fastvideo.configs.configs

Module Contents

Classes

DatasetType	Enumeration for different dataset types.
PreprocessConfig	Configuration for preprocessing operations.

Data

logger

API

class fastvideo.configs.configs.DatasetType

Bases: str, enum.Enum

Enumeration for different dataset types.

Initialization

Initialize self. See help(type(self)) for accurate signature.

HF

‘hf’

MERGED

‘merged’

classmethod choices() → list[str]

Get all available choices as strings for argparse.

classmethod from_string(value: str) → fastvideo.configs.configs.DatasetType

Convert string to DatasetType enum.

class fastvideo.configs.configs.PreprocessConfig

Configuration for preprocessing operations.

static add_cli_args(parser: fastvideo.utils.FlexibleArgumentParser, prefix: str = 'preprocess') → fastvideo.utils.FlexibleArgumentParser

Add preprocessing configuration arguments to the parser.

check_preprocess_config() → None

dataloader_num_workers: int

1

dataset_output_dir: str

‘./output’

dataset_path: str = <Multiline-String>

dataset_type: fastvideo.configs.configs.DatasetType

None

do_temporal_sample: bool

False

drop_short_ratio: float

1.0

flush_frequency: int

256

classmethod from_kwargs(kwargs: dict[str, Any]) → Optional[fastvideo.configs.configs.PreprocessConfig]

Create PreprocessConfig from keyword arguments.

max_height: int

480

max_width: int

848

model_path: str = <Multiline-String>

num_frames: int

163

preprocess_video_batch_size: int

2

samples_per_file: int

64

seed: int

42

speed_factor: float

1.0

train_fps: int

30

training_cfg_rate: float

0.0

video_length_tolerance_range: float

2.0

fastvideo.configs.configs.logger

‘init_logger(…)’