1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
|
"""Base lightning DataModule class."""
from pathlib import Path
from typing import Callable, Dict, Optional, Tuple, Type, TypeVar
import attr
from pytorch_lightning import LightningDataModule
from torch.utils.data import DataLoader
from text_recognizer.data.base_dataset import BaseDataset
from text_recognizer.data.mappings.base_mapping import AbstractMapping
T = TypeVar("T")
def load_and_print_info(data_module_class: type) -> None:
"""Load dataset and print dataset information."""
dataset = data_module_class()
dataset.prepare_data()
dataset.setup()
print(dataset)
@attr.s(repr=False)
class BaseDataModule(LightningDataModule):
"""Base PyTorch Lightning DataModule."""
def __attrs_pre_init__(self) -> None:
"""Pre init constructor."""
super().__init__()
mapping: Type[AbstractMapping] = attr.ib()
transform: Optional[Callable] = attr.ib(default=None)
test_transform: Optional[Callable] = attr.ib(default=None)
target_transform: Optional[Callable] = attr.ib(default=None)
train_fraction: float = attr.ib(default=0.8)
batch_size: int = attr.ib(default=16)
num_workers: int = attr.ib(default=0)
pin_memory: bool = attr.ib(default=True)
# Placeholders
data_train: BaseDataset = attr.ib(init=False, default=None)
data_val: BaseDataset = attr.ib(init=False, default=None)
data_test: BaseDataset = attr.ib(init=False, default=None)
dims: Tuple[int, ...] = attr.ib(init=False, default=None)
output_dims: Tuple[int, ...] = attr.ib(init=False, default=None)
@classmethod
def data_dirname(cls: T) -> Path:
"""Return the path to the base data directory."""
return Path(__file__).resolve().parents[2] / "data"
def config(self) -> Dict:
"""Return important settings of the dataset."""
return {
"input_dim": self.dims,
"output_dims": self.output_dims,
}
def prepare_data(self) -> None:
"""Prepare data for training."""
pass
def setup(self, stage: Optional[str] = None) -> None:
"""Split into train, val, test, and set dims.
Should assign `torch Dataset` objects to self.data_train, self.data_val, and
optionally self.data_test.
Args:
stage (Optional[str]): Variable to set splits.
"""
pass
def train_dataloader(self) -> DataLoader:
"""Retun DataLoader for train data."""
return DataLoader(
self.data_train,
shuffle=True,
batch_size=self.batch_size,
num_workers=self.num_workers,
pin_memory=self.pin_memory,
)
def val_dataloader(self) -> DataLoader:
"""Return DataLoader for val data."""
return DataLoader(
self.data_val,
shuffle=False,
batch_size=self.batch_size,
num_workers=self.num_workers,
pin_memory=self.pin_memory,
)
def test_dataloader(self) -> DataLoader:
"""Return DataLoader for val data."""
return DataLoader(
self.data_test,
shuffle=False,
batch_size=self.batch_size,
num_workers=self.num_workers,
pin_memory=self.pin_memory,
)
|
