diff --git a/pyproject.toml b/pyproject.toml
index b1899cd..ac68f64 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,5 +1,5 @@
 [build-system]
-requires = ["setuptools==71.0",
+requires = ["setuptools<=71.0",
             "setuptools_scm[toml]>=6.2",
             "wheel",
             "oldest-supported-numpy"]
diff --git a/spotiflow/model/pretrained.py b/spotiflow/model/pretrained.py
index 305c202..be9491a 100644
--- a/spotiflow/model/pretrained.py
+++ b/spotiflow/model/pretrained.py
@@ -1,3 +1,4 @@
+import os
 from dataclasses import dataclass
 from pathlib import Path
 from typing import Optional
@@ -6,7 +7,6 @@
 from ..utils.get_file import get_file
 
 
-
 @dataclass
 class RegisteredModel:
     """
@@ -25,8 +25,13 @@ def list_registered():
 
 
 def _default_cache_dir():
-    return Path("~").expanduser() / ".spotiflow" / "models"
-
+    default_cache_dir = os.getenv("SPOTIFLOW_CACHE_DIR", None)
+    if default_cache_dir is None:
+        return Path("~").expanduser() / ".spotiflow" / "models"
+    default_cache_dir = Path(default_cache_dir)
+    if default_cache_dir.stem != "models":
+        default_cache_dir = default_cache_dir / "models"
+    return default_cache_dir
 
 def get_pretrained_model_path(name: str, cache_dir: Optional[Path] = None) -> Path:
     """
diff --git a/spotiflow/model/spotiflow.py b/spotiflow/model/spotiflow.py
index 2ebd448..9d8ce11 100644
--- a/spotiflow/model/spotiflow.py
+++ b/spotiflow/model/spotiflow.py
@@ -193,6 +193,7 @@ def from_pretrained(
             inference_mode (bool, optional): whether to set the model in eval mode. Defaults to True.
             which (str, optional): which checkpoint to load. Defaults to "best".
             map_location (str, optional): device string to load the model to. Defaults to 'auto' (hardware-based).
+            cache_dir (Optional[Union[Path, str]], optional): directory to cache the model. Defaults to None. If None, will use the default cache directory (given by the env var SPOTIFLOW_CACHE_DIR if set, otherwise ~/.spotiflow).
 
         Returns:
             Self: loaded model
diff --git a/spotiflow/sample_data/datasets.py b/spotiflow/sample_data/datasets.py
index eba812e..731bd97 100644
--- a/spotiflow/sample_data/datasets.py
+++ b/spotiflow/sample_data/datasets.py
@@ -1,8 +1,9 @@
+import os
 from dataclasses import dataclass
 from pathlib import Path
 from typing import Optional, Union
 
-from ..utils import get_data, NotRegisteredError
+from ..utils import NotRegisteredError, get_data
 from ..utils.get_file import get_file
 
 
@@ -23,9 +24,14 @@ class RegisteredDataset:
 def list_registered():
     return list(_REGISTERED.keys())
 
-
 def _default_cache_dir():
-    return Path("~").expanduser() / ".spotiflow" / "datasets"
+    default_cache_dir = os.getenv("SPOTIFLOW_CACHE_DIR", None)
+    if default_cache_dir is None:
+        return Path("~").expanduser() / ".spotiflow" / "datasets"
+    default_cache_dir = Path(default_cache_dir)
+    if default_cache_dir.stem != "datasets":
+        default_cache_dir = default_cache_dir / "datasets"
+    return default_cache_dir
 
 
 def get_training_datasets_path(name: str, cache_dir: Optional[Path] = None) -> Path:
@@ -57,6 +63,7 @@ def load_dataset(name: str, include_test: bool=False, cache_dir: Optional[Union[
     Args:
         name (str): the name of the dataset to load.
         include_test (bool, optional): whether to include the test set in the returned data. Defaults to False.
+        cache_dir (Optional[Union[Path, str]], optional): directory to cache the model. Defaults to None. If None, will use the default cache directory (given by the env var SPOTIFLOW_CACHE_DIR if set, otherwise ~/.spotiflow).
     """
     if name not in _REGISTERED:
         raise NotRegisteredError(f"No training dataset named {name} found. Available datasets: {','.join(sorted(list_registered()))}")