diff --git a/docs/src/common_workflows/entity_embeddings/notebook.ipynb b/docs/src/common_workflows/entity_embeddings/notebook.ipynb index 24b95e2..2893222 100644 --- a/docs/src/common_workflows/entity_embeddings/notebook.ipynb +++ b/docs/src/common_workflows/entity_embeddings/notebook.ipynb @@ -29,7 +29,7 @@ "cell_type": "markdown", "source": [ "In MLJFlux, the `NeuralNetworkClassifier`, `NeuralNetworkRegressor`, and the\n", - "`MultitargetNeuralNetworkRegressor`` can be trained and evaluated with heterogenous data\n", + "`MultitargetNeuralNetworkRegressor` can be trained and evaluated with heterogenous data\n", "(i.e., containing categorical features) because they have a built-in entity embedding\n", "layer. Moreover, they offer a `transform` method which encodes the categorical features\n", "with the learned embeddings. Such embeddings can then be used as features in downstream\n", @@ -146,1045 +146,1045 @@ "\n", "\n", "\n", - " \n", + " \n", " \n", " \n", "\n", - "\n", + "\n", "\n", - " \n", - " \n", + " \n", + " \n", " \n", "\n", - "\n", + "\n", "\n", - " \n", + " \n", " \n", " \n", "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n" + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n" ] }, "metadata": {}, @@ -1368,7 +1368,7 @@ { "output_type": "execute_result", "data": { - "text/plain": "trained Machine; caches model-specific representations of data\n model: NeuralNetworkBinaryClassifier(builder = Short(n_hidden = 5, …), …)\n args: \n 1:\tSource @068 ⏎ ScientificTypesBase.Table{Union{AbstractVector{ScientificTypesBase.Continuous}, AbstractVector{ScientificTypesBase.Multiclass{4}}, AbstractVector{ScientificTypesBase.Multiclass{3}}}}\n 2:\tSource @047 ⏎ AbstractVector{ScientificTypesBase.Multiclass{2}}\n" + "text/plain": "trained Machine; caches model-specific representations of data\n model: NeuralNetworkBinaryClassifier(builder = Short(n_hidden = 5, …), …)\n args: \n 1:\tSource @878 ⏎ ScientificTypesBase.Table{Union{AbstractVector{ScientificTypesBase.Continuous}, AbstractVector{ScientificTypesBase.Multiclass{4}}, AbstractVector{ScientificTypesBase.Multiclass{3}}}}\n 2:\tSource @932 ⏎ AbstractVector{ScientificTypesBase.Multiclass{2}}\n" }, "metadata": {}, "execution_count": 11 @@ -1430,91 +1430,92 @@ "output_type": "execute_result", "data": { "text/plain": "Plot{Plots.GRBackend() n=2}", - "image/png": "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", + "image/png": "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", "text/html": [ - "" + "" ], "image/svg+xml": [ "\n", "\n", "\n", - " \n", + " \n", " \n", " \n", "\n", - "\n", + "\n", "\n", - " \n", + " \n", " \n", " \n", "\n", - "\n", + "\n", "\n", - " \n", + " \n", " \n", " \n", "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", "\n", - " \n", + " \n", " \n", " \n", "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n", - "\n" + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n", + "\n" ] }, "metadata": {}, @@ -1586,9 +1587,9 @@ { "output_type": "execute_result", "data": { - "text/plain": "\u001b[1m5×5 DataFrame\u001b[0m\n\u001b[1m Row \u001b[0m│\u001b[1m C1_1 \u001b[0m\u001b[1m C1_2 \u001b[0m\u001b[1m C2_1 \u001b[0m\u001b[1m C2_2 \u001b[0m\u001b[1m R3 \u001b[0m\n\u001b[1m \u001b[0m│\u001b[90m Float32 \u001b[0m\u001b[90m Float32 \u001b[0m\u001b[90m Float32 \u001b[0m\u001b[90m Float32 \u001b[0m\u001b[90m Float64 \u001b[0m\n─────┼─────────────────────────────────────────────────────\n 1 │ -4.73264 -2.37719 -0.0605173 0.342042 0.445107\n 2 │ -3.61518 -4.22183 0.414665 0.362535 0.841927\n 3 │ -4.73264 -2.37719 -0.0605173 0.342042 0.637375\n 4 │ 3.56793 3.30689 -0.655706 -1.89323 0.37165\n 5 │ 3.6619 2.98503 -0.655706 -1.89323 0.230953", + "text/plain": "\u001b[1m5×5 DataFrame\u001b[0m\n\u001b[1m Row \u001b[0m│\u001b[1m C1_1 \u001b[0m\u001b[1m C1_2 \u001b[0m\u001b[1m C2_1 \u001b[0m\u001b[1m C2_2 \u001b[0m\u001b[1m R3 \u001b[0m\n\u001b[1m \u001b[0m│\u001b[90m Float32 \u001b[0m\u001b[90m Float32 \u001b[0m\u001b[90m Float32 \u001b[0m\u001b[90m Float32 \u001b[0m\u001b[90m Float64 \u001b[0m\n─────┼───────────────────────────────────────────────────\n 1 │ -3.38608 3.63731 1.19541 -1.61518 0.348266\n 2 │ -3.35609 3.88752 1.20568 -1.18024 0.428207\n 3 │ -3.38608 3.63731 1.19541 -1.61518 0.430754\n 4 │ 2.92898 -4.10774 0.493239 -0.355268 0.957329\n 5 │ 4.02604 -2.68811 0.493239 -0.355268 0.241365", "text/html": [ - "
