Replace assert by ValueError of src/transformers/models/electra/modeling_{electra,tf_electra}.py and all other models that had copies (#13955)

* Replace all assert by ValueError in src/transformers/models/electra

* Reformat with black to pass check_code_quality test

* Change some assert to ValueError of modeling_bert & modeling_tf_albert

* Change some assert in multiples models

* Change multiples models assertion to ValueError in order to validate
  check_code_style test and models template test.

* Black reformat

* Change some more asserts in multiples models

* Change assert to ValueError in modeling_layoutlm.py to fix copy error in code_style_check

* Add proper message to ValueError in modeling_tf_albert.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Simplify logic in models/bert/modeling_bert.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Add ValueError message to models/convbert/modeling_tf_convbert.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Add error message for ValueError to modeling_tf_electra.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Simplify logic in models/tapas/modeling_tapas.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Simplify logic in models/electra/modeling_electra.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Add ValueError message in src/transformers/models/bert/modeling_tf_bert.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Simplify logic in src/transformers/models/rembert/modeling_rembert.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Simplify logic in src/transformers/models/albert/modeling_albert.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

src/transformers/models/albert/modeling_albert.py

@@ -186,9 +186,8 @@ def load_tf_weights_in_albert(model, config, tf_checkpoint_path):
         elif m_name == "kernel":
             array = np.transpose(array)
         try:
-            assert (
-                pointer.shape == array.shape
-            ), f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched"
+            if pointer.shape != array.shape:
+                raise ValueError(f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched")
         except AssertionError as e:
             e.args += (pointer.shape, array.shape)
             raise

src/transformers/models/albert/modeling_tf_albert.py

@@ -165,7 +165,8 @@ class TFAlbertEmbeddings(tf.keras.layers.Layer):
         Returns:
             final_embeddings (:obj:`tf.Tensor`): output embedding tensor.
         """
-        assert not (input_ids is None and inputs_embeds is None)
+        if input_ids is None and inputs_embeds is None:
+            raise ValueError("Need to provide either `input_ids` or `input_embeds`.")
 
         if input_ids is not None:
             inputs_embeds = tf.gather(params=self.weight, indices=input_ids)

src/transformers/models/bert/modeling_bert.py

@@ -153,9 +153,8 @@ def load_tf_weights_in_bert(model, config, tf_checkpoint_path):
         elif m_name == "kernel":
             array = np.transpose(array)
         try:
-            assert (
-                pointer.shape == array.shape
-            ), f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched"
+            if pointer.shape != array.shape:
+                raise ValueError(f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched")
         except AssertionError as e:
             e.args += (pointer.shape, array.shape)
             raise
@@ -450,7 +449,8 @@ class BertLayer(nn.Module):
         self.is_decoder = config.is_decoder
         self.add_cross_attention = config.add_cross_attention
         if self.add_cross_attention:
-            assert self.is_decoder, f"{self} should be used as a decoder model if cross attention is added"
+            if not self.is_decoder:
+                raise ValueError(f"{self} should be used as a decoder model if cross attention is added")
             self.crossattention = BertAttention(config)
         self.intermediate = BertIntermediate(config)
         self.output = BertOutput(config)
@@ -485,9 +485,10 @@ class BertLayer(nn.Module):
 
         cross_attn_present_key_value = None
         if self.is_decoder and encoder_hidden_states is not None:
-            assert hasattr(
-                self, "crossattention"
-            ), f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+            if not hasattr(self, "crossattention"):
+                raise ValueError(
+                    f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+                )
 
             # cross_attn cached key/values tuple is at positions 3,4 of past_key_value tuple
             cross_attn_past_key_value = past_key_value[-2:] if past_key_value is not None else None

src/transformers/models/bert/modeling_tf_bert.py

@@ -182,7 +182,8 @@ class TFBertEmbeddings(tf.keras.layers.Layer):
         Returns:
             final_embeddings (:obj:`tf.Tensor`): output embedding tensor.
         """
-        assert not (input_ids is None and inputs_embeds is None)
+        if input_ids is None and inputs_embeds is None:
+            raise ValueError("Need to provide either `input_ids` or `input_embeds`.")
 
         if input_ids is not None:
             inputs_embeds = tf.gather(params=self.weight, indices=input_ids)

src/transformers/models/convbert/modeling_tf_convbert.py

@@ -118,7 +118,8 @@ class TFConvBertEmbeddings(tf.keras.layers.Layer):
         Returns:
             final_embeddings (:obj:`tf.Tensor`): output embedding tensor.
         """
-        assert not (input_ids is None and inputs_embeds is None)
+        if input_ids is None and inputs_embeds is None:
+            raise ValueError("Need to provide either `input_ids` or `input_embeds`.")
 
