diff --git a/README.md b/README.md
index 21689331..3ec05d12 100644
--- a/README.md
+++ b/README.md
@@ -56,6 +56,33 @@ These are simple examples. For more customization details, please refer to [Note
 
 ## Training
 
+### Dataset Format
+
+1. structure
+```
+{dataset1}
+├── images
+│   ├── test
+│   │   └── 1.jpg
+│   ├── train
+│   │   └── 2.jpg
+│   └── val
+│       └── 3.jpg
+└── labels
+    ├── test
+    │   └── 1.txt
+    ├── train
+    │   └── 2.txt
+    └── val
+        └── 3.txt
+```
+
+2. label format
+id is integer, coordinates are scaled to 0 ~ 1
+```
+id xmin ymin xmax ymax
+```
+
 To train YOLO on your machine/dataset:
 
 1. Modify the configuration file `yolo/config/dataset/**.yaml` to point to your dataset.
diff --git a/yolo/utils/logging_utils.py b/yolo/utils/logging_utils.py
index f60410d4..b9558a19 100644
--- a/yolo/utils/logging_utils.py
+++ b/yolo/utils/logging_utils.py
@@ -317,7 +317,6 @@ def log_model_structure(model: Union[ModuleList, YOLOLayer, YOLO]):
     console.print(table)
 
 
-@rank_zero_only
 def validate_log_directory(cfg: Config, exp_name: str) -> Path:
     base_path = Path(cfg.out_path, cfg.task.task)
     save_path = base_path / exp_name