🎨 Run clippy on the project :)

src/algebra/mod.rs

@@ -82,7 +82,7 @@ impl<T: NeuraVectorSpace + ?Sized> NeuraVectorSpace for Box<T> {
 
 impl NeuraVectorSpace for dyn NeuraDynVectorSpace {
     fn add_assign(&mut self, other: &Self) {
-        <dyn NeuraDynVectorSpace>::add_assign(self, &*other)
+        <dyn NeuraDynVectorSpace>::add_assign(self, other)
     }
 
     fn mul_assign(&mut self, by: f64) {

src/algebra/vector.rs

@@ -34,7 +34,7 @@ impl<const LENGTH: usize, F> NeuraVector<LENGTH, F> {
         LENGTH
     }
 
-    pub fn iter<'a>(&'a self) -> std::slice::Iter<'a, F> {
+    pub fn iter(&self) -> std::slice::Iter<'_, F> {
         self.data.iter()
     }
 }

src/axis.rs

@@ -59,7 +59,7 @@ impl NeuraAxisBase for NeuraAxisAppend {
     type Err = NeuraAxisErr;
 
     fn shape(&self, inputs: &[NeuraShape]) -> Result<NeuraShape, NeuraAxisErr> {
-        let mut inputs = inputs.into_iter().map(|x| *x.borrow());
+        let mut inputs = inputs.iter().map(|x| *x.borrow());
         if let Some(mut res) = inputs.next() {
             for operand in inputs {
                 match (res, operand) {
@@ -82,7 +82,7 @@ impl<F: Clone + Default + Scalar> NeuraAxis<DVector<F>> for NeuraAxisAppend {
     type Combined = DVector<F>;
 
     fn combine(&self, inputs: &[impl Borrow<DVector<F>>]) -> Self::Combined {
-        assert!(inputs.len() > 0);
+        assert!(!inputs.is_empty());
         let mut res = Vec::with_capacity(inputs.iter().map(|vec| vec.borrow().len()).sum());
 
         for input in inputs {

src/gradient_solver/backprop.rs

@@ -44,7 +44,7 @@ where
     }
 
     fn score(&self, trainable: &Trainable, input: &Input, target: &Target) -> f64 {
-        let output = trainable.eval(&input);
+        let output = trainable.eval(input);
         self.loss.eval(target, &output).to_f64().unwrap()
     }
 }

src/gradient_solver/forward_forward.rs

@@ -129,8 +129,8 @@ trait ForwardForwardRecurse<Input, Network, Gradient> {
 
 impl<Act, Input> ForwardForwardRecurse<Input, (), ()> for NeuraForwardPair<Act> {
     #[inline(always)]
-    fn recurse(&self, _network: &(), _input: &Input) -> () {
+    fn recurse(&self, _network: &(), _input: &Input) {
-        ()
+        
     }
 }
 

src/layer/dropout.rs

@@ -51,7 +51,7 @@ impl<R: Rng + Clone + std::fmt::Debug + 'static> NeuraLayerBase for NeuraDropout
     type Gradient = ();
 
     fn default_gradient(&self) -> Self::Gradient {
-        ()
+        
     }
 
     fn output_shape(&self) -> NeuraShape {

src/layer/isolate.rs

@@ -69,7 +69,7 @@ impl NeuraLayerBase for NeuraIsolateLayer {
 
     #[inline(always)]
     fn default_gradient(&self) -> Self::Gradient {
-        ()
+        
     }
 
     fn output_shape(&self) -> NeuraShape {
@@ -87,7 +87,7 @@ impl<F: Clone + Scalar + Default> NeuraLayer<DVector<F>> for NeuraIsolateLayer {
             panic!("NeuraIsolateLayer expected a value of dimension {}, got a vector", self.start.dims());
         };
 
-        let res = DVector::from_iterator(end - start, input.iter().cloned().skip(start).take(end));
+        let res = DVector::from_iterator(end - start, input.iter().skip(start).cloned().take(end));
 
