textattack.commands package

TextAttack commands Package

AttackCommand class

class textattack.commands.attack_command.AttackCommand[source]

Bases: TextAttackCommand

The TextAttack attack module:

A command line parser to run an attack from user specifications.

static register_subcommand(main_parser: ArgumentParser)[source]

run(args)[source]

AttackResumeCommand class

class textattack.commands.attack_resume_command.AttackResumeCommand[source]

Bases: TextAttackCommand

The TextAttack attack resume recipe module:

A command line parser to resume a checkpointed attack from user specifications.

static register_subcommand(main_parser: ArgumentParser)[source]

run(args)[source]

AugmentCommand class

class textattack.commands.augment_command.AugmentCommand[source]

Bases: TextAttackCommand

The TextAttack attack module:

A command line parser to run data augmentation from user specifications.

static register_subcommand(main_parser: ArgumentParser)[source]

run(args)[source]

Reads in a CSV, performs augmentation, and outputs an augmented CSV.

Preserves all columns except for the input (augmneted) column.

BenchmarkRecipeCommand class

class textattack.commands.benchmark_recipe_command.BenchmarkRecipeCommand[source]

Bases: TextAttackCommand

The TextAttack benchmark recipe module:

A command line parser to benchmark a recipe from user specifications.

static register_subcommand(main_parser: ArgumentParser)[source]

run(args)[source]

EvalModelCommand class

class textattack.commands.eval_model_command.EvalModelCommand[source]

Bases: TextAttackCommand

The TextAttack model benchmarking module:

A command line parser to evaluatate a model from user specifications.

get_preds(model, inputs)[source]

static register_subcommand(main_parser: ArgumentParser)[source]

run(args)[source]

test_model_on_dataset(args)[source]

class textattack.commands.eval_model_command.ModelEvalArgs(dataset_by_model: str = None, dataset_from_huggingface: str = None, dataset_from_file: str = None, dataset_split: str = None, filter_by_labels: list = None, model: str = None, model_from_file: str = None, model_from_huggingface: str = None, random_seed: int = 765, batch_size: int = 32, num_examples: int = 5, num_examples_offset: int = 0)[source]