diff --git a/hyperswap/src/dataset.py b/hyperswap/src/dataset.py
index 77df6cc..3d9bdb4 100644
--- a/hyperswap/src/dataset.py
+++ b/hyperswap/src/dataset.py
@@ -1,4 +1,3 @@
-import glob
 import os
 import random
 from configparser import ConfigParser
@@ -9,7 +8,7 @@ from torch import Tensor
 from torch.utils.data import Dataset
 from torchvision import io, transforms
 
-from .helper import convert_tensor
+from .helper import convert_tensor, resolve_static_file_pattern
 from .types import Batch, BatchMode, ConvertTemplate, UsageMode
 
 
@@ -23,10 +22,9 @@ class DynamicDataset(Dataset[Tensor]):
 		self.config_batch_ratio = config_parser.getfloat('training.dataset.current', 'batch_ratio')
 		self.config_parser = config_parser
 		self.transforms = self.compose_transforms()
-		self.file_paths = glob.glob(self.config_file_pattern)
 
 	def __getitem__(self, index : int) -> Batch:
-		file_path = self.file_paths[index]
+		file_path = resolve_static_file_pattern(self.config_file_pattern)[index]
 
 		if random.random() < self.config_batch_ratio:
 			if self.config_batch_mode == 'equal':
@@ -43,7 +41,7 @@ class DynamicDataset(Dataset[Tensor]):
 		return self.prepare_different_batch(file_path)
 
 	def __len__(self) -> int:
-		return len(self.file_paths)
+		return len(resolve_static_file_pattern(self.config_file_pattern))
 
 	def compose_transforms(self) -> transforms:
 		return transforms.Compose(
@@ -79,7 +77,7 @@ class DynamicDataset(Dataset[Tensor]):
 		config_file_pattern = config_parser.get(config_section, 'file_pattern')
 		config_convert_template = cast(ConvertTemplate, config_parser.get(config_section, 'convert_template'))
 
-		target_path = random.choice(glob.glob(config_file_pattern))
+		target_path = random.choice(resolve_static_file_pattern(config_file_pattern))
 		source_tensor = io.read_image(source_path)
 		source_tensor = self.transforms(source_tensor)
 		source_tensor = self.conditional_convert_tensor(source_tensor, self.config_convert_template)
@@ -94,7 +92,7 @@ class DynamicDataset(Dataset[Tensor]):
 		config_file_pattern = config_parser.get(config_section, 'file_pattern')
 		config_convert_template = cast(ConvertTemplate, config_parser.get(config_section, 'convert_template'))
 
-		source_path = random.choice(glob.glob(config_file_pattern))
+		source_path = random.choice(resolve_static_file_pattern(config_file_pattern))
 		source_tensor = io.read_image(source_path)
 		source_tensor = self.transforms(source_tensor)
 		source_tensor = self.conditional_convert_tensor(source_tensor, config_convert_template)
@@ -104,7 +102,7 @@ class DynamicDataset(Dataset[Tensor]):
 		return source_tensor, target_tensor
 
 	def prepare_different_batch(self, source_path : str) -> Batch:
-		target_path = random.choice(self.file_paths)
+		target_path = random.choice(resolve_static_file_pattern(self.config_file_pattern))
 		source_tensor = io.read_image(source_path)
 		source_tensor = self.transforms(source_tensor)
 		source_tensor = self.conditional_convert_tensor(source_tensor, self.config_convert_template)
diff --git a/hyperswap/src/helper.py b/hyperswap/src/helper.py
index a14ea72..7d99e03 100644
--- a/hyperswap/src/helper.py
+++ b/hyperswap/src/helper.py
@@ -1,3 +1,7 @@
+import glob
+from functools import lru_cache
+from typing import List
+
 import torch
 from torch import Tensor, nn
 
@@ -55,3 +59,8 @@ def apply_noise(input_tensor : Tensor, factor : float) -> Tensor:
 	noise_tensor = torch.randn_like(input_tensor) * factor
 	output_tensor = input_tensor + noise_tensor
 	return output_tensor
+
+
+@lru_cache(maxsize = None)
+def resolve_static_file_pattern(file_pattern : str) -> List[str]:
+	return sorted(glob.glob(file_pattern))