C3_W3_UNQ_C1: Not sure how to reduce the error to under 0.5

Course Q&A AI For Medicine AI For Medical Treatment
1 
Error from reference should be less than 0.05
Your error from reference:  0.05817829

Also I don’t think I used the last two lines for this part:

grade function codes removed and continuing to post codes can cause warning from community Team, please share codes only by personal DM if a mentor wants to look at your codes

Please remove the code snippet if that is not allowed.

I will dm the full codes if needed.

Sometimes I feel like this specialisation should replace Tensorflow with PyTorch or Jax.

2

if you are old learner, then posting codes even with prior information can cause action by community team.

Avoid posting codes at all.

Check how you reshape the spatial map output and gradient.

Another place to check is while computing weight and then using this weight to calculate the dot product of spatial map values with weights

While calculating first weight remember here axis arguments because instructions mentions make sure weights have shape C

1 Like
3

the reshaping seems to be correct. I used the correct axis attribute for the global average pooling via numpy mean function.

4

I found the problem, I did not put in the category_index at one of the lines. now it passes

1 Like