I grabbed some test data from compute_cost_test and pasted it into a cell in a notebook and tried what i thought was the simplest version of vectorized logloss, but am getting an incorrect output, so I’m missing something, but don’t see what it is.

The relevant code is as follows:

```
np.random.seed(1)
Y_test = (np.random.randn(1, 5) > 0)
A2 = (np.array([[ 0.5002307 , 0.49985831, 0.50023963, 0.25, 0.7]]))
m = Y_test.shape[1]
expected_output = 0.5447066599017815
print("expected_output: %f" % expected_output)
actual_output = -np.sum((Y_test*np.log(A2)) + ((1 - Y_test) * (1 - np.log(A2))))/m
print("actual_output: %f" % actual_output)
```

I’m guessing that I have a simple error in my formula but I can’t seem to see it. I made sure that just using “*” worked the same as np. multiply by doing this

```
print("pos probs np.multiply: " +str(np.multiply(np.log(A2), Y_test)))
print("pos probs raw:" + str(Y_test*np.log(A2)))
```

The printed output is

```
expected_output: 0.544707
actual_output: -0.944606
pos probs np.multiply: [[-0.69268589 -0. -0. -0. -0.35667494]]
pos probs raw:[[-0.69268589 -0. -0. -0. -0.35667494]]
```

Any pointers would be greatly appreciated.

regards,

Michael