For the model pred = w * x with loss (pred - actual)², the gradient of the loss with respect to w is 2 * (pred - actual) * x. The 2 * (pred - actual) part is written. Fill the blank with the last factor (the input the weight was multiplied by).

Blank: