Fix unnecessary super calls (#38897)

Signed-off-by: cyy <cyyever@outlook.com>

src/transformers/models/bit/modeling_bit.py

@@ -135,7 +135,7 @@ class BitGroupNormActivation(nn.GroupNorm):
     """
 
     def __init__(self, config, num_channels, eps=1e-5, affine=True, apply_activation=True):
-        super(BitGroupNormActivation, self).__init__(config.num_groups, num_channels, eps=eps, affine=affine)
+        super().__init__(config.num_groups, num_channels, eps=eps, affine=affine)
         if apply_activation:
             self.activation = ACT2FN[config.hidden_act]
         else:

src/transformers/models/blenderbot/modeling_blenderbot.py

@@ -1183,7 +1183,7 @@ class BlenderbotModel(BlenderbotPreTrainedModel):
             )
             return BlenderbotSmallModel.from_pretrained(pretrained_model_name_or_path)
 
-        return super(BlenderbotModel, cls).from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+        return super().from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
 
     def get_input_embeddings(self):
         return self.shared
@@ -1344,9 +1344,7 @@ class BlenderbotForConditionalGeneration(BlenderbotPreTrainedModel, GenerationMi
             )
             return BlenderbotSmallForConditionalGeneration.from_pretrained(pretrained_model_name_or_path)
 
-        return super(BlenderbotForConditionalGeneration, cls).from_pretrained(
+        return super().from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
-            pretrained_model_name_or_path, *model_args, **kwargs
-        )
 
     def get_encoder(self):
         return self.model.get_encoder()

src/transformers/models/bros/modeling_bros.py

@@ -74,7 +74,7 @@ class BrosPositionalEmbedding1D(nn.Module):
     # Reference: https://github.com/kimiyoung/transformer-xl/blob/master/pytorch/mem_transformer.py#L15
 
     def __init__(self, config):
-        super(BrosPositionalEmbedding1D, self).__init__()
+        super().__init__()
 
         self.dim_bbox_sinusoid_emb_1d = config.dim_bbox_sinusoid_emb_1d
 
@@ -93,7 +93,7 @@ class BrosPositionalEmbedding1D(nn.Module):
 
 class BrosPositionalEmbedding2D(nn.Module):
     def __init__(self, config):
-        super(BrosPositionalEmbedding2D, self).__init__()
+        super().__init__()
 
         self.dim_bbox = config.dim_bbox
         self.x_pos_emb = BrosPositionalEmbedding1D(config)
@@ -112,7 +112,7 @@ class BrosPositionalEmbedding2D(nn.Module):
 
 class BrosBboxEmbeddings(nn.Module):
     def __init__(self, config):
-        super(BrosBboxEmbeddings, self).__init__()
+        super().__init__()
         self.bbox_sinusoid_emb = BrosPositionalEmbedding2D(config)
         self.bbox_projection = nn.Linear(config.dim_bbox_sinusoid_emb_2d, config.dim_bbox_projection, bias=False)
 

src/transformers/models/data2vec/modeling_data2vec_audio.py

@@ -1229,7 +1229,7 @@ class Data2VecAudioForAudioFrameClassification(Data2VecAudioPreTrainedModel):
 
 class AMSoftmaxLoss(nn.Module):
     def __init__(self, input_dim, num_labels, scale=30.0, margin=0.4):
-        super(AMSoftmaxLoss, self).__init__()
+        super().__init__()
         self.scale = scale
         self.margin = margin
         self.num_labels = num_labels

src/transformers/models/deprecated/ernie_m/modeling_ernie_m.py

@@ -484,7 +484,7 @@ ERNIE_M_INPUTS_DOCSTRING = r"""
 )
 class ErnieMModel(ErnieMPreTrainedModel):
     def __init__(self, config, add_pooling_layer=True):
-        super(ErnieMModel, self).__init__(config)
+        super().__init__(config)
         self.initializer_range = config.initializer_range
         self.embeddings = ErnieMEmbeddings(config)
         self.encoder = ErnieMEncoder(config)
@@ -964,7 +964,7 @@ class ErnieMForQuestionAnswering(ErnieMPreTrainedModel):
 )
 class ErnieMForInformationExtraction(ErnieMPreTrainedModel):
     def __init__(self, config):
-        super(ErnieMForInformationExtraction, self).__init__(config)
+        super().__init__(config)
         self.ernie_m = ErnieMModel(config)
         self.linear_start = nn.Linear(config.hidden_size, 1)
         self.linear_end = nn.Linear(config.hidden_size, 1)

