diff --git a/configs/architecture/rhythm_architecture.yaml b/configs/architecture/rhythm_architecture.yaml
index 95db5ed..72daa69 100644
--- a/configs/architecture/rhythm_architecture.yaml
+++ b/configs/architecture/rhythm_architecture.yaml
@@ -11,7 +11,7 @@ model:
 strategy: ${strategy}
 lr: ${lr}
 weight_decay: ${weight_decay}
-period: ${period}
-eta_min: ${eta_min}
+half_period: ${half_period}
+eta_min_rate: ${eta_min_rate}
 interval: step
 connected_dir: ${connected_dir}
diff --git a/configs/rhythm.yaml b/configs/rhythm.yaml
index 6808152..ac5f41e 100644
--- a/configs/rhythm.yaml
+++ b/configs/rhythm.yaml
@@ -32,8 +32,8 @@ direction: bi
 
 lr: 0.0001
 weight_decay: 0.01
-period: 2
-eta_min: 0.00001
+half_period: 1
+eta_min_rate: 0.1
 
 monitor: val_rmse_loss
 tracking_direction: min
diff --git a/configs/tuner/rhythm_tuner.yaml b/configs/tuner/rhythm_tuner.yaml
index 5ac22ef..fb2acb4 100644
--- a/configs/tuner/rhythm_tuner.yaml
+++ b/configs/tuner/rhythm_tuner.yaml
@@ -28,13 +28,13 @@ hparams:
     low: 0.001
     high: 0.01
     log: False
-  period:
+  half_period:
     low: 1
     high: 10
     log: False
-  eta_min:
-    low: 0.00005
-    high: 0.00001
+  eta_min_rate:
+    low: 0.05
+    high: 0.5
     log: False
 
 module_params:
diff --git a/src/architectures/rhythm_architecture.py b/src/architectures/rhythm_architecture.py
index 6ae2636..87edd41 100644
--- a/src/architectures/rhythm_architecture.py
+++ b/src/architectures/rhythm_architecture.py
@@ -17,8 +17,8 @@ def __init__(
         strategy: str,
         lr: float,
         weight_decay: float,
-        period: int,
-        eta_min: float,
+        half_period: int,
+        eta_min_rate: float,
         interval: str,
         connected_dir: str,
     ) -> None:
@@ -27,8 +27,8 @@ def __init__(
         self.strategy = strategy
         self.lr = lr
         self.weight_decay = weight_decay
-        self.period = period
-        self.eta_min = eta_min
+        self.half_period = half_period
+        self.eta_min_rate = eta_min_rate
         self.interval = interval
         self.connected_dir = connected_dir
 
@@ -86,11 +86,12 @@ def configure_optimizers(self) -> Dict[str, Any]:
                 lr=self.lr,
                 weight_decay=self.weight_decay,
             )
-        t_max = self.period * self.trainer.num_training_batches
+        t_max = self.half_period * self.trainer.num_training_batches
+        eta_min = self.lr * self.eta_min_rate
         scheduler = optim.lr_scheduler.CosineAnnealingLR(
             optimizer=optimizer,
             T_max=t_max,
-            eta_min=self.eta_min,
+            eta_min=eta_min,
         )
         return {
             "optimizer": optimizer,
diff --git a/src/tuners/rhythm_tuner.py b/src/tuners/rhythm_tuner.py
index f488b25..d14d912 100644
--- a/src/tuners/rhythm_tuner.py
+++ b/src/tuners/rhythm_tuner.py
@@ -116,19 +116,19 @@ def optuna_objective(
                 high=self.hparams.weight_decay.high,
                 log=self.hparams.weight_decay.log,
             )
-        if self.hparams.period:
-            params["period"] = trial.suggest_int(
-                name="period",
-                low=self.hparams.period.low,
-                high=self.hparams.period.high,
-                log=self.hparams.period.log,
+        if self.hparams.half_period:
+            params["half_period"] = trial.suggest_int(
+                name="half_period",
+                low=self.hparams.half_period.low,
+                high=self.hparams.half_period.high,
+                log=self.hparams.half_period.log,
             )
-        if self.hparams.eta_min:
-            params["eta_min"] = trial.suggest_float(
-                name="eta_min",
-                low=self.hparams.eta_min.low,
-                high=self.hparams.eta_min.high,
-                log=self.hparams.eta_min.log,
+        if self.hparams.eta_min_rate:
+            params["eta_min_rate"] = trial.suggest_float(
+                name="eta_min_rate",
+                low=self.hparams.eta_min_rate.low,
+                high=self.hparams.eta_min_rate.high,
+                log=self.hparams.eta_min_rate.log,
             )
 
         model = CustomizedRhythmNet(
@@ -143,8 +143,8 @@ def optuna_objective(
             strategy=self.module_params.strategy,
             lr=params["lr"],
             weight_decay=params["weight_decay"],
-            period=params["period"],
-            eta_min=params["eta_min"],
+            half_period=params["half_period"],
+            eta_min_rate=params["eta_min_rate"],
             interval=self.module_params.interval,
             connected_dir=self.module_params.connected_dir,
         )