diff --git a/dan/decoder.py b/dan/decoder.py
index af5c28cbba2f3de2076cf5113c8d8a9277294c3b..69e372cf98cceb5738f081be0a97acf37b0f298c 100644
--- a/dan/decoder.py
+++ b/dan/decoder.py
@@ -305,14 +305,9 @@ class FeaturesUpdater(Module):
         self.pe_2d = PositionalEncoding2D(
             params["enc_dim"], params["h_max"], params["w_max"], params["device"]
         )
-        self.use_2d_positional_encoding = (
-            "use_2d_pe" not in params or params["use_2d_pe"]
-        )
 
     def get_pos_features(self, features):
-        if self.use_2d_positional_encoding:
-            return self.pe_2d(features)
-        return features
+        return self.pe_2d(features)
 
 
 class GlobalHTADecoder(Module):
@@ -326,7 +321,6 @@ class GlobalHTADecoder(Module):
         self.dec_att_win = (
             params["attention_win"] if params["attention_win"] is not None else 1
         )
-        self.use_1d_pe = "use_1d_pe" not in params or params["use_1d_pe"]
 
         self.features_updater = FeaturesUpdater(params)
         self.att_decoder = GlobalAttDecoder(params)
@@ -361,9 +355,7 @@ class GlobalHTADecoder(Module):
         pos_tokens = self.emb(tokens).permute(0, 2, 1)
 
         # Add 1D Positional Encoding
-        if self.use_1d_pe:
-            pos_tokens = self.pe_1d(pos_tokens, start=start)
-        pos_tokens = pos_tokens.permute(2, 0, 1)
+        pos_tokens = self.pe_1d(pos_tokens, start=start).permute(2, 0, 1)
 
         if num_pred is None:
             num_pred = tokens.size(1)
diff --git a/dan/ocr/document/train.py b/dan/ocr/document/train.py
index 6f5e17b36816c7f8dd22008de8c6941e4a7c2ae0..ad4643b92506fe65befca106430d36ea1bcfa146 100644
--- a/dan/ocr/document/train.py
+++ b/dan/ocr/document/train.py
@@ -151,8 +151,6 @@ def get_config():
             "dec_pred_dropout": 0.1,  # dropout rate before decision layer
             "dec_att_dropout": 0.1,  # dropout rate in multi head attention
             "dec_dim_feedforward": 256,  # number of dimension for feedforward layer in transformer decoder layers
-            "use_2d_pe": True,  # use 2D positional embedding
-            "use_1d_pe": True,  # use 1D positional embedding
             "attention_win": 100,  # length of attention window
             # Curriculum dropout
             "dropout_scheduler": {
diff --git a/docs/get_started/training.md b/docs/get_started/training.md
index bc0764c39816df87539d1d3617e9897f2beb2a2f..15106d40c17bd3f5c3639ef04323b9e759a2fe8d 100644
--- a/docs/get_started/training.md
+++ b/docs/get_started/training.md
@@ -58,7 +58,6 @@ parameters:
     l_max: int
     dec_pred_dropout: float
     attention_win: int
-    use_1d_pe: bool
     vocab_size: int
     h_max: int
     w_max: int
diff --git a/docs/usage/train/parameters.md b/docs/usage/train/parameters.md
index bc346d86d87edb8652ba2560f1652e64b5750d92..ac4026fe205029a84d239d98ac63aa3f2de4c8b0 100644
--- a/docs/usage/train/parameters.md
+++ b/docs/usage/train/parameters.md
@@ -136,8 +136,6 @@ For a detailed description of all augmentation transforms, see the [dedicated pa
 | `model_params.dec_pred_dropout`           | Dropout rate before decision layer.                                                  | `float`       | `0.1`                                                             |
 | `model_params.dec_att_dropout`            | Dropout rate in multi head attention.                                                | `float`       | `0.1`                                                             |
 | `model_params.dec_dim_feedforward`        | Number of dimensions for feedforward layer in transformer decoder layers.            | `int`         | `256`                                                             |
-| `model_params.use_2d_pe`                  | Whether to use 2D positional embedding.                                              | `bool`        | `True`                                                            |
-| `model_params.use_1d_pe`                  | Whether to use 1D positional embedding.                                              | `bool`        | `True`                                                            |
 | `model_params.attention_win`              | Length of attention window.                                                          | `int`         | `100`                                                             |
 | `model_params.dropout_scheduler.function` | Curriculum dropout scheduler.                                                        | custom class  | `exponential_dropout_scheduler`                                   |
 | `model_params.dropout_scheduler.T`        | Exponential factor.                                                                  | `float`       | `5e4`                                                             |
diff --git a/tests/conftest.py b/tests/conftest.py
index 136cead1d494741405ee1365ce96116638e6a021..e804cb36734d4d7eb8af90ea2057b8d14e308ff0 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -99,8 +99,6 @@ def training_config():
             "dec_pred_dropout": 0.1,  # dropout rate before decision layer
             "dec_att_dropout": 0.1,  # dropout rate in multi head attention
             "dec_dim_feedforward": 256,  # number of dimension for feedforward layer in transformer decoder layers
-            "use_2d_pe": True,  # use 2D positional embedding
-            "use_1d_pe": True,  # use 1D positional embedding
             "attention_win": 100,  # length of attention window
             # Curriculum dropout
             "dropout_scheduler": {
diff --git a/tests/data/prediction/parameters.yml b/tests/data/prediction/parameters.yml
index 469afc8d157dade2a44b43891083d62796d331f0..2bad8803509baf24851dbeb2cf192da1ed6ae800 100644
--- a/tests/data/prediction/parameters.yml
+++ b/tests/data/prediction/parameters.yml
@@ -10,7 +10,6 @@ parameters:
     l_max: 15000
     dec_pred_dropout: 0.1
     attention_win: 100
-    use_1d_pe: True
     vocab_size: 96
     h_max: 500
     w_max: 1000