fixed problem with apple silicon + option to save more frequently

src/cleaner/selfclean.py

@@ -1,5 +1,6 @@
 import gc
 import os
+import platform
 from enum import Enum
 from pathlib import Path
 from typing import Optional, Union
@@ -131,6 +132,7 @@ def run_on_image_folder(
         epochs: int = 100,
         batch_size: int = 64,
         ssl_pre_training: bool = True,
+        save_every_n_epochs: int = 10,
         work_dir: Optional[str] = None,
         num_workers: int = os.cpu_count(),
         pretraining_type: PretrainingType = PretrainingType.DINO,
@@ -152,6 +154,7 @@ def run_on_image_folder(
             epochs=epochs,
             batch_size=batch_size,
             ssl_pre_training=ssl_pre_training,
+            save_every_n_epochs=save_every_n_epochs,
             work_dir=work_dir,
             num_workers=num_workers,
             pretraining_type=pretraining_type,
@@ -171,6 +174,7 @@ def run_on_dataset(
         epochs: int = 100,
         batch_size: int = 64,
         ssl_pre_training: bool = True,
+        save_every_n_epochs: int = 10,
         work_dir: Optional[str] = None,
         num_workers: int = os.cpu_count(),
         pretraining_type: PretrainingType = PretrainingType.DINO,
@@ -188,6 +192,7 @@ def run_on_dataset(
             epochs=epochs,
             batch_size=batch_size,
             ssl_pre_training=ssl_pre_training,
+            save_every_n_epochs=save_every_n_epochs,
             work_dir=work_dir,
             num_workers=num_workers,
             pretraining_type=pretraining_type,
@@ -207,6 +212,7 @@ def _run(
         epochs: int = 100,
         batch_size: int = 64,
         ssl_pre_training: bool = True,
+        save_every_n_epochs: int = 10,
         work_dir: Optional[str] = None,
         num_workers: int = os.cpu_count(),
         pretraining_type: PretrainingType = PretrainingType.DINO,
@@ -226,6 +232,7 @@ def _run(
                     epochs=epochs,
                     batch_size=batch_size,
                     ssl_pre_training=ssl_pre_training,
+                    save_every_n_epochs=save_every_n_epochs,
                     work_dir=work_dir,
                     hyperparameters=hyperparameters,
                     num_workers=num_workers,
@@ -278,6 +285,7 @@ def train_dino(
         epochs: int = 100,
         batch_size: int = 64,
         ssl_pre_training: bool = True,
+        save_every_n_epochs: int = 10,
         work_dir: Optional[str] = None,
         hyperparameters: dict = DINO_STANDARD_HYPERPARAMETERS,
         num_workers: int = os.cpu_count(),
@@ -293,6 +301,7 @@ def train_dino(
         hyperparameters["epochs"] = epochs
         hyperparameters["batch_size"] = batch_size
         hyperparameters["ssl_pre_training"] = ssl_pre_training
+        hyperparameters["save_every_n_epochs"] = save_every_n_epochs
         if work_dir is not None:
             hyperparameters["work_dir"] = work_dir
 
@@ -306,10 +315,16 @@ def train_dino(
             kwargs = {"sampler": sampler}
         else:
             kwargs = {"shuffle": True}
+
+        # due to a problem with worker spawning on apple silicon
+        # we set it here to 0
+        kwargs["num_workers"] = num_workers
+        if platform.machine().lower() == "arm64":
+            kwargs["num_workers"] = 0
+
         train_loader = DataLoader(
             dataset,
             batch_size=batch_size,
-            num_workers=num_workers,
             drop_last=False,
             pin_memory=True,
             **kwargs,