This is the only way to construct a VW Predictor. The goal here is to provide a typesafe way of getting an predictor which will return the correct ou

Helper function to create an example object associated with this Workspace instance. Args: stringOrDict: See description in :py:m

stop VW by calling finish (and, eg, write weights to disk)

<summary> Creates the VW string for this example. </summary> <returns></returns>

default iterator for values & features

Then use it as the seed for hashing features.

Prediction without learning from the example. @param example an example upon which the prediction is based @return a prediction.

Get the weight at a particular position in the (learned) weight vector. Args: index (int): position in the learned weigh

Default log method routes to driver channel for compatibility.

============================================================ Helper: learn+finish a multi_ex for cb_explore_adf tests ================================

Get the weight based on the feature name and the namespace name. Args feature_name: The name of the feature namespace

Make a prediction on the example Args: ec: Examples of which to get a prediction from. If using a single object the learner must

<summary> Injects the supplied prefix into the stream. </summary> <returns>True if another token is available, false otherwise.</returns>

Main method that converts the dataframe to the VW format. Returns: The list of parsed lines in VW format.

Fit the model according to the given training data TODO: For first pass create and store example objects. For N-1 passes use exam

* Returns the enum value of the prediction type corresponding to the problem type

The following function is a home made strtof. The differences are : - much faster (around 50% but depends on the string to parse) - less error contro

re-entering from extension