Update readme to reflect API improvements.

Author: rxwei

README.md

@@ -26,7 +26,7 @@ let hiddenSize: Int = 10
 struct Model: Layer {
     var layer1 = Dense<Float>(inputSize: 4, outputSize: hiddenSize, activation: relu)
     var layer2 = Dense<Float>(inputSize: hiddenSize, outputSize: hiddenSize, activation: relu)
-    var layer3 = Dense<Float>(inputSize: hiddenSize, outputSize: 3, activation: {$0})
+    var layer3 = Dense<Float>(inputSize: hiddenSize, outputSize: 3, activation: identity)
 
     @differentiable(wrt: (self, input))
     func applied(to input: Tensor<Float>) -> Tensor<Float> {
@@ -42,6 +42,7 @@ struct Model: Layer {
 ```swift
 let optimizer = SGD<Model, Float>(learningRate: 0.02)
 var classifier = Model()
+let context = Context(learningPhase: .training)
 let x: Tensor<Float> = ...
 let y: Tensor<Float> = ...
 ```
@@ -53,7 +54,7 @@ One way to define a training epoch is to use the [`Differentiable.gradient(in:)`
 ```swift
 for _ in 0..<1000 {
     let 𝛁model = classifier.gradient { classifier -> Tensor<Float> in
-        let ŷ = classifier.applied(to: x)
+        let ŷ = classifier.applied(to: x, in: context)
         let loss = softmaxCrossEntropy(logits: ŷ, labels: y)
         print("Loss: \(loss)")
         return loss
@@ -62,11 +63,11 @@ for _ in 0..<1000 {
 }
 ```
 
-Another way is to make use of methods on `Differentiable` or `Layer` that produce a pullback (i.e. a backpropagation function). Pullbacks allow you to compose your derivative computation with great flexibility.
+Another way is to make use of methods on `Differentiable` or `Layer` that produce a backpropagation function. This allows you to compose your derivative computation with great flexibility.
 
 ```swift
 for _ in 0..<1000 {
-    let (ŷ, backprop) = classifier.valueWithPullback(at: x)
+    let (ŷ, backprop) = classifier.appliedForBackpropagation(to: x, in: context)
     let (loss, 𝛁ŷ) = ŷ.valueWithGradient { ŷ in softmaxCrossEntropy(logits: ŷ, labels: y) }
     print("Model output: \(ŷ), Loss: \(loss)")
     let 𝛁model = backprop(𝛁ŷ)