Are there general LLMs that can translate text 'on image', as opposed to spitting out the translation as text? I feel like it would be easier to reason about in images that are very text-heavy like some of these diagrams, but from what I remember, last time I tried, chatgpt and claude would only give me a text translation
It may be $11 here in the US, but it might be harder to get and more costly in a developing nation. In which case, if it is easier to get that $2 adapter and an e-waste keyboard, that likely makes sense for them
Just to clarify something from the article and the github repo code (sorry if this is readily explained in the repo, C code using AVX stuff is not something I have any experience with): does the SIMD version still handle things like escaped quotes inbetween quotes denoting a string in a field in the csv file? It isn't mentioned in the article
The funny thing for me is that I still have the phone number I had when I was growing up, which is for a state halfway across the US. Most of my spam calls are in the area code of my phone number, making them pretty easy to recognize since I dont really know anyone from that area code anymore
That is one part of microsoft, yeah, but then you also them making various parts of windows worse to try to squeeze every penny out of people, and where that fails, doing things like ads for apps in the start menu
I feel like I remember a similar attack related to Entra ID from a while ago, although I can't remember exactly what it was (maybe [0] or [1]?).. I understand that this is a complex system, but I would be concerned with the number of relatively high severity vulnerabilities being found in it.
Hack in this context is the software development one, i.e. 'hacking on a project' meaning working on a project. You are thinking of the cybersecurity one.
Forgive me if I'm misunderstanding, but wasnt that the original purpose of the blue checkmark - a verification that the twitter account was who they are saying they are, be it a person, company, etc?
I feel like bsky and mastadon only represent a subset of users, so I'm not sure how well you would be able to create a general rec system similar to twitter's from that that is useful outside of those places
The work being OSS and done free of charge doesn't excuse them from putting their users at unnecessary risk, especially when it is done so with only a one line mention in their github README and no mention on their website, which doesn't point towards the README at all
> coding assistant: cursor, claude code, codex, etc. write some code, have an intern-level chatbot write some more code, realize you hate doing code review, ship shitty unreviewed intern code that never improves, get mad, stop using it, get fired from coinbase
I feel like poster might have a bit of a history with Coinbase.. I'm assuming this was something along the lines of 'We want you to use AI to be more productive(tm)', the poster didn't, and then Coinbase got rid of them because they were 'not getting work done fast enough'
Edit: Just to clarify, I'm on the side of the poster on this, being forced to use AI tooling when it isn't helpful is very annoying
I would argue that this is less an example of why linking out may be bad for engagement and more an example of google abusing its intermediary/market position to keep users on their own pages longer
> Some progress is better than none, and it's still nice that my ISP can't snoop on me any more. Unfortunate, TLS also took away my ability to inspect my own traffic! This makes it more difficult for me to monitor what my OS and browser vendor are doing, and as I've said previously, i trust these parties comparatively less than my ISP.
It might be more correct to say that Certificate Pinning made it so you can't inspect your own traffic - for sites with TLS but without certificate pinning, you can just as easily create your own root certificate and force the browser and OS to trust the cert by installing it at the OS level. This is (part of, atleast) how tools like Fiddler and Charles Proxy allow you to inspect HTTPS traffic, the other part being a mitm proxy that replaces the server's actual cert with one the mitm proxy generates [0]
Does this also apply if someone were to do the following:
Receive encrypted transmission -> unencrypt it -> need to pass it on, so re-encrypt it and pass it on?
I would imagine that the paraphrasing wouldn't be necessary in this case because it isn't quite as useful to compare two encrypted versions of the text versus an encrypted version and an unencrypted version (also I feel like there is some risk of a game of 'telephone' in that the meaning would change bit by bit to the point of having a different meaning over time, even if not intentionally)
That is probably another reason that e-ink laptops haven't really become a thing then - practically any use besides maybe basic web browsing and emails would require a decent refresh rate to be reasonable, removing the benefit of using e-ink over the normal display
Are there any e-ink displays with a reasonable enough refresh rate that aren't exorbitantly expensive? I know atleast as of a few years ago that was one of the issues that would've impeded an e-ink only laptop
I'm not sure that a government that goes out of its way to try to silence political dissent and enforce social order through fear should be considered interested in the wellbeing of its people. [0][1][2]
> "At first, I was presented with an endless spinner, as I hadn’t configured things properly. The documentation on Github is enough, but frustratingly, the failure mode I ran into was the video just… not loading. However, I eventually got it nailed down, and now I have a new camera in Protect."
------------------------
It is unfortunate that they decided to omit how they fixed the last issue they mentioned. That could've been useful knowledge for others
If this did happen, you would think this would be better substantiated by now. All the articles I'm finding on it are very clear at the top that it is an 'unverified' report - you'd think they'd be able to find atleast one of the 1200 engineers willing to talk to them.
Perhaps I'm too cynical but I'm not sure I believe this until a credible news source names the exec or has it substantiated by an engineer.
My guess is that MarkMonitor is mainly used for their brand-relevant domains (microsoft, office 365, github (main site), etc), as opposed to one that a small subset of a small subset of their users of one service will use - I would imagine that microsoft likely owns hundreds of domain names and doesn't pay MarkMonitor to monitor every single one
I wonder if this is why I had a recruiter at a contractor contact me fairly suddenly about a role with Walmart... (them needing new people after effectively firing 1200 contractors)