Refactor memory cleaning into a single function

train_db.py

@@ -1,7 +1,6 @@
 # DreamBooth training
 # XXX dropped option: fine_tune
 
-import gc
 import argparse
 import itertools
 import math
@@ -12,6 +11,7 @@ import toml
 from tqdm import tqdm
 import torch
 
+from library.device_utils import clean_memory
 from library.ipex_interop import init_ipex
 
 init_ipex()
@@ -138,9 +138,7 @@ def train(args):
         with torch.no_grad():
             train_dataset_group.cache_latents(vae, args.vae_batch_size, args.cache_latents_to_disk, accelerator.is_main_process)
         vae.to("cpu")
-        if torch.cuda.is_available():
-            torch.cuda.empty_cache()
-        gc.collect()
+        clean_memory()
 
         accelerator.wait_for_everyone()