Enum DataCommands 

pub enum DataCommands {
    Audit {
        file: PathBuf,
        num_classes: usize,
        input_column: String,
        label_column: String,
        preamble_prefix: Option<String>,
    },
    Split {
        file: PathBuf,
        train: f64,
        val: f64,
        test: f64,
        label_column: String,
        seed: u64,
        output: PathBuf,
    },
    Decontaminate {
        file: PathBuf,
        reference: Vec<PathBuf>,
        ngram: usize,
        threshold: f64,
        json: bool,
    },
    Balance {
        file: PathBuf,
        strategy: String,
        label_column: String,
        num_classes: Option<usize>,
        seed: u64,
        output: Option<PathBuf>,
    },
}

Data quality pipeline subcommands (powered by alimentar).

Thin CLI wrappers around alimentar’s data utilities.

Variants

Audit

Audit a JSONL classification dataset for quality issues

Fields

file: PathBuf

Path to JSONL data file

num_classes: usize

Number of output classes (for label range validation)

input_column: String

Input text column name

label_column: String

Label column name

preamble_prefix: Option<String>

Preamble prefix to detect (e.g., “#!/”)

Split

Stratified train/val/test split preserving class proportions

Fields

file: PathBuf

Path to JSONL data file

train: f64

Training set fraction

val: f64

Validation set fraction

test: f64

Test set fraction

label_column: String

Label column name for stratification

seed: u64

Random seed for deterministic split

output: PathBuf

Output directory for split files

Decontaminate

Check training data for benchmark contamination via n-gram overlap

Fields

file: PathBuf

Path to training JSONL data file

reference: Vec<PathBuf>

Reference benchmark JSONL files to check against

ngram: usize

N-gram size for overlap detection

threshold: f64

Overlap threshold (0.0-1.0) above which a sample is flagged

json: bool

Output as JSON

Balance

Resample dataset to address class imbalance

Fields

file: PathBuf

Path to JSONL data file

strategy: String

Rebalancing strategy: oversample, undersample, sqrt-inverse

label_column: String

Label column name

num_classes: Option<usize>

Number of classes (for sqrt-inverse weight computation)

seed: u64

Random seed

output: Option<PathBuf>

Output file path (required for oversample/undersample)

Trait Implementations

impl Debug for DataCommands

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl FromArgMatches for DataCommands

fn from_arg_matches(__clap_arg_matches: &ArgMatches) -> Result<Self, Error>

Instantiate Self from ArgMatches, parsing the arguments as needed.

fn from_arg_matches_mut( __clap_arg_matches: &mut ArgMatches, ) -> Result<Self, Error>

Instantiate Self from ArgMatches, parsing the arguments as needed.

fn update_from_arg_matches( &mut self, __clap_arg_matches: &ArgMatches, ) -> Result<(), Error>

Assign values from ArgMatches to self.

fn update_from_arg_matches_mut<'b>( &mut self, __clap_arg_matches: &mut ArgMatches, ) -> Result<(), Error>

Assign values from ArgMatches to self.

impl Subcommand for DataCommands

fn augment_subcommands<'b>(__clap_app: Command) -> Command

Append to Command so it can instantiate Self via FromArgMatches::from_arg_matches_mut

fn augment_subcommands_for_update<'b>(__clap_app: Command) -> Command

Append to Command so it can instantiate self via FromArgMatches::update_from_arg_matches_mut

fn has_subcommand(__clap_name: &str) -> bool

Test whether Self can parse a specific subcommand