the useage of self.count

[TAY-985]

hello,
in train.py and test() function , they all call the function save_results() , i can not understand self.count very well . What is the role of self.count ?What to use it for？thanks for your answer

[JonghwanMun]

The class "Accumulator" collects the results (e.g., [email protected]) of individual instances and return their averaged value.
So, value means score of metric while count denotes the number of instances (or the number of iterations).
Therefore, when calling add function, the count is set to 1.

Please refer the get_average function in src/utils/accumulator.py file.

[TAY-985]

but self.counter[k].add(v/nb,1); in v/nb ,average operation has been performed。in my debugging，while in each net testing , i found self.count=1;so it's just return v/nb when calling get_average function?

[JonghwanMun]

At each iteration of test time, the value (i.e., score) is computed as averaged one for the batch (\i.e., v/nb) and count is added by 1 for the iteration.
Then, after processing all the iterations, the values are divided by the number of iterations (i.e. self.count).

For example, consider three iterations with values of [3,4,5] (=v/nb).
At first iteration, self.value = 3 / self.count = 1.
At second iteration, self.value = 7 (=3+4) / self.count = 2 (=1+1).
At third iteration, self.value = 12 (=3+4+5) / self.count = 3 (=1+1+1).
Finally, get_average function returns 4 (=12 / 3).