Tensorflow Python Node breaks after weights load

[NoKot333]

Hello everyone,

I'm trying to run python node in LabView, but script return error 1672. 

With script everything ok, but LabView have some problems.

[Vazgen_H]

Hello, NoKot333. Do you use any Numpy functions?

[Vazgen_H]

What is your LabVIEW version? 

[Vazgen_H]

I found the possible solution for this error. 

https://knowledge.ni.com/KnowledgeArticleDetails?id=kA03q000000YHqWCAW&l=en-US

[alexderjuengere · Accepted solution]

---

@Vazgen_H wrote:

Hello, NoKot333. Do you use any Numpy functions?

---

I can confirm, I have seen error 1672 when returning a numpy array which is generated by the python script

in the following, I write a 2d buffer to the python node, extract the first column of this 2d buffer, convert this column vector to a numpy array and return as 1d array of doubles "y".

import numpy as np
def return_cm(data):

    y = data[:, 0]
    X = data[:, 1:]

    y = np.array(y)
    return y

the length of the y-vector can be manipulated by choosing a different number of rows, in the following screenshot 1000:

Error 1672 does halt the .vi, when the numbers of rows in this vector is bigger ~ 10.000 , but will work corretly if less e.g. 1000

[Vazgen_H]

There are some Numpy functions, which are not supported by LabVIEW.

[Vazgen_H]

Hello again. Can you send me kwhFinal.h5 for testing code on my PC? 

[NoKot333]

Yes

[NoKot333]

21.0.1 (64-bit)

[NoKot333]

https://drive.google.com/file/d/1fKdtM1uDE_eXgi-g1IY1dAi12D0ZpKr2/view?usp=sharing