diff --git a/deepmd/pt/train/training.py b/deepmd/pt/train/training.py
index 1cdc383d01..030e9ffdfb 100644
--- a/deepmd/pt/train/training.py
+++ b/deepmd/pt/train/training.py
@@ -122,8 +122,14 @@ def __init__(
         self.model_keys = (
             list(model_params["model_dict"]) if self.multi_task else ["Default"]
         )
-        self.rank = dist.get_rank() if dist.is_available() and dist.is_initialized() else 0
-        self.world_size = dist.get_world_size() if dist.is_available() and dist.is_initialized() else 1
+        self.rank = (
+            dist.get_rank() if dist.is_available() and dist.is_initialized() else 0
+        )
+        self.world_size = (
+            dist.get_world_size()
+            if dist.is_available() and dist.is_initialized()
+            else 1
+        )
         self.num_model = len(self.model_keys)
 
         # Iteration config
@@ -169,7 +175,9 @@ def get_dataloader_and_buffer(_data, _params):
                     _data,
                     sampler=_sampler,
                     batch_size=None,
-                    num_workers=NUM_WORKERS if dist.is_available() else 0,  # setting to 0 diverges the behavior of its iterator; should be >=1
+                    num_workers=NUM_WORKERS
+                    if dist.is_available()
+                    else 0,  # setting to 0 diverges the behavior of its iterator; should be >=1
                     drop_last=False,
                     pin_memory=True,
                 )
@@ -734,7 +742,11 @@ def step(_step_id, task_key="Default"):
             elif self.opt_type == "LKF":
                 if isinstance(self.loss, EnergyStdLoss):
                     KFOptWrapper = KFOptimizerWrapper(
-                        self.wrapper, self.optimizer, 24, 6, dist.is_available() and dist.is_initialized()
+                        self.wrapper,
+                        self.optimizer,
+                        24,
+                        6,
+                        dist.is_available() and dist.is_initialized(),
                     )
                     pref_e = self.opt_param["kf_start_pref_e"] * (
                         self.opt_param["kf_limit_pref_e"]
@@ -753,7 +765,9 @@ def step(_step_id, task_key="Default"):
                     # [coord, atype, natoms, mapping, shift, nlist, box]
                     model_pred = {"energy": p_energy, "force": p_force}
                     module = (
-                        self.wrapper.module if dist.is_available() and dist.is_initialized() else self.wrapper
+                        self.wrapper.module
+                        if dist.is_available() and dist.is_initialized()
+                        else self.wrapper
                     )
 
                     def fake_model():
@@ -768,10 +782,16 @@ def fake_model():
                     )
                 elif isinstance(self.loss, DenoiseLoss):
                     KFOptWrapper = KFOptimizerWrapper(
-                        self.wrapper, self.optimizer, 24, 6, dist.is_available() and dist.is_initialized()
+                        self.wrapper,
+                        self.optimizer,
+                        24,
+                        6,
+                        dist.is_available() and dist.is_initialized(),
                     )
                     module = (
-                        self.wrapper.module if dist.is_available() and dist.is_initialized() else self.wrapper
+                        self.wrapper.module
+                        if dist.is_available() and dist.is_initialized()
+                        else self.wrapper
                     )
                     model_pred = KFOptWrapper.update_denoise_coord(
                         input_dict,
@@ -924,7 +944,11 @@ def log_loss_valid(_task_key="Default"):
                 # Handle the case if rank 0 aborted and re-assigned
                 self.latest_model = Path(self.save_ckpt + f"-{_step_id + 1}.pt")
 
-                module = self.wrapper.module if dist.is_available() and dist.is_initialized() else self.wrapper
+                module = (
+                    self.wrapper.module
+                    if dist.is_available() and dist.is_initialized()
+                    else self.wrapper
+                )
                 self.save_model(self.latest_model, lr=cur_lr, step=_step_id)
                 log.info(f"Saved model to {self.latest_model}")
                 symlink_prefix_files(self.latest_model.stem, self.save_ckpt)
@@ -990,7 +1014,11 @@ def log_loss_valid(_task_key="Default"):
             prof.stop()
 
     def save_model(self, save_path, lr=0.0, step=0):
-        module = self.wrapper.module if dist.is_available() and dist.is_initialized() else self.wrapper
+        module = (
+            self.wrapper.module
+            if dist.is_available() and dist.is_initialized()
+            else self.wrapper
+        )
         module.train_infos["lr"] = lr
         module.train_infos["step"] = step
         torch.save(
diff --git a/deepmd/pt/utils/dataloader.py b/deepmd/pt/utils/dataloader.py
index 4705c6d0b4..8ebe75868e 100644
--- a/deepmd/pt/utils/dataloader.py
+++ b/deepmd/pt/utils/dataloader.py
@@ -97,7 +97,11 @@ def construct_dataset(system):
 
         with Pool(
             os.cpu_count()
-            // (int(os.environ["LOCAL_WORLD_SIZE"]) if dist.is_available() and dist.is_initialized() else 1)
+            // (
+                int(os.environ["LOCAL_WORLD_SIZE"])
+                if dist.is_available() and dist.is_initialized()
+                else 1
+            )
         ) as pool:
             self.systems = pool.map(construct_dataset, systems)
 
@@ -138,7 +142,8 @@ def construct_dataset(system):
                 num_workers=0,  # Should be 0 to avoid too many threads forked
                 sampler=system_sampler,
                 collate_fn=collate_batch,
-                shuffle=(not (dist.is_available() and dist.is_initialized())) and shuffle,
+                shuffle=(not (dist.is_available() and dist.is_initialized()))
+                and shuffle,
             )
             self.dataloaders.append(system_dataloader)
             self.index.append(len(system_dataloader))