Initial commit

work/LAST_lightning_dataSets.py

+import torch
+import torch.nn
+import glob
+import numpy as np
+from torch.utils.data import Dataset
+import pytorch_lightning as pl
+
+class PhaseDataset(pl.LightningDataModule):
+    def __init__(self, input_folder, target_folder, transform=None):
+        super().__init__()
+        self.input_folder = input_folder
+        self.target_folder = target_folder
+        self.transform = transform
+
+    def setup(self, stage=None):
+        input_files = sorted(glob.glob(self.input_folder))
+        target_files = sorted(glob.glob(self.target_folder))
+
+# By sorting the files, we establish a fixed order for loading 
+# and processing the files, making it easier to match inputs and 
+# targets correctly.
+
+        self.input_data = []
+        self.target_data = []
+
+        for input_file in input_files:
+            tensor = torch.flatten(torch.from_numpy(np.load(input_file)))
+            self.input_data.append(tensor)
+
+# ".append()" adds the specified tensor object to the end of the list. 
+# In this case, it adds the processed tensor (representing a single 
+# input sample) to the self.input_data list.
+
+        for target_file in target_files:
+            tensor = torch.flatten(torch.from_numpy(np.load(target_file)))
+            self.target_data.append(tensor)
+
+    def __len__(self):
+        return len(self.input_data)
+
+    def __getitem__(self, idx):
+        input_sample = self.input_data[idx]
+        target_sample = self.target_data[idx]
+
+        if self.transform:
+            target_sample = self.transform(target_sample)
+
+        return input_sample.float(), target_sample.float()
+#------------------------------------------------------------------------------------
+
+class PhaseDataset2(pl.LightningDataModule):
+    def __init__(self, input_folder, target_folder, transform=None):
+        super().__init__()
+        self.input_folder = input_folder
+        self.target_folder = target_folder
+        self.transform = transform
+
+    def setup(self, stage=None):
+        input_files = sorted(glob.glob(self.input_folder))
+        target_files = sorted(glob.glob(self.target_folder))
+
+        self.input_data = []
+        self.target_data = []
+
+        for input_file in input_files:
+            tensor = torch.from_numpy(np.load(input_file))
+
+            self.input_data.append(tensor)
+
+        for target_file in target_files:
+            tensor = torch.flatten(torch.from_numpy(np.load(target_file)))
+            self.target_data.append(tensor)
+
+    def __len__(self):
+        return len(self.input_data)
+
+    def __getitem__(self, idx):
+        input_sample = self.input_data[idx]
+        target_sample = self.target_data[idx]
+
+        if self.transform:
+            target_sample = self.transform(target_sample)
+
+        return input_sample.float(), target_sample.float()
+#------------------------------------------------------------------------------------
+def sinus(x):
+    y = torch.sin(x)
+    return y
+
+def arctan(x):
+    y = torch.atan(x)
+    return y
+
+#------------------------------------------------------------------------------
+"""
+control the LAST_lightning_dataSets.py / Yarkin
+"""
+dataset = PhaseDataset("C:\\Users\\yrknc\\OneDrive\\Masaüstü\\Input/*.npy", "C:\\Users\\yrknc\\OneDrive\\Masaüstü\\Target/*.npy", transform=None)
+
+# Call the setup method to load the data
+dataset.setup()
+
+# Access individual samples
+input_sample, target_sample = dataset[0]
+print(input_sample, "\nthis is length of input_sample in lightning:  ", len(input_sample))
+print(target_sample, "\nthis is length of target_sample in lightning:  ", len(target_sample))
\ No newline at end of file