sk11001

15 ms·

181.
by sk11001 3y ago
I don't see how that's relevant.
182.
by sk11001 3y ago
This isn't an argument about the capabilities of LLMs, it's an argument about whether we use the term AI - and that ship sailed a long long time ago. If your point is that LLMs won't lead to some kind of extreme AI capability
183.
by sk11001 3y ago
Recently the only people who have been saying that it’s easy to become a programmer in a few months are the ones selling you the way to get there.
184.
by sk11001 3y ago
> Originally published: 6 April 2021 The things the book talks about are not the same things that exist today.
185.
by sk11001 3y ago
You should be able to identify these things based on your interests and gaps in your knowledge e.g. “I want to learn more about backend development”, “network stuff comes up every once in a while but I’ve never studied it properly”, “this p
186.
by sk11001 3y ago
No, LLMs move fast and the people who write good technical books 1) are rare 2) take time to write them I also don't know if there is an incentive to spend a lot of time on an elaborate piece of work given how likely it is to be obsole
187.
by sk11001 3y ago
Not at all, the existence of alternative paths doesn't make the main path (degree -> first job) obsolete.
188.
by sk11001 3y ago
Is the complaint that v1 and v2 are different? Why are you learning v1 at all?
189.
by sk11001 3y ago
The difference from the EU referendum is that there's a presidential election every 4 years and 2-term limit.
190.
by sk11001 3y ago
It's not about the response after it happened, it's about whether you should set decades of foreign policy based on a single vote with a small margin.
191.
by sk11001 3y ago
I'm an ML engineer and I don't agree with the assumption that AI-related roles are somehow different. I don't think they're necessarily in more demand, or pay more, or in lower risk of being automated away. I do think th
192.
by sk11001 3y ago
Not getting responses from cold calls or applications isn't ghosting according to the author of the article (and I agree): > In order for it to be ghosting, the ghosted party has to expect the conversation will continue. This means
193.
by sk11001 3y ago
Wanted to put at least a couple of decent projects on my resume. People say that they don't matter but since I wasn't getting a lot of traction, I thought it would be a good idea. After a lot of procrastination and half-tried idea
194.
by sk11001 3y ago
I don't know if I would the generalize as much as the author about how different teams operate but there is a valid point in there - with some languages you can focus on getting things done while with Rust you spend a disproportionatel
195.
by sk11001 3y ago
The point of the article is buried under the bait that is the list of qualities to watch out for. The main problem was that these people overengineer things and work without any collaboration or business context. Out of the four listed qual
196.
by sk11001 3y ago
It's kind of fascinating to read but is there a significant takeaway from this? It's managers discussing metrics, plans and OKRs.
197.
by sk11001 3y ago
It's also fairly obvious and common sense stuff. I'm not convinced that thinking about these principles in the abstract is helpful at all.
198.
by sk11001 3y ago
TIOBE doesn't even measure engagement, it's just outright useless. For one example of its stupidity, Visual FoxPro, a proprietary language which has been dead for 8 years - https://en.wikipedia.org/wiki/Visual
199.
by sk11001 3y ago
Wrapper startups are a rounding error, what's important is that the big players in the field are releasing genuinely useful tools. Which makes it very different from crypto.
200.
by sk11001 3y ago
What do you mean concretely? What is the “what have we done” and unintended consequences of LLMs having some small % error under some circumstances?
201.
by sk11001 3y ago
It doesn’t need to be perfect to be useful.
202.
by sk11001 3y ago
A lot of these studies are observational including every single one of the cancer studies cited, and it's very difficult to separate high-calorie intake from high-protein intake. In the general population the group who eats a lot of pr
203.
by sk11001 3y ago
In my experience it works well when you discuss specifics and compare approaches - how do I do $x, compare $y and $z, I have this functionality so far, help me extend it in such and such way etc. I don't think that asking it for a big
204.
by sk11001 3y ago
I think the people you’re working with lack professional standards, no point in trying to generalize to a whole generation.
205.
by sk11001 3y ago
There's a big difference between being breakable and being representative of the web content used for training like you claimed earlier.
206.
by sk11001 3y ago
> They just produce something that looks like a human wrote it real good. That's what the base model does. To get an LLM assistant there are additional training phases to make it conversational, helpful and more correct - https:&#x
207.
by sk11001 3y ago
This isn't how LLMs work - they're trained to be helpful and give correct responses. You could argue to what extent that's achieved but the point is that it's not a weighted average of the internet, it's been fine-t
208.
by sk11001 3y ago
https://en.wikipedia.org/wiki/Intellectual_property I don't even know what my personal stance on piracy is but the subject is clearly more complex than saying that theft applies only to physical objects.
209.
by sk11001 3y ago
Type hints aren't useless - you can type check your code before you run it (and they make your code more readable). > What I would love is a mode in python where instead of duck typing everything, its statically typed, unless I do a
210.
by sk11001 3y ago
Do we really need a CLI to replace Cmd-S and a copy/paste command?

More ›