src/transformers/models/granitemoe/modeling_granitemoe.py

@@ -324,7 +324,7 @@ class GraniteMoeMoE(nn.Module):
     """
 
     def __init__(self, config: GraniteMoeConfig):
-        super(GraniteMoeMoE, self).__init__()
+        super().__init__()
 
         self.input_size = config.hidden_size
         self.hidden_size = config.intermediate_size

src/transformers/models/granitemoehybrid/modeling_granitemoehybrid.py

@@ -856,7 +856,7 @@ class GraniteMoeHybridMLP(nn.Module):
     """
 
     def __init__(self, config: GraniteMoeHybridConfig):
-        super(GraniteMoeHybridMLP, self).__init__()
+        super().__init__()
 
         self.input_size = config.hidden_size
         self.hidden_size = config.shared_intermediate_size
@@ -995,7 +995,7 @@ class GraniteMoeHybridMoE(nn.Module):
     """
 
     def __init__(self, config: GraniteMoeHybridConfig):
-        super(GraniteMoeHybridMoE, self).__init__()
+        super().__init__()
 
         self.input_size = config.hidden_size
         self.hidden_size = config.intermediate_size

src/transformers/models/granitemoeshared/modeling_granitemoeshared.py

@@ -56,7 +56,7 @@ class GraniteMoeSharedMLP(nn.Module):
     """
 
     def __init__(self, config: GraniteMoeSharedConfig):
-        super(GraniteMoeSharedMLP, self).__init__()
+        super().__init__()
 
         self.input_size = config.hidden_size
         self.hidden_size = config.shared_intermediate_size
@@ -195,7 +195,7 @@ class GraniteMoeSharedMoE(nn.Module):
     """
 
     def __init__(self, config: GraniteMoeSharedConfig):
-        super(GraniteMoeSharedMoE, self).__init__()
+        super().__init__()
 
         self.input_size = config.hidden_size
         self.hidden_size = config.intermediate_size

src/transformers/models/granitemoeshared/modular_granitemoeshared.py

@@ -43,7 +43,7 @@ class GraniteMoeSharedMLP(nn.Module):
     """
 
     def __init__(self, config: GraniteMoeSharedConfig):
-        super(GraniteMoeSharedMLP, self).__init__()
+        super().__init__()
 
         self.input_size = config.hidden_size
         self.hidden_size = config.shared_intermediate_size

src/transformers/models/jetmoe/modeling_jetmoe.py

@@ -233,7 +233,7 @@ class JetMoeMoE(nn.Module):
     """
 
     def __init__(self, config: JetMoeConfig):
-        super(JetMoeMoE, self).__init__()
+        super().__init__()
 
         self.input_size = config.hidden_size
         self.hidden_size = config.intermediate_size
@@ -291,7 +291,7 @@ class JetMoeMoA(nn.Module):
     """
 
     def __init__(self, config: JetMoeConfig):
-        super(JetMoeMoA, self).__init__()
+        super().__init__()
 
         self.num_experts = config.num_local_experts
         self.input_size = config.hidden_size

src/transformers/models/layoutlm/modeling_layoutlm.py

@@ -47,7 +47,7 @@ class LayoutLMEmbeddings(nn.Module):
     """Construct the embeddings from word, position and token_type embeddings."""
 
     def __init__(self, config):
-        super(LayoutLMEmbeddings, self).__init__()
+        super().__init__()
         self.word_embeddings = nn.Embedding(config.vocab_size, config.hidden_size, padding_idx=config.pad_token_id)
         self.position_embeddings = nn.Embedding(config.max_position_embeddings, config.hidden_size)
         self.x_position_embeddings = nn.Embedding(config.max_2d_position_embeddings, config.hidden_size)
@@ -635,7 +635,7 @@ class LayoutLMPreTrainedModel(PreTrainedModel):
 @auto_docstring
 class LayoutLMModel(LayoutLMPreTrainedModel):
     def __init__(self, config):
-        super(LayoutLMModel, self).__init__(config)
+        super().__init__(config)
         self.config = config
 
         self.embeddings = LayoutLMEmbeddings(config)

src/transformers/models/layoutlmv2/modeling_layoutlmv2.py

@@ -52,7 +52,7 @@ class LayoutLMv2Embeddings(nn.Module):
     """Construct the embeddings from word, position and token_type embeddings."""
 
