voter, make sure your politicians still overlap with your ideals. Appl
407.
▲
by
bArray
2y ago
Too right, it was far more problematic than they ever made out. > The UK government's demand came through a "technical capability notice" under the Investigatory Powers Act (IPA), requiring Apple to create a backdoor that
408.
▲
by
bArray
2y ago
I agree it's the better proposal, but it does slightly break the "drop-in"-ness of Podman.
409.
▲
by
bArray
2y ago
> I am not sure how correct this assumption is. S3 is supposed to cut power to everything but RAM, but for example Gigabyte Aorus motherboards are notorious for an NVMe SSD sleep bug that randomly prevents the system from properly sleepi
410.
▲
by
bArray
2y ago
> It makes more sense when you realize that all the money being thrown into Ukraine is likely being funneled out as tax-exempt gains by one billionaire or another who gives kickbacks to the politicians who enable this grift. The Western
411.
▲
by
bArray
2y ago
> The UK is also sending eye-watering amounts of money to Ukraine without a problem, while everyone else is pulling out of the project. The UK government has never been more obviously out of step with the rest of the West. Like it or not
412.
▲
by
bArray
2y ago
The point being made is not about removing AI investments. The OP's comment: > I am urprised how much money suddenly is available where other future-critical projects are starved of funding while needing just a fraction of that amou
413.
▲
by
bArray
2y ago
It's the same in the UK. Pensioners had their heating allowance removed to plug a £22 bn 'black hole', whilst tonnes of money is pledged to AI. I can imagine the EU is in a similar state.
414.
▲
by
bArray
2y ago
Personally, I started building tooling around Data URIs to provide permanent archives to external resources. You can only really store text or single images like this, and you have to get the size below 16KB [1] (64kB at a push) to be relia
415.
▲
by
bArray
2y ago
> - Why is it better than internet archive? With the internet archive, the purpose seems to be for public archiving. One could imagine a use-case where you want non-public archives, and are therefore not subject to any take-down requests
416.
▲
by
bArray
2y ago
Maybe I can do one better... For the 'create_chat_completion' on this line: https://github.com/simonw/llm-smollm2/blob/afed0963b3a8bcca1... I changed it to: completion = model.create_chat_comp
417.
▲
by
bArray
2y ago
Not that convinced by this model: import llm model = llm.get_model("SmolLM2") model.prompt("Write me a poem about computers").text() I ran the above to test it, and got this: In the realm of comput
418.
▲
by
bArray
2y ago
Not just most of the judges, but most of the MPs who voted on this. Let them eat their own cake. I think they could do something like what Tik Tok did, by letting users know why they can no longer provide the service. I would personally giv
419.
▲
by
bArray
2y ago
I think grid storage does care: * Size is important, because they need to be stored and maintained somewhere. The larger the energy storage, the more infrastructure required. * Weight typically scales cost (the heavier it is the more it cos
420.
▲
by
bArray
2y ago
Personally I learned with 8086/8088 via MikeOS [1] (written in NASM). Highly recommended. [1] https://mikeos.sourceforge.net/
More ›