         (res, ())
     }

src/layer/lock.rs

@@ -32,7 +32,7 @@ impl<Layer: NeuraLayerBase> NeuraLayerBase for NeuraLockLayer<Layer> {
     }
 
     fn default_gradient(&self) -> Self::Gradient {
-        ()
+        
     }
 
     fn prepare_layer(&mut self, is_training: bool) {

src/layer/mod.rs

@@ -189,7 +189,7 @@ impl NeuraLayerBase for () {
 
     #[inline(always)]
     fn default_gradient(&self) -> Self::Gradient {
-        ()
+        
     }
 
     #[inline(always)]

src/layer/normalize.rs

@@ -41,7 +41,7 @@ impl NeuraLayerBase for NeuraNormalizeLayer {
     type Gradient = ();
 
     fn default_gradient(&self) -> Self::Gradient {
-        ()
+        
     }
 }
 

src/layer/softmax.rs

@@ -33,7 +33,7 @@ impl NeuraLayerBase for NeuraSoftmaxLayer {
     }
 
     fn default_gradient(&self) -> Self::Gradient {
-        ()
+        
     }
 }
 
@@ -76,7 +76,7 @@ impl<F: Float + Scalar + NumAssignOps> NeuraLayer<DVector<F>> for NeuraSoftmaxLa
         let mut epsilon = epsilon.clone();
 
         // Compute $a_{l-1,i} ° \epsilon_{l,i}$
-        hadamard_product(&mut epsilon, &evaluated);
+        hadamard_product(&mut epsilon, evaluated);
 
         // Compute $\sum_{k}{a_{l-1,k} \epsilon_{l,k}}$
         let sum_diagonal_terms = epsilon.sum();

src/network/graph/from.rs

@@ -19,7 +19,7 @@ impl<Data> FromSequential<NeuraSequentialLast, Data> for NeuraGraph<Data> {
         Self {
             output_index: nodes.len(),
             buffer_size: nodes.len() + 1,
-            nodes: nodes,
+            nodes,
             output_shape: input_shape,
         }
     }

src/network/graph/mod.rs

@@ -28,7 +28,7 @@ impl<Data> Clone for NeuraGraphNodeConstructed<Data> {
         Self {
             node: dyn_clone::clone_box(&*self.node),
             inputs: self.inputs.clone(),
-            output: self.output.clone(),
+            output: self.output,
         }
     }
 }

src/network/graph/node.rs

@@ -10,8 +10,8 @@ use crate::{
 use super::*;
 
 pub trait NeuraGraphNodePartial<Data>: DynClone + Debug {
-    fn inputs<'a>(&'a self) -> &'a [String];
+    fn inputs(&self) -> &[String];
-    fn name<'a>(&'a self) -> &'a str;
+    fn name(&self) -> &str;
 
     fn construct(
         &self,
@@ -112,7 +112,7 @@ where
 impl<Data: Clone, Axis: NeuraAxis<Data>, Layer: NeuraLayer<Axis::Combined, Output = Data>>
     NeuraGraphNodeEval<Data> for NeuraGraphNode<Axis, Layer>
 {
-    fn eval<'a>(&'a self, inputs: &[Data]) -> Data {
+    fn eval(&self, inputs: &[Data]) -> Data {
         let combined = self.axis.combine(inputs);
         self.layer.eval(&combined)
     }
@@ -184,11 +184,11 @@ where
     Layer::Constructed: NeuraLayer<Axis::Combined, Output = Data>,
     Layer::Err: Debug,
 {
-    fn inputs<'a>(&'a self) -> &'a [String] {
+    fn inputs(&self) -> &[String] {
         &self.inputs
     }
 
