🐛 Fix broken tests, add NeuraSequentialLast

src/network/graph/from.rs

@@ -1,4 +1,7 @@
-use crate::network::residual::{NeuraAxisDefault, NeuraSplitInputs};
+use crate::network::{
+    residual::{NeuraAxisDefault, NeuraSplitInputs},
+    sequential::NeuraSequentialLast,
+};
 
 use super::*;
 
@@ -10,9 +13,9 @@ pub trait FromSequential<Seq, Data> {
     ) -> Self;
 }
 
-impl<Data> FromSequential<(), Data> for NeuraGraph<Data> {
+impl<Data> FromSequential<NeuraSequentialLast, Data> for NeuraGraph<Data> {
     fn from_sequential_rec(
-        _seq: &(),
+        _seq: &NeuraSequentialLast,
         nodes: Vec<NeuraGraphNodeConstructed<Data>>,
         input_shape: NeuraShape,
     ) -> Self {

src/network/residual/node.rs

@@ -115,7 +115,6 @@ impl<
 {
     #[inline(always)]
     fn output_shape(&self) -> NeuraShape {
-        todo!("output_shape for NeuraResidualNode is not yet ready");
         self.child_network.output_shape()
     }
 

src/network/sequential/construct.rs

@@ -2,17 +2,17 @@ use crate::err::NeuraRecursiveErr;
 
 use super::*;
 
-impl<Layer: NeuraPartialLayer> NeuraPartialLayer for NeuraSequential<Layer, ()> {
-    type Constructed = NeuraSequential<Layer::Constructed, ()>;
-    type Err = Layer::Err;
+// impl<Layer: NeuraPartialLayer> NeuraPartialLayer for NeuraSequential<Layer, ()> {
+//     type Constructed = NeuraSequential<Layer::Constructed, ()>;
+//     type Err = Layer::Err;
 
-    fn construct(self, input_shape: NeuraShape) -> Result<Self::Constructed, Self::Err> {
-        Ok(NeuraSequential {
-            layer: self.layer.construct(input_shape)?,
-            child_network: Box::new(()),
-        })
-    }
-}
+//     fn construct(self, input_shape: NeuraShape) -> Result<Self::Constructed, Self::Err> {
+//         Ok(NeuraSequential {
+//             layer: self.layer.construct(input_shape)?,
+//             child_network: Box::new(()),
+//         })
+//     }
+// }
 
 impl<Layer: NeuraPartialLayer, ChildNetwork: NeuraPartialLayer> NeuraPartialLayer
     for NeuraSequential<Layer, ChildNetwork>

src/network/sequential/layer_impl.rs

@@ -1,19 +1,11 @@
 use super::*;
 use crate::layer::{NeuraLayer, NeuraLayerBase};
 
-// impl<Layer: NeuraLayerBase> NeuraLayerBase for NeuraSequential<Layer, ()> {
-//     #[inline(always)]
-//     fn output_shape(&self) -> NeuraShape {
-//         self.layer.output_shape()
-//     }
-// }
-
 impl<Layer: NeuraLayerBase, ChildNetwork: NeuraLayerBase> NeuraLayerBase
     for NeuraSequential<Layer, ChildNetwork>
 {
     #[inline(always)]
     fn output_shape(&self) -> NeuraShape {
-        todo!("Have output_shape return Option");
         self.child_network.output_shape()
     }
 

src/network/sequential/lock.rs

@@ -8,11 +8,11 @@ pub trait NeuraSequentialLock {
     fn lock(self) -> Self::Locked;
 }
 
-impl NeuraSequentialLock for () {
-    type Locked = ();
+impl NeuraSequentialLock for NeuraSequentialLast {
+    type Locked = NeuraSequentialLast;
 
     fn lock(self) -> Self::Locked {
-        ()
+        self
     }
 }
 

src/network/sequential/mod.rs

@@ -42,7 +42,7 @@ pub use tail::*;
 /// instance.
 ///
 /// The operations on the tail end are more complex, and require recursively traversing the `NeuraSequential` structure,
-/// until an instance of `NeuraSequential<Layer, ()>` is found.
+/// until an instance of `NeuraSequential<Layer, NeuraSequentialLast>` is found.
 /// If your network feeds into a type that does not implement `NeuraSequentialTail`, then you will not be able to use those operations.
 #[derive(Clone, Debug)]
 pub struct NeuraSequential<Layer, ChildNetwork> {
@@ -76,11 +76,11 @@ impl<Layer, ChildNetwork> NeuraSequential<Layer, ChildNetwork> {
     }
 }
 
-impl<Layer> From<Layer> for NeuraSequential<Layer, ()> {
+impl<Layer> From<Layer> for NeuraSequential<Layer, NeuraSequentialLast> {
     fn from(layer: Layer) -> Self {
         Self {
             layer,
-            child_network: Box::new(()),
+            child_network: Box::new(NeuraSequentialLast::default()),
         }
     }
 }
@@ -154,7 +154,7 @@ where
 #[macro_export]
 macro_rules! neura_sequential {
     [] => {
-        ()
+        $crate::network::sequential::NeuraSequentialLast::default()
     };
 
