diff --git a/src/axolotl/cli/__init__.py b/src/axolotl/cli/__init__.py
index 6506e44f7..d07b10ce3 100644
--- a/src/axolotl/cli/__init__.py
+++ b/src/axolotl/cli/__init__.py
@@ -476,7 +476,6 @@ def load_datasets(
         tokenizer,
         processor=processor,
     )
-    print(train_dataset, eval_dataset, total_num_steps)
 
     if (
         cli_args.debug
diff --git a/src/axolotl/evaluate.py b/src/axolotl/evaluate.py
index d53d7bb5d..7fd60cb5f 100644
--- a/src/axolotl/evaluate.py
+++ b/src/axolotl/evaluate.py
@@ -14,7 +14,7 @@
 from axolotl.train import TrainDatasetMeta
 from axolotl.utils.dict import DictDefault
 from axolotl.utils.models import load_model, load_processor, load_tokenizer
-from axolotl.utils.trainer import setup_trainer
+from axolotl.utils.trainer import set_pytorch_cuda_alloc_conf, setup_trainer
 
 project_root = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))
 src_dir = os.path.join(project_root, "src")
@@ -79,14 +79,8 @@ def evaluate(
         - Dictionary of evaluation metrics
     """
     # pylint: disable=duplicate-code
-    # Set up CUDA allocation config if using PyTorch >= 2.2
-    torch_version = torch.__version__.split(".")
-    torch_major, torch_minor = int(torch_version[0]), int(torch_version[1])
-    if torch_major == 2 and torch_minor >= 2:
-        if os.getenv("PYTORCH_CUDA_ALLOC_CONF") is None:
-            os.environ[
-                "PYTORCH_CUDA_ALLOC_CONF"
-            ] = "expandable_segments:True,roundup_power2_divisions:16"
+    # Enable expandable segments for cuda allocation to improve VRAM usage
+    set_pytorch_cuda_alloc_conf()
 
     # Load tokenizer
     LOG.debug(
diff --git a/src/axolotl/train.py b/src/axolotl/train.py
index c8576f1b4..851a71e54 100644
--- a/src/axolotl/train.py
+++ b/src/axolotl/train.py
@@ -24,7 +24,7 @@
 from axolotl.utils.dict import DictDefault
 from axolotl.utils.freeze import freeze_layers_except
 from axolotl.utils.models import load_model, load_processor, load_tokenizer
-from axolotl.utils.trainer import setup_trainer
+from axolotl.utils.trainer import set_pytorch_cuda_alloc_conf, setup_trainer
 
 try:
     from optimum.bettertransformer import BetterTransformer
@@ -53,25 +53,22 @@ class TrainDatasetMeta:
 def train(
     *, cfg: DictDefault, cli_args: TrainerCliArgs, dataset_meta: TrainDatasetMeta
 ) -> Tuple[Union[PeftModel, PreTrainedModel], PreTrainedTokenizer]:
-    # enable expandable segments for cuda allocation to improve VRAM usage
-    torch_version = torch.__version__.split(".")
-    torch_major, torch_minor = int(torch_version[0]), int(torch_version[1])
-    if torch_major == 2 and torch_minor >= 2:
-        if os.getenv("PYTORCH_CUDA_ALLOC_CONF") is None:
-            os.environ[
-                "PYTORCH_CUDA_ALLOC_CONF"
-            ] = "expandable_segments:True,roundup_power2_divisions:16"
-
-    # load the tokenizer first
+    # Enable expandable segments for cuda allocation to improve VRAM usage
+    set_pytorch_cuda_alloc_conf()
+
+    # Load tokenizer
     LOG.debug(
         f"loading tokenizer... {cfg.tokenizer_config or cfg.base_model_config}",
         main_process_only=True,
     )
     tokenizer = load_tokenizer(cfg)
+
+    # Load processor for multimodal models if needed
     processor = None
     if cfg.is_multimodal:
         processor = load_processor(cfg, tokenizer)
 
+    # Get datasets
     train_dataset = dataset_meta.train_dataset
     eval_dataset = dataset_meta.eval_dataset
     total_num_steps = dataset_meta.total_num_steps
diff --git a/src/axolotl/utils/trainer.py b/src/axolotl/utils/trainer.py
index 32e54c9a8..fd09b3eb6 100644
--- a/src/axolotl/utils/trainer.py
+++ b/src/axolotl/utils/trainer.py
@@ -512,6 +512,17 @@ def prepare_opinionated_env(cfg):
         os.environ["TOKENIZERS_PARALLELISM"] = "false"
 
 
+def set_pytorch_cuda_alloc_conf():
+    """Set up CUDA allocation config if using PyTorch >= 2.2"""
+    torch_version = torch.__version__.split(".")
+    torch_major, torch_minor = int(torch_version[0]), int(torch_version[1])
+    if torch_major == 2 and torch_minor >= 2:
+        if os.getenv("PYTORCH_CUDA_ALLOC_CONF") is None:
+            os.environ[
+                "PYTORCH_CUDA_ALLOC_CONF"
+            ] = "expandable_segments:True,roundup_power2_divisions:16"
+
+
 def setup_trainer(
     cfg, train_dataset, eval_dataset, model, tokenizer, processor, total_num_steps
 ):