habitat.datasets.utils module

Tokenize and vocabulary utils originally authored by @apsdehal and are taken from Pythia.

Classes

class VocabDict

class VocabFromText

Functions

def check_and_gen_physics_config()

def get_action_shortest_path(sim: sims.habitat_simulator.habitat_simulator.HabitatSim, source_position: typing.List[float], source_rotation: typing.List[float], goal_position: typing.List[float], success_distance: float = 0.05, max_episode_steps: int = 500) -> typing.List[core.simulator.ShortestPathPoint]

def load_str_list(fname)

def tokenize(sentence, regex = re.compile('([^\\w-]+)'), keep = "'s", remove = (',', '?')) -> typing.List[str]

Data

DEFAULT_PHYSICS_CONFIG_PATH = 'data/default.physics_config.json'

SENTENCE_SPLIT_REGEX = re.compile('([^\\w-]+)')