osipov

23 ms·

61.
by osipov 7y ago
Most of the recent research is moving to GELU (Gaussian Error Linear Units) activation functions: https://arxiv.org/pdf/1606.08415.pdf
62.
by osipov 7y ago
Does this yield a more parallelizable algorithm compared to what we had previously?
63.
by osipov 7y ago
At Google Cloud the 20% time is just a recruiting tactic. Once you join you'll find out that it should be on top of your existing 100% FTE.
64.
by osipov 7y ago
You comment getting downvoted is all the evidence I need that HN went down the tubes over the past couples of years. Do you know of any good alternatives? Apparently dev.to is pretty good...
65.
by osipov 7y ago
The trouble with common sense is that is it so wicked subjective. So when we debate stuff, we at least try to have a pretense of objectivity.
66.
by osipov 7y ago
I find it fascinating that if you were to make an equivalent statement about the political realm, you'd be immediately branded a conspiracy theorist on HN. But when it comes down to labor management relations I guess HN is happy to upv
67.
by osipov 7y ago
My trusted close relative used to do classified work on "Buran", the Soviet Union's answer to the United States Space Shuttle program. According to him, the USSR military developed Buran to match what they saw as the space wa
68.
by osipov 7y ago
So do you admit that McK occasionally acts as a fluffer?
69.
by osipov 7y ago
Thanks for building this project! I'm a big fan. With that said, I think you should add a filter for "at least X searches per month" because right now Vue.JS appears right next to Juul, and these are vastly different market s
70.
by osipov 7y ago
4 day workweek is a wild goose chase. There is no consensus about which of 5 workdays to drop to get there, which means that people will lose the precious time slots that can be used by individuals to get together at the same time as a gro
71.
by osipov 7y ago
Great question. Both TensorFlow and PyTorch pretty much do nothing https://github.com/tensorflow/tensorflow/issues/20524
72.
by osipov 7y ago
I worked with DataCamp through their course authors program and found it to be a poorly run organization. They maintain a publicly accessible "wishlist" of courses that they claim to want to develop. Think twice before putting you
73.
by osipov 8y ago
Have you ever had any trauma to your head?
74.
by osipov 8y ago
Why do you think that throwing compute at the problem is absurd? Arguably that's the approach that produced the resurgence of interest in CNNs.
75.
by osipov 9y ago
Unfortunately today, Lviv remains hostile to Jews by celebrating[1] extreme nationalism. Something I experienced first hand due to my Jewish heritage. [1] https://en.wikipedia.org/wiki/Stepan_Bandera#Jews
76.
by osipov 9y ago
lyrebirds are native to Australia
77.
by osipov 9y ago
Yes. In the video there is a short snippet where the audio is shown as a Fourier transformed image on the screen and a user is annotating the image of the sound using red boxes. This is a part of the process to train the ML model to recogni
78.
by osipov 9y ago
Prove P != NP
79.
by osipov 10y ago
Some functions implemented using serverless frameworks (e.g. Lambda or OpenWhisk) are microservices.
80.
Is Serverless Computing Any Different from Heroku and Other Traditional PaaSes?(medium.com)
1 points by osipov 10y ago|0 comments
81.
by osipov 10y ago
yes i used the psycholinguistic services and tradeoff analytics. not much beyond that.
82.
by osipov 10y ago
Hi Alex. This follows on another announcement from Walmart around OneOps: https://github.com/oneops However I don't see these projects getting any sizable traction from the open source community. What's the point
83.
by osipov 10y ago
I did a cost analysis of Amazon GPUs vs. Softlayer GPUs and the latter gave me a better price/performance ratio.
84.
by osipov 10y ago
With Iraqi WMDs and Colin Powell's anthrax I think we got Russians in a distant second in this weapons race. We have a century old experience in using this weapon going at least as far back as Spanish sinking the USS Maine.
85.
by osipov 10y ago
How's this any better than https://trollingeffects.org/
86.
by osipov 10y ago
@mark_l_watson -- the paper also seems important in another respect -- in the conclusion the authors suggest abandoning loss functions as optimization objective functions in machine learning and replacing them with mutual information functi
87.
by osipov 10y ago
You're right, the "long running daemon" is the serverless platform, i.e. the process that hosts the serverless applications (microflows). In case of public cloud providers like AWS, there is ample spare capacity, so hosting (
88.
Is Serverless Computing Any Different from Heroku, OpenShift, and Other PaaSes?(medium.com)
3 points by osipov 10y ago|3 comments
89.
Is Serverless Computing Just ETL with a Makeover?(medium.com)
10 points by osipov 10y ago|2 comments
90.
Serverless Computing Architecture Patterns – Part 4(medium.com)
2 points by osipov 10y ago|0 comments

More ›