Merge pull request #173 from MilesCranmer/fix-sum-speeds

Improve aggregation speeds by using `eachindex` instead of `iterate`

Author: juliohm

src/losses.jl

@@ -35,7 +35,7 @@ include("losses/weighted.jl")
 Return sum of `loss` values over the iterables `outputs` and `targets`.
 """
 function sum(loss::SupervisedLoss, outputs, targets)
-  sum(loss(ŷ, y) for (ŷ, y) in zip(outputs, targets))
+  sum(i -> loss(outputs[i], targets[i]), eachindex(outputs, targets))
 end
 
 """
@@ -46,7 +46,7 @@ The `weights` determine the importance of each observation. The option
 `normalize` divides the result by the sum of the weights.
 """
 function sum(loss::SupervisedLoss, outputs, targets, weights; normalize=true)
-  s = sum(w * loss(ŷ, y) for (ŷ, y, w) in zip(outputs, targets, weights))
+  s = sum(i -> weights[i] * loss(outputs[i], targets[i]), eachindex(outputs, targets, weights))
   n = normalize ? sum(weights) : one(first(weights))
   s / n
 end
@@ -57,7 +57,7 @@ end
 Return mean of `loss` values over the iterables `outputs` and `targets`.
 """
 function mean(loss::SupervisedLoss, outputs, targets)
-  mean(loss(ŷ, y) for (ŷ, y) in zip(outputs, targets))
+  mean(i -> loss(outputs[i], targets[i]), eachindex(outputs, targets))
 end
 
 """
@@ -68,7 +68,7 @@ The `weights` determine the importance of each observation. The option
 `normalize` divides the result by the sum of the weights.
 """
 function mean(loss::SupervisedLoss, outputs, targets, weights; normalize=true)
-  m = mean(w * loss(ŷ, y) for (ŷ, y, w) in zip(outputs, targets, weights))
+  m = mean(i -> weights[i] * loss(outputs[i], targets[i]), eachindex(outputs, targets, weights))
   n = normalize ? sum(weights) : one(first(weights))
   m / n
 end