Change the way tensor is reshaped in BartAttention (from .view to .reshape)

src/transformers/models/bart/modeling_bart.py

@@ -229,8 +229,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py

@@ -1290,8 +1290,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/biogpt/modeling_biogpt.py

@@ -190,8 +190,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/blenderbot/modeling_blenderbot.py

@@ -216,8 +216,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/blenderbot_small/modeling_blenderbot_small.py

@@ -213,8 +213,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/data2vec/modeling_data2vec_audio.py

@@ -419,8 +419,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/gptsan_japanese/modeling_gptsan_japanese.py

@@ -451,8 +451,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/hubert/modeling_hubert.py

@@ -482,8 +482,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/m2m_100/modeling_m2m_100.py

@@ -284,8 +284,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/marian/modeling_marian.py

@@ -231,8 +231,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/mbart/modeling_mbart.py

@@ -225,8 +225,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/pegasus/modeling_pegasus.py

@@ -231,8 +231,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/pegasus_x/modeling_pegasus_x.py

@@ -240,8 +240,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/plbart/modeling_plbart.py

@@ -224,8 +224,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/sew/modeling_sew.py

@@ -482,8 +482,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/speech_to_text/modeling_speech_to_text.py

@@ -291,8 +291,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/speech_to_text_2/modeling_speech_to_text_2.py

@@ -237,8 +237,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/time_series_transformer/modeling_time_series_transformer.py

@@ -717,8 +717,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/unispeech/modeling_unispeech.py

@@ -517,8 +517,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/unispeech_sat/modeling_unispeech_sat.py

@@ -531,8 +531,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/wav2vec2/modeling_wav2vec2.py

@@ -574,8 +574,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

src/transformers/models/whisper/modeling_whisper.py

@@ -319,8 +319,8 @@ def forward(
 
         proj_shape = (bsz * self.num_heads, -1, self.head_dim)
         query_states = self._shape(query_states, tgt_len, bsz).view(*proj_shape)
-        key_states = key_states.view(*proj_shape)
-        value_states = value_states.view(*proj_shape)
+        key_states = key_states.reshape(*proj_shape)
+        value_states = value_states.reshape(*proj_shape)
 
         src_len = key_states.size(1)
         attn_weights = torch.bmm(query_states, key_states.transpose(1, 2))

tests/models/gptsan_japanese/test_modeling_gptsan_japanese.py

@@ -182,7 +182,10 @@ def test_model(self):
         self.model_tester.create_and_check_model(*config_and_inputs)
 
     @unittest.skip(
-        reason="skip for now as the computed `max_memory` by `model_split_percents` in the test method will be changed inside `from_pretrained`"
+        reason=(
+            "skip for now as the computed `max_memory` by `model_split_percents` in the test method will be changed"
+            " inside `from_pretrained`"
+        )
     )
     def test_model_parallelism(self):
         super().test_model_parallelism()

tests/test_pipeline_mixin.py

@@ -217,24 +217,24 @@ def run_pipeline_test(self, task, repo_name, model_architecture, tokenizer_name,
                 processor = processor_class.from_pretrained(repo_id)
             except Exception:
                 self.skipTest(
-                    f"{self.__class__.__name__}::test_pipeline_{task.replace('-', '_')} is skipped: Could not load the "
-                    f"processor from `{repo_id}` with `{processor_name}`."
+                    f"{self.__class__.__name__}::test_pipeline_{task.replace('-', '_')} is skipped: Could not load the"
+                    f" processor from `{repo_id}` with `{processor_name}`."
                 )
 
         # TODO: Maybe not upload such problematic tiny models to Hub.
         if tokenizer is None and processor is None:
             self.skipTest(
-                f"{self.__class__.__name__}::test_pipeline_{task.replace('-', '_')} is skipped: Could not find or load "
-                f"any tokenizer / processor from `{repo_id}`."
+                f"{self.__class__.__name__}::test_pipeline_{task.replace('-', '_')} is skipped: Could not find or load"
+                f" any tokenizer / processor from `{repo_id}`."
             )
 
-        # TODO: We should check if a model file is on the Hub repo. instead.
+        # TODO: We should check if a model file is on the Hub repo instead.
         try:
             model = model_architecture.from_pretrained(repo_id)
         except Exception:
             self.skipTest(
-                f"{self.__class__.__name__}::test_pipeline_{task.replace('-', '_')} is skipped: Could not find or load "
-                f"the model from `{repo_id}` with `{model_architecture}`."
+                f"{self.__class__.__name__}::test_pipeline_{task.replace('-', '_')} is skipped: Could not find or load"
+                f" the model from `{repo_id}` with `{model_architecture}`."
             )
 
         # validate