-    fn name<'a>(&'a self) -> &'a str {
+    fn name(&self) -> &str {
         &self.name
     }
 

src/network/graph/partial.rs

@@ -169,7 +169,7 @@ impl<Data: Clone + std::fmt::Debug + 'static> NeuraPartialLayer for NeuraGraphPa
 
             let (constructed, output_shape) = node
                 .construct(input_shapes)
-                .map_err(|e| NeuraGraphErr::LayerErr(e))?;
+                .map_err(NeuraGraphErr::LayerErr)?;
 
             shapes[index] = Some(output_shape);
 

src/network/residual/construct.rs

@@ -44,7 +44,7 @@ impl<Layer: NeuraPartialLayer, ChildNetwork: NeuraResidualConstruct, Axis: Neura
             .map_err(|e| NeuraRecursiveErr::Current(Layer(e)))?;
         let layer_shape = Rc::new(layer.output_shape());
 
-        if self.offsets.len() == 0 {
+        if self.offsets.is_empty() {
             return Err(NeuraRecursiveErr::Current(NoOutput));
         }
 
@@ -63,7 +63,7 @@ impl<Layer: NeuraPartialLayer, ChildNetwork: NeuraResidualConstruct, Axis: Neura
         let child_network = self
             .child_network
             .construct_residual(rest_inputs, rest_indices, current_index + 1)
-            .map_err(|e| NeuraRecursiveErr::Child(e))?;
+            .map_err(NeuraRecursiveErr::Child)?;
 
         Ok(NeuraResidualNode {
             layer,

src/network/residual/last.rs

@@ -67,7 +67,7 @@ impl NeuraLayerBase for NeuraResidualLast {
     type Gradient = ();
 
     fn default_gradient(&self) -> Self::Gradient {
-        ()
+        
     }
 
     fn output_shape(&self) -> NeuraShape {
@@ -98,7 +98,7 @@ impl NeuraNetworkRec for NeuraResidualLast {
     where
         Self::Layer: NeuraLayerBase,
     {
-        ()
+        
     }
 }
 
@@ -147,8 +147,7 @@ impl<Data: Clone> NeuraLayer<NeuraResidualInput<Data>> for NeuraResidualLast {
 
     fn eval_training(&self, input: &NeuraResidualInput<Data>) -> (Self::Output, ()) {
         let result: Rc<Self::Output> = input.clone().get_first()
-            .expect("Invalid NeuraResidual state: network returned no data, did you forget to link the last layer?")
+            .expect("Invalid NeuraResidual state: network returned no data, did you forget to link the last layer?");
-            .into();
 
         (unwrap_or_clone(result), ())
     }

src/network/residual/wrapper.rs

@@ -68,7 +68,7 @@ where
         epsilon: &Self::Output,
     ) -> Self::Gradient {
         self.layers
-            .get_gradient(&self.input_to_residual_input(input), intermediary, &epsilon)
+            .get_gradient(&self.input_to_residual_input(input), intermediary, epsilon)
     }
 
     fn backprop_layer(
@@ -79,7 +79,7 @@ where
     ) -> Data {
         unwrap_or_clone(
             self.layers
-                .backprop_layer(&self.input_to_residual_input(input), intermediary, &epsilon)
+                .backprop_layer(&self.input_to_residual_input(input), intermediary, epsilon)
                 .get_first()
                 .unwrap(),
         )

src/network/sequential/construct.rs

@@ -24,13 +24,13 @@ impl<Layer: NeuraPartialLayer, ChildNetwork: NeuraPartialLayer> NeuraPartialLaye
         let layer = self
             .layer
             .construct(input_shape)
-            .map_err(|e| NeuraRecursiveErr::Current(e))?;
+            .map_err(NeuraRecursiveErr::Current)?;
 
         // TODO: ensure that this operation (and all recursive operations) are directly allocated on the heap
         let child_network = self
             .child_network
             .construct(layer.output_shape())
-            .map_err(|e| NeuraRecursiveErr::Child(e))?;
+            .map_err(NeuraRecursiveErr::Child)?;
         let child_network = Box::new(child_network);
 
         Ok(NeuraSequential {

src/network/sequential/layer_impl.rs

@@ -75,10 +75,9 @@ impl<Input, Layer: NeuraLayer<Input>, ChildNetwork: NeuraLayer<Layer::Output>> N
         let transient_epsilon =
             self.child_network
                 .backprop_layer(&transient_output, &intermediary.1, incoming_epsilon);
-        let outgoing_epsilon =
-            self.layer
-                .backprop_layer(input, &intermediary.0, &transient_epsilon);
         
