Exercise 4 - w4_unittest.test_back_prop(back_prop) is wrong

Course Q&A Natural Language Processing NLP with Probabilistic Models
21

I was making my mistake in calculating grad_b1 and grad_b2.

For grad_b1 I was doing it like this,
grad_b1 = np.sum((1/batch_size)*np.dot(l1,x.T),axis=1,keepdims=True).

But I actually had to take the sum of l1 along the rows and then divide by batch_size. Hope this will help you

1 Like
22

I found my mistake. This post can be deleted.

23

hi bro i am still unable to find the solution so can you share me your code plssss

24

I’m not get it, could u plz expand it or just mention that grade_b1 line code plz

25

@Mubsi , Hi , I have tried programming exercises of NLP Course 2 and got some errors in Week 3 and Week 4.
For Week 4, Exercise 4 - back_prop, I am getting TypeError: only integer scalar arrays can be converted to a scalar index. Similarly for Exercise 5.
Please find my lab id and kindly help as I have only 2 days left for this week and have to do remaining courses by this month end. Thanks
zdyjykxmukbm

26

Hi @Riddhima_Sobti,

There were a few mistakes in your implementation.

Your calculation for grad_b1 and grad_b2 were incorrect.

As well as, for anyone else reading this, np.max is not the same as np.maximum. This is why you were getting the scalar related error @Riddhima_Sobti.

You can take a look at your notebook for more details.

Best,
Mubsi

1 Like
27

Thanks a lot for your help. I cannot express how thankful I am to you for resolving my queries so promptly.

1 Like
28

I still didn’t get the solution. can you help?

29

can you please share to me the code,
I am actually trying everything but it doesn’t work
I will be really grateful to you

30

Dear friends,
It is seems a lit confusing!

Wrote code for grad_b1 based on Video: (here is the screenshot from video)

image
image986×186 23.1 KB

Code

  1. Used np.sum(A, axis=1,keepdims=True)
  2. step(z1) => z1

Problem

Wrong output values for gradient of b1 vector.
	 Expected: [[ 0.56665733]
 [ 0.46268776]
 [ 0.1063147 ]
 [-0.17481454]
 [ 0.11041817]
 [ 0.32025188]
 [-0.51827161]
 [ 0.08430878]
 [ 0.19341   ]
 [ 0.08339139]
 [-0.35949678]
 [-0.13053946]
 [ 0.19055422]
 [ 0.56405985]
 [ 0.13321988]] 
	Got: [[ 0.46004568]
 [ 0.40483404]
 [ 0.07318845]
 [-0.0995559 ]
 [ 0.01777417]
 [ 0.14951921]
 [-0.17889287]
 [ 0.01897284]
 [ 0.11459772]
 [ 0.02604062]
 [-0.32940889]
 [-0.11874333]
 [ 0.04899498]
 [ 0.06254933]
 [ 0.02570647]].
Wrong output values for gradient of b1 vector.
	 Expected: [[ 0.01864644]
 [-0.31966546]
 [-0.3564441 ]
 [-0.31703253]
 [-0.26702975]
 [ 0.14815984]
 [ 0.25794505]
 [ 0.24893135]
 [ 0.05895103]
 [-0.15348205]] 
	Got: [[ 0.0183693 ]
 [-0.25618039]
 [-0.01923447]
 [-0.06038765]
 [-0.12080929]
 [ 0.10414779]
 [ 0.08565018]
 [ 0.0896111 ]
 [ 0.05432164]
 [-0.14636517]].
 14  Tests passed
 2  Tests failed

Looking forward for your advices and directions!

Sardor

31

If I understand you correctly, you may be doing the Relu check AFTER the np.sum step while my understanding is that it needs to be done before the sum step.

Good luck.

32

@Mubsi Hi , I am getting error with week4 back_prop function can you please tell me what am i doing wrong lab id is jlqsufmomevo.
Thank You

33

Hi @OM_GAYDHANE,

Please share the error you are getting.

Thanks,
Mubsi

34

image
image833×670 8.42 KB

image
image861×722 12.2 KB

image
image693×411 5.59 KB

This is the error @Mubsi , please help me in finding where did I go wrong.

35

Hi @OM_GAYDHANE

It looks like you have a similar error as in this post . Please check the suggested solution.

Cheers

36

Hi @arvyzukai , I tried following that solution but do not seem to find the solution please guide me.