     def __init__(self, config):
-        super(LayoutLMv2Embeddings, self).__init__()
+        super().__init__()
         self.word_embeddings = nn.Embedding(config.vocab_size, config.hidden_size, padding_idx=config.pad_token_id)
         self.position_embeddings = nn.Embedding(config.max_position_embeddings, config.hidden_size)
 

src/transformers/models/lxmert/modeling_lxmert.py

@@ -648,7 +648,7 @@ class LxmertEncoder(nn.Module):
 
 class LxmertPooler(nn.Module):
     def __init__(self, config):
-        super(LxmertPooler, self).__init__()
+        super().__init__()
         self.dense = nn.Linear(config.hidden_size, config.hidden_size)
         self.activation = nn.Tanh()
 
@@ -663,7 +663,7 @@ class LxmertPooler(nn.Module):
 
 class LxmertPredictionHeadTransform(nn.Module):
     def __init__(self, config):
-        super(LxmertPredictionHeadTransform, self).__init__()
+        super().__init__()
         self.dense = nn.Linear(config.hidden_size, config.hidden_size)
         self.transform_act_fn = ACT2FN[config.hidden_act]
         self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=1e-12)
@@ -677,7 +677,7 @@ class LxmertPredictionHeadTransform(nn.Module):
 
 class LxmertLMPredictionHead(nn.Module):
     def __init__(self, config, lxmert_model_embedding_weights):
-        super(LxmertLMPredictionHead, self).__init__()
+        super().__init__()
         self.transform = LxmertPredictionHeadTransform(config)
 
         # The output weights are the same as the input embeddings, but there is
@@ -744,7 +744,7 @@ class LxmertVisualObjHead(nn.Module):
 
 class LxmertPreTrainingHeads(nn.Module):
     def __init__(self, config, lxmert_model_embedding_weights):
-        super(LxmertPreTrainingHeads, self).__init__()
+        super().__init__()
         self.predictions = LxmertLMPredictionHead(config, lxmert_model_embedding_weights)
         self.seq_relationship = nn.Linear(config.hidden_size, 2)
 

src/transformers/models/markuplm/modeling_markuplm.py

@@ -52,7 +52,7 @@ class XPathEmbeddings(nn.Module):
     """
 
     def __init__(self, config):
-        super(XPathEmbeddings, self).__init__()
+        super().__init__()
         self.max_depth = config.max_depth
 
         self.xpath_unitseq2_embeddings = nn.Linear(config.xpath_unit_hidden_size * self.max_depth, config.hidden_size)
@@ -116,7 +116,7 @@ class MarkupLMEmbeddings(nn.Module):
     """Construct the embeddings from word, position and token_type embeddings."""
 
     def __init__(self, config):
-        super(MarkupLMEmbeddings, self).__init__()
+        super().__init__()
         self.config = config
         self.word_embeddings = nn.Embedding(config.vocab_size, config.hidden_size, padding_idx=config.pad_token_id)
         self.position_embeddings = nn.Embedding(config.max_position_embeddings, config.hidden_size)
@@ -724,9 +724,7 @@ class MarkupLMPreTrainedModel(PreTrainedModel):
 
     @classmethod
     def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.PathLike]], *model_args, **kwargs):
-        return super(MarkupLMPreTrainedModel, cls).from_pretrained(
+        return super().from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
-            pretrained_model_name_or_path, *model_args, **kwargs
-        )
 
 
 @auto_docstring

src/transformers/models/perceiver/modeling_perceiver.py

@@ -2533,7 +2533,7 @@ class Conv2dSamePadding(nn.Conv2d):
     """
 
     def __init__(self, *args, **kwargs):
-        super(Conv2dSamePadding, self).__init__(*args, **kwargs)
+        super().__init__(*args, **kwargs)
         self.zero_pad_2d = nn.ZeroPad2d(
             reduce(__add__, [(k // 2 + (k - 2 * (k // 2)) - 1, k // 2) for k in self.kernel_size[::-1]])
         )

src/transformers/models/speech_to_text/modeling_speech_to_text.py

@@ -77,7 +77,7 @@ class Conv1dSubsampler(nn.Module):
     """
 
     def __init__(self, config):
-        super(Conv1dSubsampler, self).__init__()
+        super().__init__()
         self.config = config
         self.num_layers = config.num_conv_layers
         self.in_channels = config.input_feat_per_channel * config.input_channels

src/transformers/models/swin/modeling_tf_swin.py

@@ -476,7 +476,7 @@ class TFSwinDropPath(keras.layers.Layer):
     """Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks)."""
 
     def __init__(self, drop_prob: Optional[float] = None, scale_by_keep: bool = True, **kwargs) -> None:
-        super(TFSwinDropPath, self).__init__(**kwargs)
+        super().__init__(**kwargs)
         self.drop_prob = drop_prob
         self.scale_by_keep = scale_by_keep
 

src/transformers/models/tapas/modeling_tf_tapas.py

@@ -1871,7 +1871,7 @@ class ProductIndexMap(IndexMap):
         if outer_index.batch_dims != inner_index.batch_dims:
             raise ValueError("outer_index.batch_dims and inner_index.batch_dims must be the same.")
 
