[Wav2Vec2-Conf / LLaMA] Style fix (#26188)

* torch.nn -> nn

* fix llama

* copies

src/transformers/models/deprecated/open_llama/modeling_open_llama.py

@@ -99,7 +99,7 @@ class OpenLlamaRMSNorm(nn.Module):
 
 
 # Copied from transformers.models.llama.modeling_llama.LlamaRotaryEmbedding with Llama->OpenLlama
-class OpenLlamaRotaryEmbedding(torch.nn.Module):
+class OpenLlamaRotaryEmbedding(nn.Module):
     def __init__(self, dim, max_position_embeddings=2048, base=10000, device=None):
         super().__init__()
 

src/transformers/models/llama/modeling_llama.py

@@ -89,7 +89,7 @@ class LlamaRMSNorm(nn.Module):
         return self.weight * hidden_states.to(input_dtype)
 
 
-class LlamaRotaryEmbedding(torch.nn.Module):
+class LlamaRotaryEmbedding(nn.Module):
     def __init__(self, dim, max_position_embeddings=2048, base=10000, device=None):
         super().__init__()
 

src/transformers/models/wav2vec2_conformer/modeling_wav2vec2_conformer.py

@@ -584,7 +584,7 @@ class Wav2Vec2ConformerConvolutionModule(nn.Module):
         if (config.conv_depthwise_kernel_size - 1) % 2 == 1:
             raise ValueError("`config.conv_depthwise_kernel_size` should be a odd number for 'SAME' padding")
         self.layer_norm = nn.LayerNorm(config.hidden_size)
-        self.pointwise_conv1 = torch.nn.Conv1d(
+        self.pointwise_conv1 = nn.Conv1d(
             config.hidden_size,
             2 * config.hidden_size,
             kernel_size=1,
@@ -592,8 +592,8 @@ class Wav2Vec2ConformerConvolutionModule(nn.Module):
             padding=0,
             bias=False,
         )
-        self.glu = torch.nn.GLU(dim=1)
-        self.depthwise_conv = torch.nn.Conv1d(
+        self.glu = nn.GLU(dim=1)
+        self.depthwise_conv = nn.Conv1d(
             config.hidden_size,
             config.hidden_size,
             config.conv_depthwise_kernel_size,
@@ -602,9 +602,9 @@ class Wav2Vec2ConformerConvolutionModule(nn.Module):
             groups=config.hidden_size,
             bias=False,
         )
-        self.batch_norm = torch.nn.BatchNorm1d(config.hidden_size)
+        self.batch_norm = nn.BatchNorm1d(config.hidden_size)
         self.activation = ACT2FN[config.hidden_act]
-        self.pointwise_conv2 = torch.nn.Conv1d(
+        self.pointwise_conv2 = nn.Conv1d(
             config.hidden_size,
             config.hidden_size,
             kernel_size=1,
@@ -612,7 +612,7 @@ class Wav2Vec2ConformerConvolutionModule(nn.Module):
             padding=0,
             bias=False,
         )
-        self.dropout = torch.nn.Dropout(config.conformer_conv_dropout)
+        self.dropout = nn.Dropout(config.conformer_conv_dropout)
 
     def forward(self, hidden_states):
         hidden_states = self.layer_norm(hidden_states)
@@ -798,7 +798,7 @@ class Wav2Vec2ConformerEncoderLayer(nn.Module):
 
         # Self-Attention
         self.self_attn_layer_norm = nn.LayerNorm(embed_dim)
-        self.self_attn_dropout = torch.nn.Dropout(dropout)
+        self.self_attn_dropout = nn.Dropout(dropout)
         self.self_attn = Wav2Vec2ConformerSelfAttention(config)
 
         # Conformer Convolution