diff --git a/utils/batch_viewer.py b/utils/batch_viewer.py
index 04f9666c..04885251 100644
--- a/utils/batch_viewer.py
+++ b/utils/batch_viewer.py
@@ -18,6 +18,7 @@ def view_data(
     args,
     neox_args,
     batch_fn: callable = None,
+    save_path: str = None,
 ):
     # fake MPU setup (needed to init dataloader without actual GPUs or parallelism)
     mpu.mock_model_parallel()
@@ -37,12 +38,14 @@ def view_data(
         
         if args.mode == "save":
             # save full batches for each step in the range (WARNING: this may consume lots of storage!)
-            np.save(f"./dump_data/batch{i}_bs{neox_args.train_micro_batch_size_per_gpu}", batch)
+            filename = f"batch{i}_bs{neox_args.train_micro_batch_size_per_gpu}"
+            np.save(os.path.join(save_path, filename), batch)
         elif args.mode == "custom":
             # dump user_defined statistic to a jsonl file (save_fn must return a dict)
             log = batch_fn(batch, i)
 
-            with open("./dump_data/stats.jsonl", "w+") as f:
+            filename = "stats.jsonl"
+            with open(os.path.join(save_path, filename), "w+") as f:
                 f.write(json.dumps(log) + "\n")
         else:
             raise ValueError(f'mode={mode} not acceptable--please pass either "save" or "custom" !')
@@ -74,6 +77,12 @@ def view_data(
         choices=["save", "custom"],
         help="Choose mode: 'save' to log all batches, and 'custom' to use user-defined statistic"
     )
+    parser.add_argument(
+        "--save_path",
+        type=str,
+        default=0,
+        help="Save path for files"
+    )
     args = parser.parse_known_args()[0]
 
     # init neox args
@@ -86,10 +95,11 @@ def save_fn(batch: np.array, iteration: int):
         # define your own logic here
         return {"iteration": iteration, "text": None}
 
+    os.makedirs(args.save_path, exist_ok=True)
 
     view_data(
         args,
         neox_args,    
         batch_fn=save_fn,
+        save_path=args.save_path,
     )
-