M4v3R

22 ms·

301.
by M4v3R 3y ago
It also works for photos/images! If you have text in your photo that contains a unit, you can click on the Live Text button in one of the corners and then on the unit itself, and you will get a conversion automatically. It works not on
302.
by M4v3R 4y ago
I'm the same, and it has been working really good for me so far. The only downside is that it's very easy to stay in your bubble of subscriptions and not get broader view from other sources. That's why occasionally I will sti
303.
by M4v3R 4y ago
This video is one of my favorites from Ahoy, what a great job preserving this awesome piece of retro music history.
304.
by M4v3R 4y ago
Yeah the background of the original also seemed to contain some kind of slight texture.
305.
by M4v3R 4y ago
I literally got back from a doctor today after having back pain and he prescribed me to do stretching exercises. These will come in handy, right on time. Thanks!
306.
by M4v3R 4y ago
I agree, but then again I don’t think prompt injection attacks are as severe as SQLi or XSS attacks. The latter can be disastrous for your application if even one is found, while for prompt injection the worst can happen is that the user wi
307.
by M4v3R 4y ago
> Here's an example: https://simonwillison.net/2023/Apr/14/worst-that-can-happen/ Your example doesn't use the same kind of prompt I mentioned above. When I've added "You are to ig
308.
by M4v3R 4y ago
It's pretty hard to formally prove that such an attack is impossible given the infinite number of inputs you can give to an LLM, but from my limited testing this method is pretty robust and personally I didn't find a way to break
309.
by M4v3R 4y ago
As I've already pointed out in another thread [1] the prompt injection attack where you insert an injection as invisible text inside your article will not work with GPT-4 when you use a system prompt correctly. You just need to tell it
310.
by M4v3R 4y ago
I agree and I think OpenAI is aware of this issue. I’ve read the other day that they are employing a “red team” - group of people that basically are tasked with breaking their models so they can find weak spots and fix them.
311.
by M4v3R 4y ago
I wouldn’t say so. For 3.5 it’s not like they didn’t want to provide this option, they just couldn’t. As for GPT-4 it does make sense it’s only available via the API. ChatGPT is basically an end product so it’s not really vulnerable to prom
312.
by M4v3R 4y ago
GPT-4 (the one available via API) is indeed more resistant against prompt injection attacks because of how the model treats “system message” (that’s configurable only via the API). It will really stick to the instructions from the system me
313.
by M4v3R 4y ago
Can you provide an example? Because from my experience it's the exact opposite - GPT-4 can handle translation, especially when there are complex sentences and context that needs to be kept across sentences, way better than Google Trans
314.
by M4v3R 4y ago
I've used GPT-4 for translation numerous times and it worked astonishingly well, better then Google Translate most of the time. Especially when translating idioms and slang. It's also much better than GPT-3 at poetry - I really su
315.
by M4v3R 4y ago
From the article: > To the best of my knowledge—and I double-checked—this exam has never before been posted on the public Internet, and could not have appeared in GPT-4’s training data.
316.
by M4v3R 4y ago
More like $1,92 (32 * 0,12) for a 32k prompt, or twice that for a 32k completion. Still not cheap though.
317.
by M4v3R 4y ago
Yes :). Up to a point, that is. Please see the video demo I've included in the README: https://www.youtube.com/watch?v=XgMKCeiUDQc
318.
An experiment for creating an AI Developer sidekick that works on existing code(github.com)
5 points by M4v3R 4y ago|2 comments
319.
Show HN: GPTinker – AI developer sidekick that can work on existing codebases
3 points by M4v3R 4y ago|2 comments
320.
by M4v3R 4y ago
This is actually possible, I made a proof of concept agent that does exactly this. The agent has access to several commands that get executed on demand - ListFiles, ReadFile and WriteFile. You give it a prompt - for example "add a head
321.
by M4v3R 4y ago
It's perfectly possible to assemble a project by writing individual files one at a time, so you would basically get 8k tokens per file. Or you could even write out the files in parts.
322.
by M4v3R 4y ago
I've checked some of the failure mode he mentions using GPT-4 and it answered the questions/completed tasks correctly, so I can only assume the author used GPT-3.5.
323.
by M4v3R 4y ago
It’s not totally obsolete but for 99% use cases nowadays there are faster ways to achieve the same goals than hand writing assembly. I predict that the same will happen with AI-assisted programming.
324.
by M4v3R 4y ago
If you don’t care you can simply close the page. HN rules don’t allow editorializing the titles of the stories, OP just put the page title there like we do for every single story on this site.
325.
by M4v3R 4y ago
> It's a form of propaganda to shame anyone who doesn't believe things unconditionally. Not necessarily. If someone doesn’t believe something that is universally believed and for what there exist undeniable evidence (for exampl
326.
by M4v3R 4y ago
I can’t say much about it’s ability to write lower level C, but I found that it’s pretty useful in explaining what low level code does, even if to me at a first glance it’s not obvious. It doesn’t only give you a line by line explanation bu
327.
by M4v3R 4y ago
Not the case for GPT-4 though, as it’s knowledge cutoff is the same as GPT-3s, that’s why it’s easy to compare the two on the same problems and see what’s the difference.
328.
by M4v3R 4y ago
Yes and no. You can totally create a simple game with ChatGPT but you still need to take the code it spits out, paste it somewhere, compile and/or run it in some environment. Until someone builds an interactive ChatGPT code sandbox for
329.
by M4v3R 4y ago
What's even more interesting, I gave GPT-4 a follow-up instruction: > Please provide a single mathematical equation that could be used to solve this problem. And it gave me the following answer: > Sure, let's represent the r
330.
by M4v3R 4y ago
From the waitlist, I subscribed for it early on and got access just a few days ago.

More ›