carlossouza

7 ms·

31.
by carlossouza 3y ago
> We observed that all the VLMs tend to be confident while being wrong. Interestingly, we observed that even when the entropy was high, models tried to provide a nonsensical rational, instead of acknowledging their inability to perform t
32.
by carlossouza 3y ago
> We find inspiration not from the size of a market, but from the importance of the work. Because the importance of the work is the early indicator of a future market. > You are probably one ArXiv paper away from figuring this thing o
33.
by carlossouza 3y ago
> But that mistakes what a book is for. A book is a tool. It’s a machine for thinking. Best quote in the article. Taking the time to read, absorb and apply a book changes us in a way no summary or audiobook could ever change.
34.
by carlossouza 3y ago
The non-profit-good-for-humanity speech is such a hypocrisy. All incentives all different stakeholders (employees, leadership, board, investors) have are all aligned with the full for-profit side of the organization. In fact, that’s the onl
35.
by carlossouza 3y ago
Congrats on the project! Here’s an idea to monetize it: implement collaborative filtering (example: funds in rows, assets in columns). Once you do it, you will be able to cluster similar funds. Let’s say X funds have the asset A, but Y fund
36.
by carlossouza 3y ago
Answer.ai is truly open AI. :)
37.
Machine learning and information theory concepts towards an AI Mathematician(arxiv.org)
3 points by carlossouza 3y ago|1 comments
38.
by carlossouza 3y ago
Yoshua Bengio, Nikolay Malkin The current state-of-the-art in artificial intelligence is impressive, especially in terms of mastery of language, but not so much in terms of mathematical reasoning. What could be missing? Can we learn somethi
39.
by carlossouza 3y ago
Pg_vector is great! I’m using it at scale here: https://trendingpapers.com/ But it’s Postgres + pg_vector only, no Supabase
40.
by carlossouza 3y ago
Their name is as brilliant as “The Democratic People's Republic of Korea” (AKA North Korea)
41.
by carlossouza 3y ago
> The Open in openAI means that everyone should benefit from the fruits of AI after its built, but it's totally OK to not share the science... Based on this line of reasoning, ANY company that builds any given technology and intends
42.
by carlossouza 3y ago
> Sutherland was impressed by the significant dose of luck that had led to the discovery of murmurations. Talented people + hard work… + LUCK! > Even then, the murmurations were only found because of Pozdnyakov’s inexperience. Also, f
43.
by carlossouza 3y ago
Totally. People tend to forget it's just code. In fact, these public failures from Google AI provide an important public service: they remind us that there's nothing magical about LLMs. It's just code, and it is hard as hell
44.
by carlossouza 3y ago
On AWS Cost and usage control panel: > Forecasted month end costs > There isn't enough historical data to forecast your spend
45.
by carlossouza 3y ago
> The only regret I have is that I did not start selling my software earlier and only published free, open-source software. Well… better late than never. Congrats!
46.
by carlossouza 3y ago
Someone smart once said: If it is good, we call it "creativity." If it is bad, we call it "hallucination." This isn't a bug (or limitation, as the authors say). It's a feature.
47.
How Susceptible Are Large Language Models to Ideological Manipulation?(arxiv.org)
2 points by carlossouza 3y ago|0 comments
48.
by carlossouza 3y ago
Congrats on the launch! Does it search within pdf files?
49.
by carlossouza 3y ago
The Cobra Effect https://en.wikipedia.org/wiki/Perverse_incentive Classic
50.
by carlossouza 3y ago
https://www.youtube.com/watch?v=bRFLE9qi3t8 Here's the CEO of Builder.io supporting your comment: he says they tried LLMs/agents, and it didn't work. Then, they collected a dataset and developed an in-house m
51.
by carlossouza 3y ago
Great comment. I bet there are at least two markets (or niches): 1. People who want the absolute best performance and the latest possible version and are willing to pay the premium for it; 2. People who want to trade performance by cost and
52.
by carlossouza 3y ago
Because the supply for this market is constrained. It's a pure business decision based on simple math. If the estimated revenues from selling to the underserved market are higher than the cost of funding the project (they probably are,
53.
by carlossouza 3y ago
How old is he?
54.
by carlossouza 3y ago
Not yet... great idea, thanks! We’ll add to the roadmap
55.
by carlossouza 3y ago
Yes, totally! That's a great input, thanks!
56.
by carlossouza 3y ago
Thanks! That's something we are considering, given how many people mentioned it and offered help. Right now, the help we need is writing a paper detailing all aspects of the system and comparing it with similar systems. We have a group
57.
by carlossouza 3y ago
Thanks! Yes, that's in the roadmap; I believe it will be complete in 3-4 cycles. Right now, we are finishing the topics module. It will provide us with more granular filtering tools, and enable us to also discover trending topics, etc.
58.
by carlossouza 3y ago
The growing pace of new papers, especially in CS, is a hard problem to solve. We created this tool to help: https://trendingpapers.com/ The website was launched in October. It is already used by researchers in most of the l
59.
by carlossouza 3y ago
Google is far from doing just fine. Rome wasn't built in a day, and neither was it destroyed by a single incident or individual. The downfall of empires gradually happens over years... decades... But make no mistake: the sign of the fi
60.
by carlossouza 3y ago
Difficult, yes… but not impossible :) I just extract the titles and look for their respective ids. The real challenge was how to do that at scale. Only in CS there are well over half a million papers

More ›