« first day (2721 days earlier)      last day (2064 days later) » 

03:16
@halirutan "a real word application" :D
 
1 hour later…
04:43
Does anyone have a good way to make your own insertfunctionhereLayer? @CarlLange maybe? Or know of a decent guide to fussing with them? I think ElementwiseLayer would be a good one to use, I saw a guide on Community, but I am just not confident about it, would like some more input from those who use it more often than I.
@halirutan @b3m2a1 I'll mention something. I did meet someone today who made a guide for feeding his cat/other how-to-do-this type tasks for when he is away from home so that anyone who is watching his cat/house can access it.
Why not make a workaround @halirutan? I know this is beside the point, but from what I gather, yes the cloud notebook interfaces are a great asset in the formulation of the wolframcloud architecture, but there is gold in them hills if you will. As I'm sure you know. You can do forms and such, and many other things that you can integrate to be invisible to the end user, but I understand your end here. @b3m2a1 that's a great point I don't think anyone has mentioned before. I, too, would enjoy
using something like wolfr.am/b3m2a1 or similar--couldn't you do this though? To simplify some expression, couldn't you use WolframAlpha? ;) isn't it the loss of data n whatnots after some months of no-use? You could write a script to poke it every so often, yeah? shake the contents if you will. I have also seen (and will now implement my own) the creation of a financial data checker-thingie. I'm still a novice at-best, though. But I think this is all valuable info that should be passed on.
@b3m2a1 would the features you mentioned be included with a paid account? I wonder if it is some level of giving the mouse a cookie? Like a free sample sort-of deal? In that, the cookie represents the things you now miss, and if you bought the pack of cookies, you'd find that they were all there? I would describe yourself and @halirutan as power-users, though, and I know you've done some pretty crazy stuff with the system in the past. I'd love to see it continue, and I'm sure it will, but it
would be cool if you got to relay this and it was heard :)
 
3 hours later…
07:58
@CATrevillian There used to be a way using the undocumented "MXLayer", but it seems significantly different in 12.
It's a feature that's been coming for forever, hopefully in 12.1 with netfunction
Ah, but I suppose it depends on what you want. You can use ElementwiseLayer to map a simple function, but that function has to be composed of only a small group of math symbols.
@C.E. More "real world" than
WordFrequencyData[{"dog", "cat"}]
:)
08:17
@CarlLange Essentially I'd solve a 2nd order ODE and pop it in there, I'm a bit of a slow worker though so I'll report back in a few hours after a nap and see what I did/do! I'll ask about MXLayer--I was told ElementwiseLayer lets you use any function that can be either made up of ones in that documented list or out of a pure/anonymous function.
 
9 hours later…
17:30
"Relation to square roots"
(*start*)
a = 1;(*"a" is a positive small integer*)
b = 0;
nn = 100;
TableForm[
L2 = Table[
LinearProgramming[
Table[1/n, {n, 1, k}], {Table[If[n == 1, k, 1], {n, 1, k}]}, {{1,
0}}, Table[
If[n == 1, {-1, 1}, {-a*(n - 1), b*(n - 1)}], {n, 1, k}]], {k,
1, nn}]];
t1 = Table[Sum[If[L2[[n, k]] <= -1, 1, 0], {k, 2, n}], {n, 2, nn}]
t2 = Table[Round[Sqrt[2/a*(n - 1)]], {n, 2, nn}]
t2 - t1
Count[%, 0] - nn + 1
(*end*)
17:58
Setting the variable b to a positive integer gives a linear instead of a square root bound on the partial sums of the Möbius inverse of the Harmonic numbers.
18:20
Is anyone else getting this weird behavior?
0
Q: Calling NetPortGradient changes the value of the neural network

TreFoxI'm using Wolfram Desktop 12. Let's load a simple neural net model and an image to test it on: imageId = NetModel["Wolfram ImageIdentify Net V1"] image=ExampleData[{"TestImage", "Apples"}] Now, when I run imageId[image, "TopProbabilities"] I get: {Red Delicious -> 0.959278} Now, let's get t...

18:48
Nevermind, my question was a dupe of this: mathematica.stackexchange.com/questions/145115/…
But the problem still doesn't seem to be fixed
18:59
How does one download a file (e.g. a paclet file) from Google Drive via Mathematica? File sits at my Drive account. I understand that the Google Drive url-s do not contain the file name - how does one automatically get that?
Quick question before I ask on the main site. I have a (black box) function $f$ that takes a permutation $P$ of the integers 1-100, and returns a score $f(P)$. I want to find the maximum of this function. Is there an easy way to set that up?
@TreFox version?
19:25
@IstvánZachar can you use whatever link you'd use to share to someone who could download it? just get from there?
It's incredibly hard to kill time on an intercontinental flight if seats are too short...
too.....short? @kirma
@CATrevillian Well... I'm probably the tallest guy on the flight. :I
(observing interesting swamplands of Siberia during the night...)
19:40
@CATrevillian Newest. Desktop 12
 
2 hours later…
21:15
@CATrevillian I did try to use the sharing link, but whatever is downloaded via e.g. URLDownload is a <!DOCTYPE html> instead of a paclety file (which it is) and resulting file size does not matter either.
 
1 hour later…
22:19
@IstvánZachar try renaming the URL as described at labnol.org/internet/direct-links-for-google-drive/28356
that didn't work by itself, but then try:
RenameFile[
URLDownload[
"https://drive.google.com/uc?export=download&id=FILEIDHERE"], "~/tmp.paclet"]
PacletInstall["tmp.paclet"]
maybe this could be streamlined

« first day (2721 days earlier)      last day (2064 days later) »