shoyer

30 ms·

91.
by shoyer 9y ago
That comment states that TensorFlow does not handle a large number of operations well, not a large number of variables. There's a large difference between those statements :). But yes, the general point is correct, especially for mod
92.
by shoyer 9y ago
Agreed! Sadly it is difficult to fully emulate in Python without some form of delayed evaluation in the language.
93.
by shoyer 9y ago
Another pandas developer here. Numba can give you similar performance to what you'll see for highly optimized operations in pandas (e.g., for groupby-sum or a moving average), but you would have to write the low-level loops yourself. L
94.
by shoyer 9y ago
I think it's a Go joke. Just be glad it's not called "m"!
95.
by shoyer 9y ago
Take a look at their "Convenience API" ( http://toyplot.readthedocs.io/en/stable/convenience-api.html ) that does indeed make simple two-axis plots a one-liner. toyplot.plot(x, y) seems pretty hard to beat
96.
by shoyer 9y ago
I haven't used Grumpy at all, and unless it starts supporting C extension modules like NumPy I doubt I ever will. Google's numerical computing / machine learning stack (e.g., TensorFlow) is based on Python/C++.
97.
by shoyer 9y ago
This is spot on. There are plenty of jobs applying open source technologies, but far fewer building those libraries themselves. (I'm NumPy dev who works at Google on machine learning.)
98.
by shoyer 9y ago
If you're excited about non-scalar values in DataFrames, you should take a look at xarray ( http://xarray.pydata.org ), which implements a very similar idea in its Dataset class.
99.
by shoyer 9y ago
Matrix chain multiplication was added to NumPy recently, see numpy.linalg.multi_dot: https://docs.scipy.org/doc/numpy-1.13.0/reference/generated/...
100.
by shoyer 9y ago
There's no naming and shaming in "Yes, at Google". From the article: "Usually, the people in the complaints are not named."
101.
by shoyer 9y ago
It's an intriguing idea, but where would the money come from? You'd need to convince companies that this is worth spending on. Also, who funds development of essential upstream libraries? In my experience, such projects have much
102.
by shoyer 9y ago
Companies like Gilead have made a fortune on anti-viral cures. Case in point: their recent Hepatitis C vaccines, which pulled in $4B in one quarter last year [1]. So a lack of money to be made on cures is not why this isn't funded. [1]
103.
by shoyer 9y ago
Hey, that's my PhD advisor on the steering board! So clearly it must be legit :)
104.
by shoyer 10y ago
I imagine there might be serious anti-trust concerns with such an acquisition.
105.
by shoyer 10y ago
One of my favorite tricks as maintainer for a moderately popular project is asking every user to give themselves credit in the release notes, e.g., - Fixed a terrible bug that crashes lots of crashes. By John Smith. Contributors feel great
106.
by shoyer 10y ago
Most news venues accept advertisements, even from somewhat controversial sources. This does not mean that sponsors have any editorial influence, which would be a serious allegation. Unless you are accusing STAT of violating journalistic eth
107.
by shoyer 10y ago
The Python API (pystan) is also GPLv3.
108.
by shoyer 10y ago
Google's mission has been "to organize the world's information and make it universally accessible and useful" since day one [1]. You're confusing mission with strategy. [1] http://www.businessinsider.com&
109.
by shoyer 10y ago
Before you try any sort of ML, explore your data [1, 2]. Exploratory data analysis may very well tell you that there is absolutely no point in making a fancy predictive model at all. If a few heuristics get you 90% of the way to an optimal
110.
by shoyer 10y ago
Looking at the map from the page you link to, I see significant amounts of snowpack at the stations above Lake Oroville: https://www.wcc.nrcs.usda.gov/webmap_beta/index.html#version... We see the same thing in NOAA&#x2
111.
by shoyer 10y ago
It looks like they install (sell?) racks of computers as household "heaters". Scroll down to the "Win, win, win!" section on their homepage with a video. This is a cute idea but I am skeptical that it makes sense from ei
112.
by shoyer 10y ago
Send me an email (my username at google.com). We may be hiring an intern this summer for a physics project. Yes, Google does interesting science research as well as hiring physicists, though we have plenty of those, too. I'm one of fou
113.
by shoyer 10y ago
It's entirely standard at tech companies to have a non-solicitation clause in employment contracts that prohibits soliciting other employees up through a fixed period (e.g., one year) after you leave. IANAL but I did some research on t
114.
by shoyer 10y ago
"lib pandas" has become pandas 2.0. See https://github.com/pandas-dev/pandas2 for ongoing design discussions and development.
115.
Deep Learning for Detection of Diabetic Eye Disease(research.googleblog.com)
14 points by shoyer 10y ago|0 comments
116.
Trumpian Cure for the FDA’s Chronic Lethargy(wsj.com)
1 points by shoyer 10y ago|0 comments
117.
by shoyer 10y ago
Criticism that GMOs have "not accelerated increases in crop yields" misses the point. If you pick weeds by hand, even organic farming can have great yields! Are GMOs making food cheaper to produce? The answer is almost certainly y
118.
by shoyer 10y ago
This is a cute example, but it misses the mark. The efficient way to do fast nearest neighbor search is with a search tree (e.g., KDTree or BallTree), which brings down query time from linear to logarithmic in the number of items.
119.
by shoyer 10y ago
This is almost certainly against the Google Apps TOS. So do so at your own risk.
120.
by shoyer 10y ago
> Later stage investment is increasingly focused on science > > Even five years ago it was incredibly hard to get funding for science-based ventures... This seems wrong to me. This may be true for consumer facing products or tech i

More ›