kremi

5 ms·

1.
by kremi 7mo ago
« Connaître est reconnaître » I don’t think essentialist explanations about how LLMs work are very helpful. It doesn’t give any meaningful explanation of the high level nature of the pattern matching that LLMs are capable of. And it draws a
2.
by kremi 7mo ago
> The model can be prompted to talk about competitive dynamics. It can produce text that sounds like adversarial reasoning. But the underlying knowledge is not in the training data. It’s in outcomes that were never written down. With all
3.
by kremi 8mo ago
Interesting, I didn't expect that. I guess stock markets have a good reason to worry about video game studios then. Now, makes me wonder if movie companies stocks were similarly affected when major video models were announced.
4.
by kremi 8mo ago
To understand whether AI world models could replace traditional video games, it might be useful to compare this to how LLMs have affected old-school role-playing games. LLMs are more mature and have been around longer than AI world models s
5.
by kremi 1y ago
Yes, the term "middleware" is unfortunate, this is abundantly clear. > you wouldn't get a new tool simply because you might need it No but you get a framework precisely because it's "batteries included": man
6.
by kremi 1y ago
To people who downvote my comment (and there are many): please elaborate. I like to learn and improve. A lot of comments here are just baseless negative comments. Please let’s have a real discussion.
7.
by kremi 1y ago
I think “middleware” is a bit of a misnomer in Next.js. It’s really an edge function that runs before your request hits the app -- quick header checks, routing, and other lightweight guards. It runs on the edge runtime, not on the app serve
8.
by kremi 2y ago
It’d be hard to draw any conclusion. A whistleblower must be under extreme stress and pressure which in itself in some way or other will increase the risk of death — so that has to be taken account before saying the plausible cause for the
9.
by kremi 2y ago
This is anecdotal but - A company I worked for wanted to be 100% focused on doing one thing. It was spending 10x more than it was making revenue. It went bankrupt. - Another company I worked for always insisted on not having all eggs in one
10.
by kremi 2y ago
Some of the replies here are pretty good, I basically agree with “if it works for your data scientists then why not”. I’m actually a software developer with 10 years experience and also happen to do data science. And found myself in situati
11.
by kremi 2y ago
I guess it won’t take very long before you can do PPC (or similar paid advertising) in LLM results. Google had to turn to paid advertising for profitability. OpenAI and the likes might have to do the same — competition is fierce and price a
12.
by kremi 2y ago
Sorry I don't know what to answer. I don't think what I do qualifies as "workload". I have a process that generates lots of data. I put it in a huge multi-indexed dataframe that luckily fits in RAM. I then slice out the
13.
by kremi 2y ago
> Take a look at the indexing code in pandas As the end-user, not quite my concern. > and the staggering complexity of what's possible to put inside square brackets and how to decipher its meaning I might not be aware of everythi
14.
by kremi 2y ago
Pandas has been working fine for me. The most powerful feature that makes me stick to it is the multi-index (hierarchical indexes) [1]. Can be used for columns too. Not sure how the cool new kids like polars or ibis would fare in that categ
15.
by kremi 2y ago
I’m pretty sure the thumbs up emoji or the slightly smiling emoji won’t get different interpretations based on font. And here the issue is not in font differences (or different pictures of the same thing getting different interpretations):
16.
by kremi 2y ago
Your question sounds like you want to know how the word is spelled, and no one would put two r’s at straw, so the model could be assuming that you’re asking whether it’s strawbery or strawberry. What happens if you ask the total number of o
17.
by kremi 2y ago
Checks are quite foolproof too - spelling out the number (a hundred eighty eight) and writing the number (188), like on checks. The paragraph you mention seems to worry more about mistakes more than tempering attempts though (taking 10x the
18.
by kremi 3y ago
Very cool. When I was living in Paris I wanted to do something similar (but without the driving / directions part). I just liked walking randomly in the city and wished someone could tell me about all the interesting little bits of cul
19.
by kremi 3y ago
That’s pretty interesting. Which makes me wonder: what do search engines do to keep fake news out of its search results? I guess it’s based on human moderators?
20.
by kremi 3y ago
> It correctly identifies him as a whistleblower, and specifically says he "was not involved in the payment of bribes... as claimed by an AI chatbot called ChatGPT". How can the bing search version of ChatGPT quote chatgpt? Is
21.
by kremi 4y ago
Some more drawbacks : 1. the plants are usually grown using fertilizers that are made out of, ironically, fossil fuels. 2. the process of turning the plant into fuel might also create emissions that need to be taken into account.
22.
by kremi 6y ago
Somehow, thinking about how much money Besos earns (is it 78 billions this year? [1]), I find smile's slogan "Experience feel-good shopping" a bad joke at best. [1] https://www.businessinsider.fr/us/what-
23.
by kremi 6y ago
What are the advantages of Influx-iOX compared to Apache Druid? There seems to be much overlap in feature set and target use cases: seperate storage from compute, using parquet for columnar storage, good at analytical workloads on timeserie
24.
by kremi 7y ago
Indeed, the author does recognize it: > The next major problem Carell wants to tackle is what reactions could have formed the sugar ribose, which needs to link to nucleobases before RNA can form. Possibly something resembling the Calvin
25.
by kremi 9y ago
Made by my brother to learn React and showcase his coding to prospective employers. All feedback welcome!
26.
Show HN: Blob War, fun and original game(blobwar.herokuapp.com)
1 points by kremi 9y ago|1 comments
27.
by kremi 12y ago
Kayak's explore is pretty nice when you have uncertainty: http://www.kayak.com/explore/BER/ Although unlike Google Flights, it doesn't let you search for specific dates.
28.
by kremi 12y ago
The problem with any paid API is that the API consumer needs to make money somehow so that he can offer the price of the service. How will the API consumer make money? I don't think Ads will generate enough revenue -- especially not on
29.
by kremi 12y ago
Yeah debuggex's output looks slightly better, on top of having a lot more options. Regulex however is open-source with nicely written javascript: https://github.com/JexCheng/regulex Here's a side by side comp