xmprt

17 ms·

301.
by xmprt 2y ago
In my experience using LLMs, the 90% is less about buggy code and more about just ignoring 10% of the features that you require. So it will write code that's mostly correct in 100-1000 lines of code (not buggy) but then no matter how h
302.
by xmprt 2y ago
I agree with you but as I've matured as a programmer, I feel like it's very hard to get abstractions for boilerplate right. Every library I've seen attempt to do it has struggled.
303.
by xmprt 2y ago
Is it just me or does the author here sound like they're hating just to hate? The writing doesn't sound that terrible. Maybe it's a bit amateur but isn't that what you're expect from an engineering blog post written
304.
by xmprt 2y ago
I personally haven't worked with many of the github apps that you seem to refer to but the few that I've used are only limited to access the specific repositories that I give and within those repositories their access control is s
305.
by xmprt 2y ago
On the flipside, you might end up getting scammed even worse because of incorrect analysis. For example if ChatGPT hallucinates some data/features through faulty research then you might be surprised when you actually make the decision.
306.
by xmprt 2y ago
If people are using them outside of speculation then I'd consider that legitimate. I'll concede that most of the market cap of cryptocurrencies is from speculation but at least Bitcoin, Ethereum, and a few other coins have use cas
307.
by xmprt 2y ago
Bitcoin, ETH, and Monero all have utility in one way or another. Bitcoin is accepted by most black markets (and Monero is even better for privacy). And software is built on top of the ETH chain. No one is buying stuff using DOGE or Trump co
308.
by xmprt 2y ago
I find your videos fascinating and I'm surprised that more people haven't checked them out or read your articles. I personally use Android so the videos are technically not applicable for me but I still get a lot of value out of l
309.
by xmprt 2y ago
If anyone knows someone similar for Android, I'd be interested in learning more. I've seen some of the impressive things that can be done using Revanced but there don't seem to be good guides for how to get started with somet
310.
Debugging an Undebuggable App(bryce.co)
351 points by xmprt 2y ago|58 comments
311.
by xmprt 2y ago
Attention works better on smaller contexts since there's less confounding tokens so even if the LLM can see the entire context, it's better to keep the amount of confounding context lower. And at some point the source code will ex
312.
by xmprt 2y ago
If your abstractions match common design patterns then you've solved your problem. It's ridiculous to assume that an LLM will understand 1k LOC of standard library code better than 10 lines of a custom abstraction which uses a com
313.
by xmprt 2y ago
I'd love to have those answers too but it seems like DOGE doesn't care about transparency as much as they claim to want that. Elon keeps touting open source and transparency but the transparency is only in the form of poorly resea
314.
by xmprt 2y ago
Zig doesn't have a 1.0 release yet so I'd be hesitant to use it on any production projects. I've dabbled in it for some toy projects just to get a feel for the language and so far I think I still prefer Rust although the simp
315.
by xmprt 2y ago
I think you feel that way because you're comparing Node to Python. Neither of them are particularly good languages for servers. Languages like Go and Java/Kotlin are much better when you're building complicated systems that n
316.
by xmprt 2y ago
I think a lot of frontend engineers can do backend work and vice versa, but it takes a lot to be very good at both. It's not impossible but also I think it's rare that a backend engineer would voluntarily decide to also become a f
317.
I Automated My Taxes Using AutoHotKey(preethamrn.com)
2 points by xmprt 2y ago|1 comments
318.
by xmprt 2y ago
To be fair, I can't think of a single context where I would want to truncate a password before hashing so interoperability with other systems isn't worth letting by a dangerous edge case in my opinion. I'd rather have the sys
319.
by xmprt 2y ago
I wouldn't say 2019 is particularly old. I didn't even know IE still existed at the time - I thought it had been replaced by Edge.
320.
by xmprt 2y ago
Perhaps I suck at prompting but what I've noticed is that if an LLM has hallucinated something or learned a fake fact, it will use that fact no matter what you say to try to steer it away. The only way to get out of the loop is to know
321.
by xmprt 2y ago
I agree. I just think we need to focus more on congress' role in this. The checks and balances only work if we force congress to act as a check and as long as congress keeps voting along party lines and doing the president's biddi
322.
by xmprt 2y ago
I get how the system works but I also don't understand how congress can force the executive branch to take out a loan but also sets a debt ceiling which could shut down the government unless it's raised by congress. So congress bl
323.
by xmprt 2y ago
I think the solution depends on the type of problem you have. If your problem is reliability then increased testing and handling edge cases (or avoiding code that even hits those edge cases) will help. But if the problem is uncertain requir
324.
by xmprt 2y ago
Democrats have bigger fish to fry and DOGE isn't a real department so it doesn't have a whole lot of authority to do things on its own. It can only advise the government so in the end, until an executive order is signed or some ot
325.
by xmprt 2y ago
I, for one, think this is a valuable piece of information and somewhat interesting analysis. You can take the cynical point of view that this was released just to tarnish their reputation or you can assume that it's security researcher
326.
by xmprt 2y ago
Nah. I'd almost argue that discovery coding is what helps define the architecture. I've seen way too many cases of over complexity designing a system for scale that will never be hit which results in an architecture that is too ri
327.
by xmprt 2y ago
> human slop is typically easy to detect by grammar and other clues I'm not sure this is true. There have been a lot of times where I see a very well made video or article about some interesting topic but when I go to the comments I
328.
by xmprt 2y ago
Also, it's not counting many of the startups from the last few years which haven't had enough time to fail. If the denominator is the total number of exits then it's probably much lower than 4k.
329.
by xmprt 2y ago
ChatGPT helps me get an initial grasp of the topic from which I can come up with more questions and research to do on my own. Usually I don't use ChatGPT as a source but rather a springboard for ideas.
330.
by xmprt 2y ago
And businesses that knowingly sell alcohol to minors are charged with a crime.

More ›