zawerf

16 ms·

91.
by zawerf 8y ago
This reminds me of the audio clip where you can flip between whether you heard Yanny or Laurel: https://www.nytimes.com/interactive/2018/05/16/upshot/audio-...
92.
by zawerf 8y ago
I remember seeing a video demoing how to make sand flow like water by bubbling air into it: https://youtube.com/watch?v=My4RA5I0FKs
93.
by zawerf 8y ago
This is a similar project that showed up on hn a while ago: http://omrelli.ug/g9/gallery/
94.
by zawerf 8y ago
Is there any other way at google scale? I haven't written more than 100k LOC of C++ and yet there are already lots of code (mostly template meta programming) that I wrote that even I can't read anymore. Now imagine blowing that up
95.
by zawerf 8y ago
Does it have to be the face? Just wondering if they could've used another combination of body parts with sufficient variation and use those as the "fingerprint" instead. Rather than "the bristles, the snout, the eyes and
96.
by zawerf 8y ago
I think that's just a corollary of "63% Of Americans Don't Have Enough Savings To Cover A $500 Emergency": https://www.forbes.com/sites/maggiemcgrath/2016/01/06/63-of-... EDIT: M
97.
by zawerf 8y ago
If you can't make it to singapore, a lot of malaysian restaurants have them in NYC: https://www.yelp.com/search?find_desc=stingray&find_loc=nyc&...
98.
by zawerf 8y ago
It's pretty interesting that their training set consists of "outbound links from Reddit which received at least 3 karma". There are definitely large subreddits which are flooded by highly voted fake news which you don't
99.
by zawerf 8y ago
Probably one of the most unique designs for a personal blog I have ever seen (the animated page headers, page transitions, and a lot of genuinely great articles and talks): https://acko.net
100.
by zawerf 8y ago
Google's own tutorials for node.js recommends using passport.js with google oauth: https://cloud.google.com/nodejs/getting-started/authenticate... But that library hasn't been updated to not use the depr
101.
by zawerf 8y ago
I do something like this with their CLI to ignore node_modules folders for js dev on mac/linux: exclude_folders=$(find . -type d -name "node_modules") dropbox exclude add $exclude_folders dropbox exclude list #
102.
AlphaStar: Mastering the Real-Time Strategy Game StarCraft II(deepmind.com)
805 points by zawerf 8y ago|453 comments
103.
by zawerf 8y ago
If I remember correctly, in the mathematical model a global queue only helps with variance, but not the expected waiting time or throughput. In the real world you won't have the same service time for both cases since you need to add in
104.
by zawerf 8y ago
You can do a generator returning promises. It's actually preferred over async/await in the frontend world in libraries such as redux-saga and mobx flow. One advantage is that you can cancel the iteration early.
105.
by zawerf 8y ago
I am not sure you should put too much confidence in the "pwned password" search. I know one of the weak password I stupidly reuse everywhere was compromised since I had someone buy something with my paypal account. But it comes up
106.
by zawerf 8y ago
This reminded me of another guy who does graphics study on how frames are rendered in GTA V, DOOM, etc: http://www.adriancourreges.com/blog/2015/11/02/gta-v-graphic... Other games/discussions can be
107.
by zawerf 8y ago
He manually overrided some large subs: https://old.reddit.com/r/dataisbeautiful/comments/ae88pk/int...
108.
by zawerf 8y ago
Do you have a source on this? I was under the impression that machine learning is only good at learning things that can be approximated by continuous functions. A RNG is almost the complete opposite of that. It's everywhere discontinuo
109.
by zawerf 8y ago
Won't you be incentivized to prefer quantity over quality? You can easily scale the number of classrooms by 2x but it isn't as easy to 2x the starting salary of your graduates. The worry here is that you will over-focus on how to
110.
by zawerf 8y ago
Do compilers really try to prove UB so they can use it as an excuse to silently murder kittens? That sounds so counterproductive. If it can prove UB, it should just pop an error and stop.
111.
by zawerf 8y ago
A while back I wanted to learn about SLAM, computer vision, robotics, drones, etc. I did all the following courses which had significant overlap in materials: https://www.edx.org/course/autonomous-navigation-flying-robo
112.
by zawerf 8y ago
In particular are there any new materials that has been invented or widely adopted since the older textbooks? For example, bloom filters were rarely taught maybe 10 years ago but probabilistic data structures are now pretty mandatory in a d
113.
by zawerf 8y ago
See https://js.tensorflow.org/ for automatically porting tensorflow models to the browser
114.
by zawerf 8y ago
Yea the infinite scaling was referring to deployment of inference, not training.
115.
by zawerf 8y ago
> First of all, what you have described is far from the reality. Did you try any of the links I included? This is the reality for all of them and they are a few years old. They have a model file loaded from bucket url and never make anot
116.
by zawerf 8y ago
Cost as in money. To deploy a model in js (as a web page) all you need is a static S3 or GCS bucket. You don't even need a webserver and it can automatically handle infinite scale. Show me a python solution that can do the same.
117.
by zawerf 8y ago
One very practical reason is to save on computation costs. GPU servers are not cheap and if you're going to run on CPU anyway, you might as well run it locally on the user's browser. Some demos such as real time object detection i
118.
by zawerf 8y ago
I have not read his book but have seen that particular point float around from links to his blog: https://dilbertblog.typepad.com/the_dilbert_blog/2007/07/car... > Everyone has at least a few areas in whic
119.
by zawerf 8y ago
I found this summary of the book pretty good if you don't have time: https://www.reddit.com/r/slatestarcodex/comments/9w1b19/the_...
120.
by zawerf 8y ago
> Unsurprisingly lots of people have been in similar situations According to this article[1], you need $8.4 million to be in the top 1% of net worth so 1 out of 100 HN lurkers can probably also give useful first hand advice! (assuming be

More ›