cosminro

5 ms·

1.
A CUDA Kernel for Motion Blur(mlcrumbs.com)
2 points by cosminro 2y ago|0 comments
2.
by cosminro 5y ago
One important aspect of leetcode/coding contest problems is that they have a input size constraint and a time limit constraint. You can use the two to figure out what is the time complexity for a solution that would work. This simplifi
3.
by cosminro 9y ago
A lot of people reading the paper miss this. I guess it's not emphasized enough. In the first paper, the selfplay trained policy is about 1500 in elo rating, while darkforest2 a supervised trained policy from Facebook is around the sam
4.
by cosminro 9y ago
Attention is used in machine translation since 2014 as far as i know (Dzmitry Bahdanau, Kyunghyun Cho, Yoshua Bengio: Neural Machine Translation by Jointly Learning to Align and Translate) in various forms.
5.
by cosminro 9y ago
He is referring to this paper https://research.googleblog.com/2017/06/multimodel-multi-tas... Where they used one model to do image recognition, speech recognition and translation in the same network.
6.
by cosminro 10y ago
No batch norm for LSTMs
7.
by cosminro 10y ago
they probably used the meet in the middle technique: http://www.infoarena.ro/blog/meet-in-the-middle
8.
Ioffe and Szegedy get 4.8% top-5 test on ImageNet(beating recent MSR result)(arxiv.org)
1 points by cosminro 12y ago|0 comments
9.
Why We Think GoLang Is Ready For Early Stage Startups(jellolabs.com)
43 points by cosminro 13y ago|12 comments
10.
Numbers every algorithm designer should know(infoarena.ro)
17 points by cosminro 13y ago|0 comments
11.
by cosminro 14y ago
Here's some more background around Petr http://www.quora.com/Petr-Mitrichev/What-it-is-like-to-meet-... And some discussion around Russia and China dominating in coding contests http://www.quora.com/Why-do-people-from-Eastern-Europe-and-
12.
by cosminro 14y ago
Any others?
13.
by cosminro 15y ago
It takes time and practice. You won't get the practice at work. I dealt with a network flow problem a while back and a spell correction for URLs problem before that. In the last year I haven't done anything algorithmically challenging. Most
14.
What does "it's a good start" really mean.(okvivi.com)
200 points by cosminro 15y ago|43 comments
15.
by cosminro 15y ago
Picking up algorithm skills on the job is not that easy. You really have to spend some time, think problems though, play with them and internalize the learnings. Your job in any company doesn't deal with algorithms that much but it's very u
16.
by cosminro 15y ago
2002 is quite a while back, the problems have steadily increased in difficulty, try and look at a div 1 set now. Also the competition is much tougher as back then I think the site was mostly open to people from the US while now you can see
17.
by cosminro 15y ago
Yup, was thinking the same. He didn't do enough contests to escape higher volatility in the low ratings.
18.
by cosminro 15y ago
That rating is pretty low. Your average guy that gets hired at Google or Facebook is >1500 in rating probably.
19.
by cosminro 15y ago
Adam D'Angelo's (former FaceBook CTO) TopCoder profile: dangelo Algorithm Rating: 2351 Total Earnings: $3,082.50 School: California Institute of Technology http://www.topcoder.com/tc?module=MemberProfile&cr=26098...