Update ruff to 0.11.2 (#36962)

* update

* update

* update

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

tests/models/fuyu/test_image_processing_fuyu.py

@@ -42,9 +42,9 @@ class TestFuyuImageProcessor(unittest.TestCase):
         expected_num_patches = self.processor.get_num_patches(image_height=self.height, image_width=self.width)
 
         patches_final = self.processor.patchify_image(image=self.image_input)
-        assert (
-            patches_final.shape[1] == expected_num_patches
-        ), f"Expected {expected_num_patches} patches, got {patches_final.shape[1]}."
+        assert patches_final.shape[1] == expected_num_patches, (
+            f"Expected {expected_num_patches} patches, got {patches_final.shape[1]}."
+        )
 
     def test_scale_to_target_aspect_ratio(self):
         # (h:450, w:210) fitting (160, 320) -> (160, 210*160/450)

tests/models/gpt2/test_modeling_gpt2.py

@@ -431,9 +431,9 @@ class GPT2ModelTester:
         model.eval()
 
         # We want this for SDPA, eager works with a `None` attention mask
-        assert (
-            model.config._attn_implementation == "sdpa"
-        ), "This test assumes the model to have the SDPA implementation for its attention calculations."
+        assert model.config._attn_implementation == "sdpa", (
+            "This test assumes the model to have the SDPA implementation for its attention calculations."
+        )
 
         # Prepare cache and non_cache input, needs a full attention mask
         cached_len = input_ids.shape[-1] // 2

tests/models/gpt_neox/test_modeling_gpt_neox.py

@@ -222,9 +222,9 @@ class GPTNeoXModelTester:
         model.eval()
 
         # We want this for SDPA, eager works with a `None` attention mask
-        assert (
-            model.config._attn_implementation == "sdpa"
-        ), "This test assumes the model to have the SDPA implementation for its attention calculations."
+        assert model.config._attn_implementation == "sdpa", (
+            "This test assumes the model to have the SDPA implementation for its attention calculations."
+        )
 
         # Prepare cache and non_cache input, needs a full attention mask
         cached_len = input_ids.shape[-1] // 2

tests/models/mask2former/test_image_processing_mask2former.py

@@ -315,7 +315,7 @@ class Mask2FormerImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase
             inst2class = {}
             for label in class_labels:
                 instance_ids = np.unique(instance_seg[class_id_map == label])
-                inst2class.update({i: label for i in instance_ids})
+                inst2class.update(dict.fromkeys(instance_ids, label))
 
             return instance_seg, inst2class
 

tests/models/maskformer/test_image_processing_maskformer.py

@@ -269,7 +269,7 @@ class MaskFormerImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase)
             inst2class = {}
             for label in class_labels:
                 instance_ids = np.unique(instance_seg[class_id_map == label])
-                inst2class.update({i: label for i in instance_ids})
+                inst2class.update(dict.fromkeys(instance_ids, label))
 
             return instance_seg, inst2class
 

tests/pipelines/test_pipelines_automatic_speech_recognition.py

@@ -1458,9 +1458,9 @@ class AutomaticSpeechRecognitionPipelineTests(unittest.TestCase):
 
         chunked_output = speech_recognizer(inputs.copy(), chunk_length_s=30)
         non_chunked_output = speech_recognizer(inputs.copy())
-        assert (
-            chunked_output.keys() == non_chunked_output.keys()
-        ), "The output structure should be the same for chunked vs non-chunked versions of asr pipelines."
+        assert chunked_output.keys() == non_chunked_output.keys(), (
+            "The output structure should be the same for chunked vs non-chunked versions of asr pipelines."
+        )
 
     @require_torch
     def test_return_timestamps_ctc_fast(self):

tests/sagemaker/scripts/pytorch/run_glue_model_parallelism.py

@@ -145,9 +145,9 @@ class DataTrainingArguments:
             train_extension = self.train_file.split(".")[-1]
             assert train_extension in ["csv", "json"], "`train_file` should be a csv or a json file."
             validation_extension = self.validation_file.split(".")[-1]
-            assert (
-                validation_extension == train_extension
-            ), "`validation_file` should have the same extension (csv or json) as `train_file`."
+            assert validation_extension == train_extension, (
+                "`validation_file` should have the same extension (csv or json) as `train_file`."
+            )
 
 
 @dataclass
@@ -265,9 +265,9 @@ def main():
             if data_args.test_file is not None:
                 train_extension = data_args.train_file.split(".")[-1]
                 test_extension = data_args.test_file.split(".")[-1]
-                assert (
-                    test_extension == train_extension
-                ), "`test_file` should have the same extension (csv or json) as `train_file`."
+                assert test_extension == train_extension, (
+                    "`test_file` should have the same extension (csv or json) as `train_file`."
+                )
                 data_files["test"] = data_args.test_file
             else:
                 raise ValueError("Need either a GLUE task or a test file for `do_predict`.")

tests/test_modeling_common.py

@@ -3234,9 +3234,9 @@ class ModelTesterMixin:
         for model_class in self.all_model_classes:
             model = model_class(config)
             num_params = model.num_parameters()
-            assert (
-                num_params < 1000000
-            ), f"{model_class} is too big for the common tests ({num_params})! It should have 1M max."
+            assert num_params < 1000000, (
+                f"{model_class} is too big for the common tests ({num_params})! It should have 1M max."
+            )
 
     @require_flash_attn
     @require_torch_gpu

tests/trainer/test_trainer.py

@@ -3005,9 +3005,9 @@ class TrainerIntegrationTest(TestCasePlus, TrainerIntegrationCommon):
         )
         trainer.train()
         # Check that we have the last known step:
-        assert os.path.exists(
-            os.path.join(tmp_dir, f"checkpoint-{trainer.state.max_steps}")
-        ), f"Could not find checkpoint-{trainer.state.max_steps}"
+        assert os.path.exists(os.path.join(tmp_dir, f"checkpoint-{trainer.state.max_steps}")), (
+            f"Could not find checkpoint-{trainer.state.max_steps}"
+        )
         # And then check the last step
         assert os.path.exists(os.path.join(tmp_dir, "checkpoint-9")), "Could not find checkpoint-9"
 

tests/trainer/test_trainer_seq2seq.py

@@ -180,9 +180,9 @@ class Seq2seqTrainerTester(TestCasePlus):
         for num_return_sequences in range(3, 0, -1):
             gen_config.num_return_sequences = num_return_sequences
             metrics = trainer.evaluate(eval_dataset=prepared_dataset, generation_config=gen_config)
-            assert (
-                metrics["eval_samples"] == dataset_len * num_return_sequences
-            ), f"Got {metrics['eval_samples']}, expected: {dataset_len * num_return_sequences}"
+            assert metrics["eval_samples"] == dataset_len * num_return_sequences, (
+                f"Got {metrics['eval_samples']}, expected: {dataset_len * num_return_sequences}"
+            )
 
     @require_torch
     def test_bad_generation_config_fail_early(self):