diff --git a/setup.py b/setup.py
index 7d6161e567..11f12cd8bb 100644
--- a/setup.py
+++ b/setup.py
@@ -22,7 +22,7 @@
 
 setuptools.setup(
     name="TensorFlowASR",
-    version="0.7.7",
+    version="0.7.8",
     author="Huy Le Nguyen",
     author_email="nlhuy.cs.16@gmail.com",
     description="Almost State-of-the-art Automatic Speech Recognition using Tensorflow 2",
diff --git a/tensorflow_asr/losses/rnnt_losses.py b/tensorflow_asr/losses/rnnt_losses.py
index d6cbcb2b06..e8a2486a6e 100644
--- a/tensorflow_asr/losses/rnnt_losses.py
+++ b/tensorflow_asr/losses/rnnt_losses.py
@@ -235,7 +235,6 @@ def rnnt_loss_tf(logits, labels, label_length, logit_length, name=None):
     name = "rnnt_loss" if name is None else name
     with tf.name_scope(name):
         logits = tf.convert_to_tensor(logits, name="logits")
-        logits = tf.nn.log_softmax(logits)
         labels = tf.convert_to_tensor(labels, name="labels")
         label_length = tf.convert_to_tensor(label_length, name="label_length")
         logit_length = tf.convert_to_tensor(logit_length, name="logit_length")