Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
woko
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
15 ms
·
121.
▲
A bundle of O'Reilly about AI
(humblebundle.com)
2 points
by
woko
6y ago
|
0 comments
122.
▲
by
woko
6y ago
I think this post means the same as this tweet: > Transformers are a special case of Graph Neural Networks. This may be obvious to some. https://twitter.com/OriolVinyalsML/status/123378359362695168...
123.
▲
by
woko
6y ago
As others mentioned, it is a good habit to request the page to be archived. You don't have to link to the archive, but you would have the option to if the page were to disappear in the future. I wish I had done this 15 years ago for a
124.
▲
by
woko
6y ago
> do anarchists really work as professors at the London School of Economics? And Yale? I know a banker who is communist. It does not mean that he behaves like a communist in his actions, but he does in his words. From my perspective, he
125.
▲
by
woko
6y ago
> we’ve never moved permanently backwards before, so it would be rather strange if we started doing that now. It would not be strange. It would be unexpected. It reminds me of this funny little story: > "Every morning on a turkey
126.
▲
by
woko
6y ago
I have the same issue. I never waste time trying to find a way to forwards these emails to the right people, because I am extremely suspicious, so I often assume there is a phishing attempt behind these emails. I mean, often, money is menti
127.
▲
by
woko
6y ago
The issue is that new competitors have to go against... Apple and Google. Given what happened to Epic (retaliation on Unreal Engine after #FreeFortnite), you can be sure that any competitor which would get a little bit of traction would be
128.
▲
by
woko
6y ago
> What data and models were used to synthesize these predictions? You have to ask the creator of ArtBreeder ( https://www.artbreeder.com/about ) which apparently uses BigGAN, StyleGAN, etc., trained on different datasets t
129.
▲
by
woko
6y ago
> I'd be careful with computer crimes on the Internet though. Exactly. Let us say you break into a shop owned by some mafia to put out a fire, then you might be fine w.r.t. authority, but you might be in trouble w.r.t. criminals. Si
130.
▲
by
woko
6y ago
The webpage is weird. For these interested, it seems to be about: https://github.com/DanielBMarkham/OnePageChess
131.
▲
by
woko
6y ago
The videos are very impressive. I wish you let us move the 3D scene with the mouse in the browser to get a better idea of the result.
132.
▲
by
woko
6y ago
Not a doctor, but I have had kidney stones. The answer is a loss of calcium, so high level found in urine test: if you had way too much vitamin D, then you would start releasing the calcium from your bones. I don't think that taking ca
133.
▲
by
woko
6y ago
I would call it "inheritance" of discrimination rather than "replication" of discrimination: the skin-colour bias arises because of the Ivy-League bias. Quotes below: > "[...] Facebook staffers say that despite a
134.
▲
by
woko
6y ago
As a non-native speaker, this website is useful to learn English slang.
135.
▲
by
woko
6y ago
I get your point, but I don't get how it diverges from what was described in the article. Based on what you wrote, it looks to me that you are actually very advanced in the process of forgiveness: telling your story, expressing your fe
136.
▲
by
woko
6y ago
True. I realize that I wrongly assumed that "revenge" was retaliation perpetrated by the victim outside of any legal system. If I understand right, some people consider that justice is a form of revenge. If the process of administ
137.
▲
by
woko
6y ago
You kind of answered the question: revenge does not help the victim. It might feel necessary, but it does not actually help.
138.
▲
by
woko
6y ago
Listen to some samples: - https://openai.com/blog/jukebox/ (2020, quite good, but no classical music) - https://openai.com/blog/musenet/ (2019 so not as good as the 2020 one, but showcas
139.
▲
by
woko
6y ago
> I'm not aware of parents that specifically believe learning music makes their kid more intelligent in non-musical areas. It is all anecdotal evidence, but I am aware of some parents who "forced" their baby (and then litt
140.
▲
by
woko
6y ago
Disclaimer: I only realize now that my comment is about images returned in image search results, not about reverse image search. Bing definitely uses facial recognition: faces of celebrities are detected, there is a bounding box around the
141.
▲
by
woko
6y ago
For me, it is a setting to tick, which mentions the country: https://i.imgur.com/fML3x0l.png (in French) Once it is set to my country (France), then I get almost identical results. I don't know how to deactivate this f
142.
▲
by
woko
6y ago
In my opinion, it is due to location. Bing returns 8 results on the first page, and DDG returns 10 results. At the top of the Bing page, I can see that Bing already takes my location into account (and it is precise, it is my city), so there
143.
▲
by
woko
6y ago
It is troubling that there was nobody in the meeting room to raise concerns about pushing notifications to ~250 millions[1] of users. [1] Monthly Active users at https://data.firefox.com/dashboard/user-activity
144.
▲
by
woko
6y ago
Thank you. I appreciate your conciseness. The presentation is similar to the (nice yet more verbose) notebook "01-g-h-filter.ipynb" in github.com/rlabbe/Kalman-and-Bayesian-Filters-in-Python/
145.
▲
by
woko
6y ago
I find it weird that this (free) book was omitted: - Goodfellow et al. Deep learning. MIT press, 2016. https://www.deeplearningbook.org/ And there is the book about DL with Python published by Manning in 2017, but not the
146.
▲
by
woko
6y ago
From your link: > The way this process works is that, while averaging all of the pixels in a series of, say, 10 images, the program mathematically calculates which pixels fall far away from the mean value because they're much bright
147.
▲
by
woko
6y ago
> generating a barrage of spam or fake news that uncritical readers can't distinguish from human output. Which could indirectly bring good things: what if this decreases our tolerance threshold for nonsensical texts? For instance, p
148.
▲
by
woko
6y ago
> I briefly skimmed (not expecting very much from a Bitcoin-themed article), read the end, and only then read more carefully. It sounds to me like you must be an academic, or someone with good habits for being efficient at reading articl
149.
▲
by
woko
6y ago
Same, except that I skipped straight to the last line to check whether it was a generated text after I noticed the first sentence made no sense (GPT-2 already generates grammatically correct English sentences).
150.
▲
by
woko
6y ago
What if the 8 non-verified accounts are alt-accounts used by celebs/VIPs for personal communication? Let us imagine that I am Jeff Bezos, why would I use my official account to DM people? I would rather use one where I look like everyb
More ›