-        super(ProductIndexMap, self).__init__(
+        super().__init__(
             indices=(
                 inner_index.indices
                 + outer_index.indices * tf.cast(inner_index.num_segments, inner_index.indices.dtype)

src/transformers/models/udop/modeling_udop.py

@@ -847,7 +847,7 @@ class UdopBlock(nn.Module):
 
 class UdopCellEmbeddings(nn.Module):
     def __init__(self, max_2d_position_embeddings=501, hidden_size=1024):
-        super(UdopCellEmbeddings, self).__init__()
+        super().__init__()
         self.max_2d_position_embeddings = max_2d_position_embeddings
 
         self.x_position_embeddings = nn.Embedding(max_2d_position_embeddings, hidden_size)
@@ -911,7 +911,7 @@ class RelativePositionBiasBase(nn.Module, ABC):
         prefix_bucket=False,
         expand=False,
     ):
-        super(RelativePositionBiasBase, self).__init__()
+        super().__init__()
         self.prefix_bucket = prefix_bucket
         self.augmentation = augmentation
         self.level = level
@@ -1499,7 +1499,7 @@ class UdopModel(UdopPreTrainedModel):
     ]
 
     def __init__(self, config):
-        super(UdopModel, self).__init__(config)
+        super().__init__(config)
 
         # text and image embeddings
         self.shared = nn.Embedding(config.vocab_size, config.d_model)
@@ -1695,7 +1695,7 @@ class UdopForConditionalGeneration(UdopPreTrainedModel, GenerationMixin):
     ]
 
     def __init__(self, config):
-        super(UdopForConditionalGeneration, self).__init__(config)
+        super().__init__(config)
 
         # text and image embeddings
         self.shared = nn.Embedding(config.vocab_size, config.d_model)

src/transformers/models/unispeech_sat/modeling_unispeech_sat.py

@@ -1670,7 +1670,7 @@ class UniSpeechSatForAudioFrameClassification(UniSpeechSatPreTrainedModel):
 
 class AMSoftmaxLoss(nn.Module):
     def __init__(self, input_dim, num_labels, scale=30.0, margin=0.4):
-        super(AMSoftmaxLoss, self).__init__()
+        super().__init__()
         self.scale = scale
         self.margin = margin
         self.num_labels = num_labels

src/transformers/models/wav2vec2/modeling_wav2vec2.py

@@ -2203,7 +2203,7 @@ class Wav2Vec2ForAudioFrameClassification(Wav2Vec2PreTrainedModel):
 
 class AMSoftmaxLoss(nn.Module):
     def __init__(self, input_dim, num_labels, scale=30.0, margin=0.4):
-        super(AMSoftmaxLoss, self).__init__()
+        super().__init__()
         self.scale = scale
         self.margin = margin
         self.num_labels = num_labels

src/transformers/models/wav2vec2_bert/modeling_wav2vec2_bert.py

@@ -1358,7 +1358,7 @@ class Wav2Vec2BertForAudioFrameClassification(Wav2Vec2BertPreTrainedModel):
 
 class AMSoftmaxLoss(nn.Module):
     def __init__(self, input_dim, num_labels, scale=30.0, margin=0.4):
-        super(AMSoftmaxLoss, self).__init__()
+        super().__init__()
         self.scale = scale
         self.margin = margin
         self.num_labels = num_labels

src/transformers/models/wav2vec2_conformer/modeling_wav2vec2_conformer.py

@@ -1751,7 +1751,7 @@ class Wav2Vec2ConformerForAudioFrameClassification(Wav2Vec2ConformerPreTrainedMo
 
 class AMSoftmaxLoss(nn.Module):
     def __init__(self, input_dim, num_labels, scale=30.0, margin=0.4):
-        super(AMSoftmaxLoss, self).__init__()
+        super().__init__()
         self.scale = scale
         self.margin = margin
         self.num_labels = num_labels

src/transformers/models/wavlm/modeling_wavlm.py

@@ -1514,7 +1514,7 @@ class WavLMForAudioFrameClassification(WavLMPreTrainedModel):
 
 class AMSoftmaxLoss(nn.Module):
     def __init__(self, input_dim, num_labels, scale=30.0, margin=0.4):
-        super(AMSoftmaxLoss, self).__init__()
+        super().__init__()
         self.scale = scale
         self.margin = margin
         self.num_labels = num_labels