diff --git a/examples/csmsc/vits/conf/default.yaml b/examples/csmsc/vits/conf/default.yaml
index a897569a..7e9e9c1d 100644
--- a/examples/csmsc/vits/conf/default.yaml
+++ b/examples/csmsc/vits/conf/default.yaml
@@ -181,5 +181,5 @@ generator_first: False # whether to start updating generator first
 num_snapshots: 10            # max number of snapshots to keep while training
 max_epoch: 1000              # Number of training epochs.
 save_interval_epochs: 1      # Interval epochs to save checkpoint.
-eval_interval_steps: 250     # Interval steps to evaluate the network.
+eval_interval_epochs: 1      # Interval steps to evaluate the network.
 seed: 777                    # random seed number
diff --git a/paddlespeech/t2s/exps/vits/train.py b/paddlespeech/t2s/exps/vits/train.py
index fefa6a01..cdfd3003 100644
--- a/paddlespeech/t2s/exps/vits/train.py
+++ b/paddlespeech/t2s/exps/vits/train.py
@@ -234,7 +234,7 @@ def train_sp(args, config):
 
     if dist.get_rank() == 0:
         trainer.extend(
-            evaluator, trigger=(config.eval_interval_steps, 'iteration'))
+            evaluator, trigger=(config.eval_interval_epochs, 'epoch'))
         trainer.extend(VisualDL(output_dir), trigger=(1, 'iteration'))
     trainer.extend(
         Snapshot(max_size=config.num_snapshots),