diff --git a/NEWS.md b/NEWS.md
index 2c18102661..9bf97ddb3b 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -1,8 +1,5 @@
 # Flux Release Notes
 
-## v0.13.5
-* `loadmodel!` now supports a [`filter` keyword argument](https://github.com/FluxML/Flux.jl/pull/2041)
-
 ## v0.13.4
 * Added [`PairwiseFusion` layer](https://github.com/FluxML/Flux.jl/pull/1983)
 
diff --git a/src/loading.jl b/src/loading.jl
index 26aa34f104..74d9c35685 100644
--- a/src/loading.jl
+++ b/src/loading.jl
@@ -33,7 +33,7 @@ _filter_children(f, children::NamedTuple) =
 _filter_children(f, children) = filter(f, children)
 
 """
-    loadmodel!(dst, src; filter = _ -> true)
+    loadmodel!(dst, src)
 
 Copy all the parameters (trainable and non-trainable) from `src` into `dst`.
 
@@ -43,9 +43,6 @@ Non-array elements (such as activation functions) are not copied and need not ma
 Zero bias vectors and `bias=false` are considered equivalent
 (see extended help for more details).
 
-Specify the predicate function `filter` to control what is recursed.
-A child node `x` in either `dst` and `src` is skipped when `filter(x) == false`.
-
 # Examples
 ```julia
 julia> dst = Chain(Dense(Flux.ones32(2, 5), Flux.ones32(2), tanh), Dense(2 => 1; bias = [1f0]))
@@ -66,19 +63,6 @@ false
 
 julia> iszero(dst[2].bias)
 true
-
-julia> src = Chain(Dense(5 => 2), Dropout(0.2), Dense(2 => 1))
-Chain(
-  Dense(5 => 2),                        # 12 parameters
-  Dropout(0.2),
-  Dense(2 => 1),                        # 3 parameters
-)                   # Total: 4 arrays, 15 parameters, 348 bytes.
-
-julia> Flux.loadmodel!(dst, src; filter = x -> !(x isa Dropout)) # skips loading Dropout
-Chain(
-  Dense(5 => 2, tanh),                  # 12 parameters
-  Dense(2 => 1),                        # 3 parameters
-)                   # Total: 4 arrays, 15 parameters, 316 bytes.
 ```
 
 # Extended help