         if input_ids is not None:
             inputs_embeds = tf.gather(params=self.weight, indices=input_ids)

src/transformers/models/electra/modeling_electra.py

@@ -139,9 +139,8 @@ def load_tf_weights_in_electra(model, config, tf_checkpoint_path, discriminator_
             elif m_name == "kernel":
                 array = np.transpose(array)
             try:
-                assert (
-                    pointer.shape == array.shape
-                ), f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched"
+                if pointer.shape != array.shape:
+                    raise ValueError(f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched")
             except AssertionError as e:
                 e.args += (pointer.shape, array.shape)
                 raise
@@ -447,7 +446,8 @@ class ElectraLayer(nn.Module):
         self.is_decoder = config.is_decoder
         self.add_cross_attention = config.add_cross_attention
         if self.add_cross_attention:
-            assert self.is_decoder, f"{self} should be used as a decoder model if cross attention is added"
+            if not self.is_decoder:
+                raise ValueError(f"{self} should be used as a decoder model if cross attention is added")
             self.crossattention = ElectraAttention(config)
         self.intermediate = ElectraIntermediate(config)
         self.output = ElectraOutput(config)
@@ -482,9 +482,10 @@ class ElectraLayer(nn.Module):
 
         cross_attn_present_key_value = None
         if self.is_decoder and encoder_hidden_states is not None:
-            assert hasattr(
-                self, "crossattention"
-            ), f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+            if not hasattr(self, "crossattention"):
+                raise ValueError(
+                    f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+                )
 
             # cross_attn cached key/values tuple is at positions 3,4 of past_key_value tuple
             cross_attn_past_key_value = past_key_value[-2:] if past_key_value is not None else None

src/transformers/models/electra/modeling_tf_electra.py

@@ -404,7 +404,8 @@ class TFElectraEmbeddings(tf.keras.layers.Layer):
         Returns:
             final_embeddings (:obj:`tf.Tensor`): output embedding tensor.
         """
-        assert not (input_ids is None and inputs_embeds is None)
+        if input_ids is None and inputs_embeds is None:
+            raise ValueError("Need to provide either `input_ids` or `input_embeds`.")
 
         if input_ids is not None:
             inputs_embeds = tf.gather(params=self.weight, indices=input_ids)

src/transformers/models/layoutlm/modeling_layoutlm.py

@@ -362,7 +362,8 @@ class LayoutLMLayer(nn.Module):
         self.is_decoder = config.is_decoder
         self.add_cross_attention = config.add_cross_attention
         if self.add_cross_attention:
-            assert self.is_decoder, f"{self} should be used as a decoder model if cross attention is added"
+            if not self.is_decoder:
+                raise ValueError(f"{self} should be used as a decoder model if cross attention is added")
             self.crossattention = LayoutLMAttention(config)
         self.intermediate = LayoutLMIntermediate(config)
         self.output = LayoutLMOutput(config)
@@ -397,9 +398,10 @@ class LayoutLMLayer(nn.Module):
 
         cross_attn_present_key_value = None
         if self.is_decoder and encoder_hidden_states is not None:
-            assert hasattr(
-                self, "crossattention"
-            ), f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+            if not hasattr(self, "crossattention"):
+                raise ValueError(
+                    f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+                )
 
             # cross_attn cached key/values tuple is at positions 3,4 of past_key_value tuple
             cross_attn_past_key_value = past_key_value[-2:] if past_key_value is not None else None

src/transformers/models/rembert/modeling_rembert.py

@@ -135,9 +135,8 @@ def load_tf_weights_in_rembert(model, config, tf_checkpoint_path):
         elif m_name == "kernel":
             array = np.transpose(array)
         try:
-            assert (
-                pointer.shape == array.shape
-            ), f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched"
+            if pointer.shape != array.shape:
+                raise ValueError(f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched")
         except AssertionError as e:
             e.args += (pointer.shape, array.shape)
             raise
@@ -420,7 +419,8 @@ class RemBertLayer(nn.Module):
         self.is_decoder = config.is_decoder
         self.add_cross_attention = config.add_cross_attention
         if self.add_cross_attention:
-            assert self.is_decoder, f"{self} should be used as a decoder model if cross attention is added"
+            if not self.is_decoder:
+                raise ValueError(f"{self} should be used as a decoder model if cross attention is added")
             self.crossattention = RemBertAttention(config)
         self.intermediate = RemBertIntermediate(config)
         self.output = RemBertOutput(config)
@@ -455,9 +455,10 @@ class RemBertLayer(nn.Module):
 
         cross_attn_present_key_value = None
         if self.is_decoder and encoder_hidden_states is not None:
-            assert hasattr(
-                self, "crossattention"
-            ), f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+            if not hasattr(self, "crossattention"):
+                raise ValueError(
+                    f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+                )
 