5×5 DataFrame
RowC1_1C1_2C2_1C2_2R3
Float32Float32Float32Float32Float64
1-4.73264-2.37719-0.06051730.3420420.445107
2-3.61518-4.221830.4146650.3625350.841927
3-4.73264-2.37719-0.06051730.3420420.637375
43.567933.30689-0.655706-1.893230.37165
53.66192.98503-0.655706-1.893230.230953
" + "
5×5 DataFrame
RowC1_1C1_2C2_1C2_2R3
Float32Float32Float32Float32Float64
1-3.386083.637311.19541-1.615180.348266
2-3.356093.887521.20568-1.180240.428207
3-3.386083.637311.19541-1.615180.430754
42.92898-4.107740.493239-0.3552680.957329
54.02604-2.688110.493239-0.3552680.241365
" ] }, "metadata": {}, diff --git a/docs/src/common_workflows/entity_embeddings/notebook.jl b/docs/src/common_workflows/entity_embeddings/notebook.jl index e62b46a..68a05bd 100644 --- a/docs/src/common_workflows/entity_embeddings/notebook.jl +++ b/docs/src/common_workflows/entity_embeddings/notebook.jl @@ -9,7 +9,7 @@ # employed in NLP architectures. # In MLJFlux, the `NeuralNetworkClassifier`, `NeuralNetworkRegressor`, and the -# `MultitargetNeuralNetworkRegressor`` can be trained and evaluated with heterogenous data +# `MultitargetNeuralNetworkRegressor` can be trained and evaluated with heterogenous data # (i.e., containing categorical features) because they have a built-in entity embedding # layer. Moreover, they offer a `transform` method which encodes the categorical features # with the learned embeddings. Such embeddings can then be used as features in downstream diff --git a/docs/src/common_workflows/entity_embeddings/notebook.md b/docs/src/common_workflows/entity_embeddings/notebook.md index 14bd2bd..7e65805 100644 --- a/docs/src/common_workflows/entity_embeddings/notebook.md +++ b/docs/src/common_workflows/entity_embeddings/notebook.md @@ -13,7 +13,7 @@ categorical feature into a dense continuous vector in a similar fashion to how t employed in NLP architectures. In MLJFlux, the `NeuralNetworkClassifier`, `NeuralNetworkRegressor`, and the -`MultitargetNeuralNetworkRegressor`` can be trained and evaluated with heterogenous data +`MultitargetNeuralNetworkRegressor` can be trained and evaluated with heterogenous data (i.e., containing categorical features) because they have a built-in entity embedding layer. Moreover, they offer a `transform` method which encodes the categorical features with the learned embeddings. Such embeddings can then be used as features in downstream diff --git a/docs/src/common_workflows/entity_embeddings/notebook.unexecuted.ipynb b/docs/src/common_workflows/entity_embeddings/notebook.unexecuted.ipynb index ec7fc0e..cffefd1 100644 --- a/docs/src/common_workflows/entity_embeddings/notebook.unexecuted.ipynb +++ b/docs/src/common_workflows/entity_embeddings/notebook.unexecuted.ipynb @@ -29,7 +29,7 @@ "cell_type": "markdown", "source": [ "In MLJFlux, the `NeuralNetworkClassifier`, `NeuralNetworkRegressor`, and the\n", - "`MultitargetNeuralNetworkRegressor`` can be trained and evaluated with heterogenous data\n", + "`MultitargetNeuralNetworkRegressor` can be trained and evaluated with heterogenous data\n", "(i.e., containing categorical features) because they have a built-in entity embedding\n", "layer. Moreover, they offer a `transform` method which encodes the categorical features\n", "with the learned embeddings. Such embeddings can then be used as features in downstream\n",