I see that the error suggests the variables shapes and outputs are wrong. I’ve been staring at my formulas for a while but couldn’t figure out what’s wrong yet…
~/work/release/W4A1/test_utils.py in multiple_test(test_cases, target)
140 print(’\033[92m’, success," Tests passed")
141 print(’\033[91m’, len(test_cases) - success, " Tests failed")
→ 142 raise AssertionError(“Not all tests were passed for {}. Check your equations and avoid using global variables inside the function.”.format(target.name))
143
AssertionError: Not all tests were passed for L_model_backward. Check your equations and avoid using global variables inside the function.
I think one of my mistakes was in the second loop (corrected from l-1 to l-2). Now the error code seems to suggest that I made a mistake with dZ (but I don’t think that’s the case…)
in L_model_backward(AL, Y, caches)
59 # YOUR CODE STARTS HERE
60 current_cache = caches[l-2]
—> 61 dA_prev_temp, dW_temp, db_temp = linear_activation_backward(dAL,current_cache,activation=“relu”)
62 grads[“dA” + str(l)] = dA_prev_temp
63 grads[“dW” + str(l + 1)] = dW_temp
in linear_activation_backward(dA, cache, activation)
22 # dA_prev, dW, db = …
23 # YOUR CODE STARTS HERE
—> 24 dZ=relu_backward(dA,activation_cache)
25 dA_prev,dW, db = linear_backward(dZ, linear_cache)
26
~/work/release/W4A1/dnn_utils.py in relu_backward(dA, cache)
54
55 # When z <= 0, you should set dz to 0 as well.
—> 56 dZ[Z <= 0] = 0
57
58 assert (dZ.shape == Z.shape)
IndexError: boolean index did not match indexed array along dimension 0; dimension is 1 but corresponding boolean dimension is 3
Hi, @Kimberly_Chan. It looks like you have an indexing error. In your backprop loop, you are picking up the l-2 element from the current cache. That’s a problem.
The loop initiates with for l in reversed(range(L-1)). The reversed function makes it a bit tricky. To see what’s going on, you can create your own cell with “Insert → Cell Below” and enter
for l in reversed(range(L-1)):
print(l)
You will see that you will want to change your index for caches. in your current_cache = ... statement.
in linear_activation_backward(dA, cache, activation)
22 # dA_prev, dW, db = …
23 # YOUR CODE STARTS HERE
—> 24 dZ = relu_backward(dA, activation_cache)
25 dA_prev, dW, db = linear_backward(dZ, linear_cache)
26
~/work/release/W4A1/dnn_utils.py in relu_backward(dA, cache)
54
55 # When z <= 0, you should set dz to 0 as well.
—> 56 dZ[Z <= 0] = 0
57
58 assert (dZ.shape == Z.shape)
IndexError: boolean index did not match indexed array along dimension 0; dimension is 1 but corresponding boolean dimension is 3
This is a different kind of “indexing” problem than the one Ken was pointing out earlier on this thread. You can have a look at the logic of the relu_backward function by clicking “File → Open” and then opening the appropriate python file (see the “import” block at the beginning of the notebook to figure out the name). What you will see is that this error means that the dA value and the cache value (which is just Z in this case) that you passed from linear_activation_backward to relu_backward don’t match. So now the question is why that happened. It could be that it’s the dA value, not the activation cache value, that is incorrect. Put some print statements to show the shapes and then compare that to your “dimensional analysis”.
I checked my functions and after following the logic I finally noticed that I was passing the wrong dA value to the linar_activation_backwards-function in the inner for loop. Thank you so much!
for anyone suffers from this error:
i have faced the same problem
the mistake i have done is passing dAL to the linear_activation_backward function of relu, you should pass grads[“dA” + str(l + 1)]
and pay your attention to current caches index as we loop over the reversed of L-1 there is no need to start the index of caches by l-1 or l-2, just use l like the following