             # cross_attn cached key/values tuple is at positions 3,4 of past_key_value tuple
             cross_attn_past_key_value = past_key_value[-2:] if past_key_value is not None else None

src/transformers/models/roberta/modeling_roberta.py

@@ -389,7 +389,8 @@ class RobertaLayer(nn.Module):
         self.is_decoder = config.is_decoder
         self.add_cross_attention = config.add_cross_attention
         if self.add_cross_attention:
-            assert self.is_decoder, f"{self} should be used as a decoder model if cross attention is added"
+            if not self.is_decoder:
+                raise ValueError(f"{self} should be used as a decoder model if cross attention is added")
             self.crossattention = RobertaAttention(config)
         self.intermediate = RobertaIntermediate(config)
         self.output = RobertaOutput(config)
@@ -424,9 +425,10 @@ class RobertaLayer(nn.Module):
 
         cross_attn_present_key_value = None
         if self.is_decoder and encoder_hidden_states is not None:
-            assert hasattr(
-                self, "crossattention"
-            ), f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+            if not hasattr(self, "crossattention"):
+                raise ValueError(
+                    f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+                )
 
             # cross_attn cached key/values tuple is at positions 3,4 of past_key_value tuple
             cross_attn_past_key_value = past_key_value[-2:] if past_key_value is not None else None

src/transformers/models/roformer/modeling_roformer.py

@@ -167,9 +167,8 @@ def load_tf_weights_in_roformer(model, config, tf_checkpoint_path):
         elif m_name == "kernel":
             array = np.transpose(array)
         try:
-            assert (
-                pointer.shape == array.shape
-            ), f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched"
+            if not pointer.shape == array.shape:
+                raise ValueError(f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched")
         except AssertionError as e:
             e.args += (pointer.shape, array.shape)
             raise
@@ -463,7 +462,8 @@ class RoFormerLayer(nn.Module):
         self.is_decoder = config.is_decoder
         self.add_cross_attention = config.add_cross_attention
         if self.add_cross_attention:
-            assert self.is_decoder, f"{self} should be used as a decoder model if cross attention is added"
+            if not self.is_decoder:
+                raise ValueError(f"{self} should be used as a decoder model if cross attention is added")
             self.crossattention = RoFormerAttention(config)
         self.intermediate = RoFormerIntermediate(config)
         self.output = RoFormerOutput(config)

src/transformers/models/splinter/modeling_splinter.py

@@ -329,7 +329,8 @@ class SplinterLayer(nn.Module):
         self.is_decoder = config.is_decoder
         self.add_cross_attention = config.add_cross_attention
         if self.add_cross_attention:
-            assert self.is_decoder, f"{self} should be used as a decoder model if cross attention is added"
+            if not self.is_decoder:
+                raise ValueError(f"{self} should be used as a decoder model if cross attention is added")
             self.crossattention = SplinterAttention(config)
         self.intermediate = SplinterIntermediate(config)
         self.output = SplinterOutput(config)
@@ -364,9 +365,10 @@ class SplinterLayer(nn.Module):
 
         cross_attn_present_key_value = None
         if self.is_decoder and encoder_hidden_states is not None:
-            assert hasattr(
-                self, "crossattention"
-            ), f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+            if not hasattr(self, "crossattention"):
+                raise ValueError(
+                    f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+                )
 
             # cross_attn cached key/values tuple is at positions 3,4 of past_key_value tuple
             cross_attn_past_key_value = past_key_value[-2:] if past_key_value is not None else None

src/transformers/models/tapas/modeling_tapas.py

@@ -252,9 +252,8 @@ def load_tf_weights_in_tapas(model, config, tf_checkpoint_path):
         elif m_name == "kernel":
             array = np.transpose(array)
         try:
-            assert (
-                pointer.shape == array.shape
-            ), f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched"
+            if pointer.shape != array.shape:
+                raise ValueError(f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched")
         except AssertionError as e:
             e.args += (pointer.shape, array.shape)
             raise
@@ -548,7 +547,8 @@ class TapasLayer(nn.Module):
         self.is_decoder = config.is_decoder
         self.add_cross_attention = config.add_cross_attention
         if self.add_cross_attention:
-            assert self.is_decoder, f"{self} should be used as a decoder model if cross attention is added"
+            if not self.is_decoder:
+                raise ValueError(f"{self} should be used as a decoder model if cross attention is added")
             self.crossattention = TapasAttention(config)
         self.intermediate = TapasIntermediate(config)
         self.output = TapasOutput(config)
@@ -583,9 +583,10 @@ class TapasLayer(nn.Module):
 
         cross_attn_present_key_value = None
         if self.is_decoder and encoder_hidden_states is not None:
-            assert hasattr(
-                self, "crossattention"
-            ), f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+            if not hasattr(self, "crossattention"):
+                raise ValueError(
+                    f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`"
+                )
 
             # cross_attn cached key/values tuple is at positions 3,4 of past_key_value tuple
             cross_attn_past_key_value = past_key_value[-2:] if past_key_value is not None else None