File size: 4,546 Bytes
093b0a5 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 | import numpy as np
import pandas as pd
from torch.utils.data import Dataset, DataLoader
from data_provider.data_loader import (
Dataset_Custom,
Dataset_Pred,
# Dataset_ETT_hour,
# Dataset_ETT_minute,
)
from utils.tools import dotdict
import pytorch_lightning as pl
class CustomDataModule(pl.LightningDataModule):
def __init__(self, config: dotdict, num_workers: int = 0):
super().__init__()
self.data_train: Dataset | None = None
self.data_val: Dataset | None = None
self.data_test: Dataset | None = None
self.config = config
# pl makes self.batch_size special
self.batch_size = config.batch_size
self.num_workers = num_workers
assert (
not config.inverse
) or config.scale, "Can't enable inverse without enabling scale"
def prepare_data(self):
"""Download data if needed. This method is called only from a single GPU.
Do not use it to assign state (self.x = y)."""
pass
def setup(self, stage: str | None = None):
"""Load data. Set variables: `self.data_train`, `self.data_val`, `self.data_test`.
This method is called by lightning twice for `trainer.fit()` and `trainer.test()`, so be careful if you do a random split!
The `stage` can be used to differentiate whether it's called before trainer.fit()` or `trainer.test()`.
"""
self.data_train = Dataset_Custom(self.config, flag="train")
self.data_val = Dataset_Custom(self.config, flag="val")
self.data_test = Dataset_Custom(self.config, flag="test")
# self.data_pred = Dataset_Pred(self.config, flag="pred")
print(
f"LOADED DATASETS for {stage}: train: {len(self.data_train)}\tval: {len(self.data_val)}\ttest: {len(self.data_test)}"
)
def train_dataloader(self):
return DataLoader(
self.data_train,
batch_size=self.batch_size,
shuffle=not self.config.dont_shuffle_train,
num_workers=self.num_workers,
drop_last=True,
)
def val_dataloader(self):
# assert self.batch_size <= len(
# self.data_val
# ), f"Batch size larger than val data set, batch size: {self.batch_size}, val size: {len(self.data_val)}"
return [
DataLoader(
self.data_val,
batch_size=self.batch_size,
shuffle=False,
drop_last=False,
num_workers=self.num_workers,
),
DataLoader(
self.data_test,
batch_size=self.batch_size,
shuffle=False,
drop_last=False,
num_workers=self.num_workers,
),
]
def test_dataloader(self):
return [
DataLoader(
self.data_train,
batch_size=self.config.batch_size,
shuffle=False,
drop_last=False,
num_workers=self.num_workers,
),
DataLoader(
self.data_val,
batch_size=self.config.batch_size,
shuffle=False,
drop_last=False,
num_workers=self.num_workers,
),
DataLoader(
self.data_test,
batch_size=self.config.batch_size,
shuffle=False,
drop_last=False,
num_workers=self.num_workers,
),
]
def predict_dataloader(self):
return (
DataLoader(
self.data_train,
batch_size=self.config.batch_size,
shuffle=False,
drop_last=False,
num_workers=self.num_workers,
),
DataLoader(
self.data_val,
batch_size=self.config.batch_size,
shuffle=False,
drop_last=False,
num_workers=self.num_workers,
),
DataLoader(
self.data_test,
batch_size=self.config.batch_size,
shuffle=False,
drop_last=False,
num_workers=self.num_workers,
),
# DataLoader(
# self.data_pred,
# batch_size=self.config.batch_size,
# shuffle=False,
# drop_last=False,
# num_workers=self.num_workers,
# ),
)
|