     [ .. $network:expr $(,)? ] => {

src/network/sequential/tail.rs

@@ -1,5 +1,127 @@
 use super::*;
 
+/// Last element of a NeuraSequential network
+#[derive(Clone, Debug, PartialEq, Copy)]
+pub struct NeuraSequentialLast {
+    shape: Option<NeuraShape>,
+}
+
+impl NeuraPartialLayer for NeuraSequentialLast {
+    type Constructed = NeuraSequentialLast;
+
+    type Err = ();
+
+    fn construct(mut self, input_shape: NeuraShape) -> Result<Self::Constructed, Self::Err> {
+        self.shape = Some(input_shape);
+        Ok(self)
+    }
+}
+
+impl NeuraLayerBase for NeuraSequentialLast {
+    type Gradient = ();
+
+    #[inline(always)]
+    fn output_shape(&self) -> NeuraShape {
+        self.shape
+            .expect("Called NeuraSequentialLast::output_shape() without building it")
+    }
+
+    #[inline(always)]
+    fn default_gradient(&self) -> Self::Gradient {
+        ()
+    }
+}
+
+impl<Input: Clone> NeuraLayer<Input> for NeuraSequentialLast {
+    type Output = Input;
+    type IntermediaryRepr = ();
+
+    #[inline(always)]
+    fn eval_training(&self, input: &Input) -> (Self::Output, Self::IntermediaryRepr) {
+        (input.clone(), ())
+    }
+
+    #[inline(always)]
+    fn backprop_layer(
+        &self,
+        _input: &Input,
+        _intermediary: &Self::IntermediaryRepr,
+        epsilon: &Self::Output,
+    ) -> Input {
+        epsilon.clone()
+    }
+}
+
+impl NeuraNetworkBase for NeuraSequentialLast {
+    type Layer = ();
+
+    #[inline(always)]
+    fn get_layer(&self) -> &Self::Layer {
+        &()
+    }
+}
+
+impl NeuraNetworkRec for NeuraSequentialLast {
+    type NextNode = ();
+
+    #[inline(always)]
+    fn get_next(&self) -> &Self::NextNode {
+        &()
+    }
+
+    #[inline(always)]
+    fn merge_gradient(
+        &self,
+        rec_gradient: <Self::NextNode as NeuraLayerBase>::Gradient,
+        _layer_gradient: <Self::Layer as NeuraLayerBase>::Gradient,
+    ) -> Self::Gradient
+    where
+        Self::Layer: NeuraLayerBase,
+    {
+        rec_gradient
+    }
+}
+
+impl<Input: Clone> NeuraNetwork<Input> for NeuraSequentialLast {
+    type LayerInput = Input;
+    type NodeOutput = Input;
+
+    fn map_input<'a>(&'_ self, input: &'a Input) -> Cow<'a, Self::LayerInput> {
+        Cow::Borrowed(input)
+    }
+
+    fn map_output<'a>(
+        &'_ self,
+        _input: &'_ Input,
+        layer_output: &'a Input,
+    ) -> Cow<'a, Self::NodeOutput> {
+        Cow::Borrowed(layer_output)
+    }
+
+    fn map_gradient_in<'a>(
+        &'_ self,
+        _input: &'_ Input,
+        gradient_in: &'a Self::NodeOutput,
+    ) -> Cow<'a, Input> {
+        Cow::Borrowed(gradient_in)
+    }
+
+    fn map_gradient_out<'a>(
+        &'_ self,
+        _input: &'_ Input,
+        _gradient_in: &'_ Self::NodeOutput,
+        gradient_out: &'a Self::LayerInput,
+    ) -> Cow<'a, Input> {
+        Cow::Borrowed(gradient_out)
+    }
+}
+
+impl Default for NeuraSequentialLast {
+    fn default() -> Self {
+        Self { shape: None }
+    }
+}
+
 /// Operations on the tail end of a sequential network
 pub trait NeuraSequentialTail {
     type TailTrimmed;
@@ -10,13 +132,13 @@ pub trait NeuraSequentialTail {
 }
 
 // Trimming the last layer returns an empty network
-impl<Layer> NeuraSequentialTail for NeuraSequential<Layer, ()> {
-    type TailTrimmed = ();
+impl<Layer> NeuraSequentialTail for NeuraSequential<Layer, NeuraSequentialLast> {
+    type TailTrimmed = NeuraSequentialLast;
     // GAT :3
-    type TailPushed<T> = NeuraSequential<Layer, NeuraSequential<T, ()>>;
+    type TailPushed<T> = NeuraSequential<Layer, NeuraSequential<T, NeuraSequentialLast>>;
 
     fn trim_tail(self) -> Self::TailTrimmed {
-        ()
+        NeuraSequentialLast::default()
     }
 
     fn push_tail<T>(self, layer: T) -> Self::TailPushed<T> {
@@ -24,7 +146,7 @@ impl<Layer> NeuraSequentialTail for NeuraSequential<Layer, ()> {
             layer: self.layer,
             child_network: Box::new(NeuraSequential {
                 layer,
-                child_network: Box::new(()),
+                child_network: Box::new(NeuraSequentialLast::default()),
             }),
         }
     }