Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
inductive_magic
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
12 ms
·
31.
▲
by
inductive_magic
3y ago
>Any tips on building robust LLM software? Honestly, it's difficult. One reason why we are very happy with our results is because the person in charge of experimenting with the LLM-flow, while highly intelligent, is not talented wit
32.
▲
by
inductive_magic
3y ago
>There’s nothing stopping Jian Yang from creating newpipeline.ai As someone leading a product team which "only calls apis" in this context: this is a very premature take. Hear me out :-) Robust LLM-powered software requires *
33.
▲
by
inductive_magic
3y ago
> I don’t trust this post The post appended the raw data provided by GPT, allowing you to verify the integrity of the data. This makes the post trustworthy from a methodological pov. > I don’t trust anything published on the Internet
34.
▲
by
inductive_magic
3y ago
Lovely how the article doesn't even attempt to clarify what kinds of "assets were stolen". They just go with "someone stole 200 Million" like someone walked out of the office with a duffle bag, when in reality there
35.
▲
by
inductive_magic
3y ago
> I would assume training an LLM to do the same would definitely be doable. I wouldn't be so sure. The reasoning process of Geoguessr pros is symbolic, not statistical inference. /edit: as other commenters pointed out, somethin
36.
▲
by
inductive_magic
3y ago
> "why not?" > Microsoft ...
37.
▲
by
inductive_magic
3y ago
> I never understood the point of hookup apps like Tinder. Access to hundreds+ of interesting people from the comfort of your couch? I've met many interesting women on tinder, that I'd never have met "in real life", i
38.
▲
by
inductive_magic
3y ago
The -highly entertaining- comments are rightfully bashing the de facto state of defi in the context of GPs comment. Remind me when a digital currency actually exists, because I haven't seen one yet. The problem isn't just that the
39.
▲
by
inductive_magic
3y ago
As other people here are recommending computer museums: I can recommend two in Germany: the Heinz-Nixdorf museum in Paderborn and the computer museum in Kiel. The one in Paderborn is quite big and has lots to see. The one in Kiel is located
40.
▲
by
inductive_magic
3y ago
I find it fascinating that we have no way of telling if a user is human just by looking at their signal. All the meaning we attribute to a signal depends entirely on something that is not detectable from the signal itself itself: wether an
41.
▲
by
inductive_magic
3y ago
I mean, from what I hear Berlin does take the cake on this one, but yes: its a country-wide phenomenon. For what its worth, I was pleasantly surprised by the things you can do digitally in Kiel. Schleswig-Holstein, Niedersachsen, Hamburg an
42.
▲
by
inductive_magic
3y ago
>What's wrong with DDG and local searches Maybe its different in the US or just not optimised in Germany, but the quality of search results isn't comparable. Query: "All you can eat Kiel" Google: 1. a map with the top
43.
▲
by
inductive_magic
3y ago
I don't trust them in an emotional way, but I think that our interests are aligned.
44.
▲
by
inductive_magic
3y ago
It depends on the usecase. If you're querying for something local like a restaurant or a store, google wins by orders of magnitude. If you're researching, duckduckgo is fine. Duckduckgo doesn't really compare to Kagi though.
45.
▲
by
inductive_magic
3y ago
It means something different than you're insinuating: https://www.collinsdictionary.com/dictionary/english/english...
46.
▲
by
inductive_magic
3y ago
Is "Englishism" slang? The term I know is "anglicism"
47.
▲
by
inductive_magic
3y ago
Apple solves this problem for me entirely. No need to tag documents etc, just iCloud everything -> everything is searchable.
48.
▲
by
inductive_magic
3y ago
I appreciate your perspective on the potential benefits of ads. However, I’m emphasizing a much deeper, existential concern here. Our lifes, essentially, are the sum of what we pay attention to. In this context, advertising doesn’t just sel
49.
▲
by
inductive_magic
3y ago
I wish more people understood how horrible advertising truly is. Your life is essentially a compilation of what captures your attention. Embracing this perspective, any entity auctioning off your attention to the highest bidder is trading a
50.
▲
by
inductive_magic
3y ago
>How is this data tracking any different from what google, twitter, or Facebook use? To use an application from one of these providers, you explicitly have to agree to their terms and services: a contract between you and the company is e
51.
▲
by
inductive_magic
3y ago
Great/inspiring leadership and VSOPs come to mind
52.
▲
by
inductive_magic
3y ago
According to ChatGPT, in the context of gps post: "CAC" stands for "Customer Acquisition Cost." It represents the cost a company incurs to acquire a new customer, including all the costs associated with marketing and adv
53.
▲
Mercedes Benz integrates ChatGPT in in-car voice control
(media.mercedes-benz.com)
1 points
by
inductive_magic
3y ago
|
1 comments
54.
▲
by
inductive_magic
3y ago
Randomly scrolled past that announcement, thought I'd share -
55.
▲
by
inductive_magic
3y ago
I mean, it refutes the argument. Everyone has access to the internet, to conv nets, to C/python/pandas/TCP, you name it. Yet, nobody would seriously argue that one can not build a moat with products based on those stacks for
56.
▲
by
inductive_magic
3y ago
Re: smart people avoiding wrapper work - sure! Not just because the result might be short lived, but also because there is a lot more prestige in building something from scratch rather than wrapping. However, the "very very temporary s
57.
▲
by
inductive_magic
3y ago
I wasn't trying to understate OpenAI's contribution, far from it. I can hardly express my appreciation for their work. That doesn't mean that everything else in the stack is window dressing though - custom, domain specific wr
58.
▲
by
inductive_magic
3y ago
As your wording implies, finetuning is restricted to the smaller models, i.e. babbage, curie etc. You can generate the training data for this with 3.5 and 4 and tune smaller models with the resulting data. For lots of tasks, this results in
59.
▲
by
inductive_magic
3y ago
100% of all existing code is just calling an OS-API that everyone else has access to. That's just nonsense
60.
▲
by
inductive_magic
3y ago
OpenAI builds an LLM and an api-interface to that model. The design of abstractions, prompt engineering, custom fine-tunes and software engineering required to ship a valuable application on top of that interface counts as "building an
More ›