Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
ec109685
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
30 ms
·
451.
▲
by
ec109685
2y ago
There are similarities to the approach in this paper (though they trained a model from scratch): https://arxiv.org/pdf/2309.07062 How well would an LLM trained with a huge number of examples do on this test? Essentiall
452.
▲
by
ec109685
2y ago
I know the published stats show it to be relatively safe, but it’s poorly regulated and not clear if all accidents are reported: “28 people died from skydiving near Lodi. How does the Parachute Center stay in business?” https://w
453.
▲
by
ec109685
2y ago
“Excellent query good sir! <said slowly enough to let the LLM catch up>…” And more seriously, it seems like the LLM could be used to precreate lots of filler prefixes that correspond to the rag’d document that are being sent to the mo
454.
▲
by
ec109685
2y ago
If only Salesforce was better software, billions of capital funding “a better Salesforce” could be deployed in more productive ways.
455.
▲
by
ec109685
2y ago
Companies really don’t like being sued for hundreds of millions in punitive damages just for the benefit of training on the small percentage of people that opt out.
456.
▲
by
ec109685
2y ago
They’re letting OpenAI upsell to a professional version, so there is a lot in it for OpenAI to offer this for free, even without the data.
457.
▲
by
ec109685
2y ago
They have tech to obscure IPs in a way that prevents any one entity from being able to de-obfuscate: https://support.apple.com/en-us/102602
458.
▲
by
ec109685
2y ago
You can permanently disable OpenAI from training with your chat data for your account: “To disable model training, navigate to your profile icon on the bottom-left of the page and select Settings > Data Controls, and disable “Improve the
459.
▲
by
ec109685
2y ago
Then you’re checking in artifacts generated from source code (for this method that relies on git push to work).
460.
▲
by
ec109685
2y ago
It’s a materialized view (so it’s the data structure providing key lookup).
461.
▲
by
ec109685
2y ago
At least ChatGPT 4o still looks like it is using dalle. https://x.com/krishnanrohit/status/1755123169353236848?s=46
462.
▲
by
ec109685
2y ago
9 months ago! https://tabular.io/blog/the-case-for-independent-storage/
463.
▲
by
ec109685
2y ago
This seems like a temporary problem as AI models improve. If we can figure out how to extract meaning from distorted images, AI will be able to as well. OCR for distorted text is a solved problem. Relatively simple math based on normalizing
464.
▲
by
ec109685
2y ago
Fools who believe there is risk free 2x yields and their money are soon parted: “Users save money, get over 2x the national average in interest and weekly chances to win additional prizes up to $10 million through weekly random number drawi
465.
▲
by
ec109685
2y ago
Imagine a world where employees can’t take a better offer, but a business could have their employees invent the next Google search, and then immediately fire all the workers that made that happen and hire cheaper ones to operate it. That wo
466.
▲
by
ec109685
2y ago
Given you’ll eventually get that full software deduction tax break back, just spread across several years, I am surprised there aren’t companies that will finance that to have the money upfront (with a fee of course :( )
467.
▲
by
ec109685
2y ago
Obesity is such a huge factor in many diseases it is logical that that reducing it, has these other beneficial properties.
468.
▲
by
ec109685
2y ago
Blog post about the approach here: https://sillycross.github.io/2023/05/12/2023-05-12/
469.
▲
by
ec109685
2y ago
Impressive. Here is the blog post (also linked from their github) that talks about the bypass (the tldr; is that they rotate through public keys in a deterministic way, so Apple’s protections think it’s a new device each time and thus don’t
470.
▲
by
ec109685
2y ago
Yeah good point, I have seen this before with sites, but is suss.
471.
▲
by
ec109685
2y ago
It seems like those breaches all revolve around misconfiguration. Aren’t all clouds vulnerable to issues like this?
472.
▲
by
ec109685
2y ago
Accidentally left auto wipers on at the car wash. Never enabled them again.
473.
▲
by
ec109685
2y ago
Saudi Arabia should tax port fees a bit and use that to send people like this back to their home country. Inhumane.
474.
▲
by
ec109685
2y ago
Autonomous Trains. If you are not paying for the conductor, can’t you make trains much more appealing? They could run every five minutes, and last mile can be solved with autonomous car that is waiting for you when you arrive.
475.
▲
by
ec109685
2y ago
That graph those numbers are pulled from should incorporate speed. Otherwise, you could say sidewalks are more efficient than cars.
476.
▲
by
ec109685
2y ago
This type of behavior from stripe is super upsetting, most glaringly the lack of being able to talk to a human and discuss alternatives. That said, if the service is adding value, they shouldn’t lose half their revenue if they need users to
477.
▲
by
ec109685
2y ago
Without two phases: figure out what you need for the resource, execute query, how can you avoid N+1 problem, raw sql or not.
478.
▲
by
ec109685
2y ago
That wasn’t the case here. They had a horrible bug and no checks and balances when their automation ran on the account.
479.
▲
by
ec109685
2y ago
Pretty clickbait title then.
480.
▲
by
ec109685
2y ago
Oops, someone’s script was too greedy when uploading those elixir api documents.
More ›