In course 1 week 3 optional lab “Gradient descent for logistic regression”:

```
def compute_gradient_logistic(X, y, w, b):
...
for i in range(m):
...
*err_i* = f_wb_i - y[i] # What does it refer to?!
...
```