diff --git a/paddleformers/transformers/ernie4_5/modeling.py b/paddleformers/transformers/ernie4_5/modeling.py
index 8c9c4b831bb..3c66681892b 100644
--- a/paddleformers/transformers/ernie4_5/modeling.py
+++ b/paddleformers/transformers/ernie4_5/modeling.py
@@ -984,7 +984,7 @@ def forward(
             )
 
         # if labels is None，means we need full output, instead of tensor_parallel_output
-        # tensor_parallel_output is togather with ParallelCrossEntropy
+        # tensor_parallel_output is together with ParallelCrossEntropy
         logits = self.lm_head(hidden_states)
 
         if return_dict:  # aka Generate Decoding
diff --git a/paddleformers/transformers/model_utils.py b/paddleformers/transformers/model_utils.py
index d4f8817db88..2b9b17be496 100644
--- a/paddleformers/transformers/model_utils.py
+++ b/paddleformers/transformers/model_utils.py
@@ -2319,7 +2319,7 @@ def _fuse_or_split_keys(
                             base_model_prefix=prefix,
                         )
                     if post_quantize:
-                        # Split -> quantize(Not support mdoel save)
+                        # Split -> quantize(Not support model save)
                         state_dict = load_state_dict(
                             shard_file,
                             tp_actions if pre_tensor_parallel_split else None,
@@ -2334,7 +2334,7 @@ def _fuse_or_split_keys(
                             dtype,
                         )
                     else:
-                        # quantize -> split(Support mdoel save)
+                        # quantize -> split(Support model save)
                         state_dict = load_state_dict(
                             shard_file,
                             tp_actions if pre_tensor_parallel_split else None,
diff --git a/tests/nn/test_linear.py b/tests/nn/test_linear.py
index 52fcab82d1f..70ef7f6a5f0 100644
--- a/tests/nn/test_linear.py
+++ b/tests/nn/test_linear.py
@@ -67,7 +67,7 @@ def _test_create_parallel_linear(config):
 
     assert isinstance(linear, mpu.RowParallelLinear)
     config.sequence_parallel = False
-    # Test createing squence row parallel linear
+    # Test creating sequence row parallel linear
     linear = Linear.create(
         in_features=config.in_features,
         out_features=config.out_features,
diff --git a/tests/transformers/test_configuration_common.py b/tests/transformers/test_configuration_common.py
index b63860662d8..6b006a18cac 100644
--- a/tests/transformers/test_configuration_common.py
+++ b/tests/transformers/test_configuration_common.py
@@ -215,7 +215,7 @@ class ConfigTestUtils:
     def test_config_common_kwargs_is_complete(self):
         base_config = PretrainedConfig()
         missing_keys = [key for key in base_config.__dict__ if key not in config_common_kwargs]
-        # If this part of the test fails, you have arguments to addin config_common_kwargs above.
+        # If this part of the test fails, you have arguments to add in config_common_kwargs above.
         self.assertListEqual(
             missing_keys,
             [