label1,label2,...labelk ft1:ft1_val ft2:ft2_val ft3:ft3_val .. ftd:ftd_val

$ cargo build --release

$ craftml train --help

craftml-train
Train a new CRAFTML model

USAGE:
    craftml train [FLAGS] [OPTIONS] <training_data>

FLAGS:
    -h, --help                 Prints help information
        --test_trees_singly    Test forest tree by tree, freeing each before training the next to reduce memory usage.
                               Model cannot be saved.
    -V, --version              Prints version information

OPTIONS:
        --centroid_min_n_preserve <centroid_min_n_preserve>
            The minimum number of entries to preserve from puning, regardless preserve ratio setting. [default: 10]

        --centroid_preserve_ratio <centroid_preserve_ratio>
            A real number between 0 and 1, which is the ratio of entries with largest absoulte values to preserve. The
            rest of the entries are pruned. [default: 0.1]
        --cluster_sample_size <cluster_sample_size>
            Number of examples drawn for clustering on a branching node [default: 20000]

        --cv_splits_path <PATH>
            Path to the k-fold cross validation splits file, with k space-separated columns of indices (starting from 1)
            for training splits.
        --k_clusters <k_clusters>                              Number of clusters on a branching node [default: 10]
        --leaf_max_size <leaf_max_size>
            Maximum number of distinct examples on a leaf node [default: 10]

        --model_path <PATH>                                    Path to which the trained model will be saved if provided
        --n_cluster_iters <n_cluster_iters>
            Number of clustering iterations to run on each branching node [default: 2]

        --n_feature_buckets <n_feature_buckets>
            Number of buckets into which features are hashed [default: 10000]

        --n_label_buckets <n_label_buckets>
            Number of buckets into which labels are hashed [default: 10000]

        --n_threads <n_threads>
            Number of worker threads. If 0, the number is selected automatically. [default: 0]

        --n_trees <n_trees>                                    Number of trees in the random forest [default: 50]
        --out_path <PATH>
            Path to the which predictions will be written, if provided

        --test_data <PATH>
            Path to test dataset file used to calculate metrics if provided (in the format of the Extreme Classification
            Repository)

ARGS:
    <training_data>    Path to training dataset file (in the format of the Extreme Classification Repository)

$ craftml test --help

craftml-test
Test an existing CRAFTML model

USAGE:
    craftml test [OPTIONS] <model_path> <test_data>

FLAGS:
    -h, --help       Prints help information
    -V, --version    Prints version information

OPTIONS:
        --k_top <k_top>            Number of top predictions to write out for each test example [default: 5]
        --n_threads <n_threads>    Number of worker threads. If 0, the number is selected automatically. [default: 0]
        --out_path <PATH>          Path to the which predictions will be written, if provided

ARGS:
    <model_path>    Path to the trained model
    <test_data>     Path to test dataset file (in the format of the Extreme Classification Repository)