Could not broadcast input array from shape (80,2) into shape (200,2)


#1

Whenever I hit the Start Training button this error comes up. I am not really sure what is wrong.
Is it something wrong with my model or does it have to do with the dataset?


#2

yes there is something wrong with your model.
please share the snapshot of the model used along with its configuration.

Regards
Rajat


#3


This is the model


#4

Ok your model looks fine to me.
I can’t see any block with (200,2) dimension.
In that case please tell us more about the dataset you have used. It might possible few of your image have different input dimension. Try using resize option available in the data tab in order to remove size difference issue.

Thanks


#5

Hi there
I am seeing a similar error message - could not broadcast input array from shape (16,2) into shape (32,2).

All my input JPG images have already been resized to 100 x 100.

May i seek your assistance on this? Thank u very much.


#6

Hi can you check that is your training size is greater than your batch size.
This might be because your input training data have 32 samples and you are using batch size of 32.

Regards
Rajat


#7

Hi Rajat

My training size is 16 (8 in each of two categories - Ok & Positive). As for the remainders 2 goes to Validation and another 2 goes to Tests. Thus, the total number of images is 20.

I am not sure about the setting for ‘batch size’.


#8

Yes that’s why you are getting this error.
Go to hyperparameters tab and then change your batch size from 32 to less than 2. Since you have 16 images only so set your batch size to 1.