vladf

15 ms·

121.
by vladf 8y ago
> Calling TensorFlow API is probably easier then calling Basilica API 99% of the case. Maybe, but training/curating data appropriate for your application isn't. It's not in that state right now but I think this service cou
122.
Systems for the New Normal(sisu.ai)
10 points by vladf 8y ago|0 comments
123.
by vladf 8y ago
Do you mind elaborating?
124.
by vladf 8y ago
(This may be obvious, but I want to hear it from you). Why add "kind"? Why not just "actionable and specific"?
125.
by vladf 8y ago
Your definition of tolerant is incompatible with availability from CAP. The Paxos 3/7 minority in your example can't make progress when partitioned from the other 4 machines.
126.
by vladf 9y ago
Eig inv svd det, at least, in 2010. https://dpinte.wordpress.com/2010/01/15/numpy-performance-im...
127.
by vladf 9y ago
I'm sorry, isn't "objects belonging to class C may have property P" a fair way to say there are some members which have it and some which do not? I don't see how that's misleading. I was correcting your stateme
128.
by vladf 9y ago
well, also np.linalg routines that call LAPACK may be multithreaded.
129.
by vladf 9y ago
> Any other interpretation of the debate would be paradox, because, a descriptivist would have to describe it as prescriptive if many users think it's prescriptive. Why is this a paradox? Why is prescriptivity a result of the actio
130.
by vladf 9y ago
I disagree on all counts. The first program that would be generated by such a procedure would very likely be the constant program that happens to always respond with the correct questions to your test. It depends how you randomly generate c
131.
by vladf 9y ago
I'm interested but "Lean programming languages" and its varieties are difficult to Google. Would you mind posting a complete reference?
132.
by vladf 9y ago
Ah, that's a shame, it looks like that notebook is no longer available. Thanks for pointing it out. I'll remove the reference.
133.
NumPy Tricks and Pitfalls(nbviewer.jupyter.org)
262 points by vladf 9y ago|26 comments
134.
Beating TensorFlow Training In-VRAM(vlad17.github.io)
8 points by vladf 9y ago|0 comments
135.
by vladf 9y ago
It's math characters are nearly illegible. The font itself is OK, but the math it causes is rarely used for a reason.
136.
by vladf 9y ago
Warning: font is Palatino
137.
by vladf 9y ago
Simplicity of implementation and single threaded speed seem to be, well, implementation issues. Nonetheless, they are reasonable doubts about the project. However, my comment was mostly aimed at the other commenters who were saying multipro
138.
by vladf 9y ago
Are you saying that because a language is missing something, when considering a fix for that thing, the existence of other languages/solutions is an argument against that fix?
139.
by vladf 9y ago
There seem to be a lot of naysayers in the comments about removing the GIL. Multiprocess parallelism isn't always appropriate, so I find this to be a very promising change that will definitely make me want to switch to PyPy. Here are t
140.
by vladf 9y ago
Worked first try, with no problems! Thank you, really great solution!
141.
by vladf 9y ago
To be honest, I don't see how an increase in algorithmic judgments will necessarily lead to a culture where we are more blindly trusting of them. I think it's up in the air, and there are forces working both ways: the more present
142.
by vladf 9y ago
I wholeheartedly agree with the linked page (we should encourage this kind of scrutiny in algorithmic judgements). However, where there seems to be disagreement is that an algorithmic judgement is a step forward compared to offline, human
143.
by vladf 9y ago
Sure, but what I'm saying is that the above holds for non-mass surveillance as well. So if you're OK with offline systems of social reputation (which most people are, and it seems essential to the function of society), then you ow
144.
by vladf 9y ago
Still, why is that distinct from offline reputation (hence my original question still holds)? All second-hand information about another person's reputation is just as out-of-context when repeated by another person. Consider (1) your f
145.
by vladf 9y ago
I might be misunderstanding the point being made here, but at one part doesn't make sense to me. The digital reputation argument seems to be saying "big data is bad because your reputation (i.e., people's valuations of your p
146.
by vladf 9y ago
Yeah, looking at it with Bayes rule makes it sound more convincing. Here's what's assumed: * The guard can't express whether or not A was pardoned to A. * The guard is telling the truth. Then the named (named to prisoner A) e