diff --git a/.riahub/workflows/workflow.yaml b/.riahub/workflows/workflow.yaml
index 4c71765..8cb64a2 100644
--- a/.riahub/workflows/workflow.yaml
+++ b/.riahub/workflows/workflow.yaml
@@ -46,7 +46,6 @@ jobs:
 
       - name: 2. Train Model
         run: |
-          mkdir -p data/dataset
           PYTHONPATH=. python data/training/train.py
           echo "training model"
 
diff --git a/conf/app.yaml b/conf/app.yaml
index 9010e0c..3c5cb2c 100644
--- a/conf/app.yaml
+++ b/conf/app.yaml
@@ -15,7 +15,8 @@ training:
   batch_size: 64
   epochs: 50
   learning_rate: 0.001
-  checkpoint_path: checkpoints/inference_recognition_model.ckpt
+  checkpoint_dir: checkpoints
+  checkpoint_filename: inference_recognition_model
   use_gpu: true
 
 inference:
diff --git a/data/training/train.py b/data/training/train.py
index 9184cd4..9943aba 100644
--- a/data/training/train.py
+++ b/data/training/train.py
@@ -33,7 +33,8 @@ def train_model():
     batch_size = 128
     epochs = 1
 
-    checkpoint_filename = f"{training_cfg.checkpoint_path}"
+    checkpoint_dir = training_cfg.checkpoint_dir
+    checkpoint_filename = training_cfg.checkpoint_filename
 
     train_data = (
         f"{dataset_cfg.output_dir}/train.h5"
@@ -130,6 +131,7 @@ def train_model():
     )
 
     checkpoint_callback = L.pytorch.callbacks.ModelCheckpoint(
+        dirpath=checkpoint_dir,
         filename=checkpoint_filename,
         save_top_k=True,
         verbose=True,
diff --git a/helpers/app_settings.py b/helpers/app_settings.py
index 4b593b0..ee52c47 100644
--- a/helpers/app_settings.py
+++ b/helpers/app_settings.py
@@ -26,7 +26,8 @@ class TrainingConfig:
     batch_size: int
     epochs: int
     learning_rate: float
-    checkpoint_path: str
+    checkpoint_dir: str
+    checkpoint_filename: str
     use_gpu: bool