NN Classifier training seems not copying data completely

#1

I have created 3 spectral feature blocks each having 1,31,932 samples. But the training seems using only 18,778 samples. Below is the log:

Copying features from processing blocks...
[     1/131932] Merging DSP blocks...
Still copying features from processing blocks...
[ 15393/131932] Merging DSP blocks...
[18778/131932] Merging DSP blocks...
Copying features from processing blocks OK

Job started
Splitting data into training and validation sets...
Splitting data into training and validation sets OK

Training model...
Training on 15022 inputs, validating on 3756 inputs

When I was using single spectral block with same number of samples it was working correctly.

#2

Hi @naveen,

Thanks for posting! I was able to take a look at your project and it appears the number of windows is now correct, at least when training NN Classifier All:

image

Are you still seeing the issue? Which block did your screenshot come from?

Warmly,
Dan

#3

Hi @dansitu,

I have removed those spectral blocks which were causing issues. What you have seen is newly created blocks which are working correctly. I will recreate the problematic blocks again later today and let you know to check.

#4

Thank you, let me know and I’ll take a look.

#5

@dansitu Please check the same project. I have reproduced the issue.

#6

@naveen thanks, we can reproduce now. Working on a fix.

1 Like
#7

@naveen Fix is deploying now, will be live within an hour or two!

#8

@janjongboom Thanks a lot!

1 Like