update readme

README.md

@@ -21,7 +21,7 @@ problem). While each MLJ model has a simple default builder, users
 will generally need to define their own builders to get good results,
 and this will require familiarity with the [Flux
 API](https://fluxml.ai/Flux.jl/stable/) for defining a neural network
-chain. 
+chain.
 
 In the future MLJFlux may provided an assortment of more sophisticated
 canned builders.
@@ -37,7 +37,7 @@ standardization of input features.
 
 ```julia
 using MLJ
-import RDatasets 
+import RDatasets
 iris = RDatasets.dataset("datasets", "iris");
 y, X = unpack(iris, ==(:Species), colname -> true, rng=123);
 @load NeuralNetworkClassifier
@@ -58,7 +58,41 @@ NeuralNetworkClassifier(
     optimiser_changes_trigger_retraining = false) @ 1…60
 ```
 
-#### Inspecting evolution of out-of-sample performance 
+#### Incremental training
+
+```julia
+import Random.seed!; seed!(123)
+mach = machine(clf, X, y)
+fit!(mach)
+
+julia> training_loss = cross_entropy(predict(mach, X), y) |> mean
+0.89526004f0
+
+# reduce learning rate and add iterations:
+clf.optimiser.eta = clf.optimiser.eta / 5
+clf.epochs = clf.epochs + 5
+
+julia> fit!(mach, verbosity=2)
+[ Info: Updating Machine{NeuralNetworkClassifier{Short,…}} @ 1…42.
+[ Info: Loss is 0.9638
+[ Info: Loss is 0.9226
+[ Info: Loss is 0.9197
+[ Info: Loss is 0.9477
+[ Info: Loss is 0.9517
+Machine{NeuralNetworkClassifier{Short,…}} @ 1…42
+
+julia> training_loss = cross_entropy(predict(mach, X), y) |> mean
+0.87368965f0
+```
+
+#### Accessing the Flux chain (model)
+
+```julia
+julia> fitted_params(mach).chain
+Chain(Chain(Dense(4, 3, σ), Flux.Dropout{Float64}(0.5, false), Dense(3, 3)), softmax)
+```
+
+####  Evolution of out-of-sample performance
 
 ```julia
 r = range(clf, :epochs, lower=1, upper=200, scale=:log10)
@@ -104,7 +138,7 @@ model type | prediction type | `scitype(X) <: _` | `scitype(y) <: _`
 
 > Table 1. Input and output types for MLJFlux models
 
-#### Matrix input
+#### Non-tabular input
 
 Any `AbstractMatrix{<:AbstractFloat}` object `Xmat` can be forced to
 have scitype `Table(Continuous)` by replacing it with ` X =