-        outgoing_epsilon
+
+        self.layer
+                .backprop_layer(input, &intermediary.0, &transient_epsilon)
     }
 }

src/network/sequential/mod.rs

@@ -70,7 +70,7 @@ impl<Layer, ChildNetwork> NeuraSequential<Layer, ChildNetwork> {
         Layer: NeuraLayer<Input>,
     {
         NeuraSequential {
-            layer: layer,
+            layer,
             child_network: Box::new(self),
         }
     }
@@ -80,7 +80,7 @@ impl<Layer> From<Layer> for NeuraSequential<Layer, NeuraSequentialLast> {
     fn from(layer: Layer) -> Self {
         Self {
             layer,
-            child_network: Box::new(NeuraSequentialLast::default()),
+            child_network: Box::<NeuraSequentialLast>::default(),
         }
     }
 }

src/network/sequential/tail.rs

@@ -2,6 +2,7 @@ use super::*;
 
 /// Last element of a NeuraSequential network
 #[derive(Clone, Debug, PartialEq, Copy)]
+#[derive(Default)]
 pub struct NeuraSequentialLast {
     shape: Option<NeuraShape>,
 }
@@ -28,7 +29,7 @@ impl NeuraLayerBase for NeuraSequentialLast {
 
     #[inline(always)]
     fn default_gradient(&self) -> Self::Gradient {
-        ()
+        
     }
 }
 
@@ -116,11 +117,7 @@ impl<Input: Clone> NeuraNetwork<Input> for NeuraSequentialLast {
     }
 }
 
-impl Default for NeuraSequentialLast {
+
-    fn default() -> Self {
-        Self { shape: None }
-    }
-}
 
 /// Operations on the tail end of a sequential network
 pub trait NeuraSequentialTail {
@@ -146,7 +143,7 @@ impl<Layer> NeuraSequentialTail for NeuraSequential<Layer, NeuraSequentialLast>
             layer: self.layer,
             child_network: Box::new(NeuraSequential {
                 layer,
-                child_network: Box::new(NeuraSequentialLast::default()),
+                child_network: Box::<NeuraSequentialLast>::default(),
             }),
         }
     }

src/train.rs

@@ -123,10 +123,10 @@ impl NeuraBatchedTrainer {
 
             for _ in 0..self.batch_size {
                 if let Some((input, target)) = iter.next() {
-                    let gradient = gradient_solver.get_gradient(&network, &input, &target);
+                    let gradient = gradient_solver.get_gradient(network, &input, &target);
                     gradient_sum.add_assign(&gradient);
 
-                    train_loss += gradient_solver.score(&network, &input, &target);
+                    train_loss += gradient_solver.score(network, &input, &target);
                 } else {
                     break 'd;
                 }
@@ -151,7 +151,7 @@ impl NeuraBatchedTrainer {
                 network.prepare_layer(false);
                 let mut val_loss = 0.0;
                 for (input, target) in test_inputs {
-                    val_loss += gradient_solver.score(&network, input, target);
+                    val_loss += gradient_solver.score(network, input, target);
                 }
                 val_loss /= test_inputs.len() as f64;
                 train_loss /= (self.batch_size * self.log_iterations) as f64;

src/utils.rs

@@ -26,7 +26,7 @@ impl<J: Iterator> Iterator for Chunked<J> {
             }
         }
 
-        if result.len() > 0 {
+        if !result.is_empty() {
             Some(result)
         } else {
             None
@@ -54,8 +54,8 @@ where
         if let Some(next) = self.iter.next() {
             // Base iterator is not empty yet
             self.buffer.push(next.clone());
-            return Some(next);
+            Some(next)
-        } else if self.buffer.len() > 0 {
+        } else if !self.buffer.is_empty() {
             if self.index == 0 {
                 // Shuffle the vector and return the first element, setting the index to 1
                 self.buffer.shuffle(&mut self.rng);