Remove unnecessary list comprehension (#41305)

Remove unnecessary comprehension

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

examples/pytorch/contrastive-image-text/run_clip.py

@@ -387,7 +387,7 @@ def main():
         return
 
     # 6. Get the column names for input/target.
-    dataset_columns = dataset_name_mapping.get(data_args.dataset_name, None)
+    dataset_columns = dataset_name_mapping.get(data_args.dataset_name)
     if data_args.image_column is None:
         image_column = dataset_columns[0] if dataset_columns is not None else column_names[0]
     else:

examples/pytorch/question-answering/run_qa_beam_search_no_trainer.py

@@ -933,7 +933,7 @@ def main():
             all_end_top_index.append(accelerator.gather_for_metrics(end_top_index).cpu().numpy())
             all_cls_logits.append(accelerator.gather_for_metrics(cls_logits).cpu().numpy())
 
-    max_len = max([x.shape[1] for x in all_end_top_log_probs])  # Get the max_length of the tensor
+    max_len = max(x.shape[1] for x in all_end_top_log_probs)  # Get the max_length of the tensor
 
     # concatenate all numpy arrays collected above
     start_top_log_probs_concat = create_and_fill_np_array(all_start_top_log_probs, eval_dataset, max_len)
@@ -993,7 +993,7 @@ def main():
                 all_end_top_index.append(accelerator.gather_for_metrics(end_top_index).cpu().numpy())
                 all_cls_logits.append(accelerator.gather_for_metrics(cls_logits).cpu().numpy())
 
-        max_len = max([x.shape[1] for x in all_end_top_log_probs])  # Get the max_length of the tensor
+        max_len = max(x.shape[1] for x in all_end_top_log_probs)  # Get the max_length of the tensor
 
         # concatenate all numpy arrays collected above
         start_top_log_probs_concat = create_and_fill_np_array(all_start_top_log_probs, predict_dataset, max_len)

examples/pytorch/question-answering/run_seq2seq_qa.py

@@ -416,7 +416,7 @@ def main():
         return
 
     # Get the column names for input/target.
-    dataset_columns = question_answering_column_name_mapping.get(data_args.dataset_name, None)
+    dataset_columns = question_answering_column_name_mapping.get(data_args.dataset_name)
     if data_args.question_column is None:
         question_column = dataset_columns[0] if dataset_columns is not None else column_names[0]
     else:

examples/pytorch/summarization/run_summarization.py

@@ -531,7 +531,7 @@ def main():
         model.config.forced_bos_token_id = forced_bos_token_id
 
     # Get the column names for input/target.
-    dataset_columns = summarization_name_mapping.get(data_args.dataset_name, None)
+    dataset_columns = summarization_name_mapping.get(data_args.dataset_name)
     if data_args.text_column is None:
         text_column = dataset_columns[0] if dataset_columns is not None else column_names[0]
     else:

examples/pytorch/summarization/run_summarization_no_trainer.py

@@ -476,7 +476,7 @@ def main():
     column_names = raw_datasets["train"].column_names
 
     # Get the column names for input/target.
-    dataset_columns = summarization_name_mapping.get(args.dataset_name, None)
+    dataset_columns = summarization_name_mapping.get(args.dataset_name)
     if args.text_column is None:
         text_column = dataset_columns[0] if dataset_columns is not None else column_names[0]
     else:

src/transformers/models/deepseek_vl_hybrid/image_processing_deepseek_vl_hybrid_fast.py

@@ -65,7 +65,7 @@ class DeepseekVLHybridImageProcessorFast(BaseImageProcessorFast):
         if kwargs.get("image_mean") is None:
             background_color = (127, 127, 127)
         else:
-            background_color = tuple([int(x * 255) for x in kwargs.get("image_mean")])
+            background_color = tuple(int(x * 255) for x in kwargs.get("image_mean"))
         if kwargs.get("high_res_image_mean") is None:
             high_res_background_color = (127, 127, 127)
         else:

src/transformers/models/deepseek_vl_hybrid/modular_deepseek_vl_hybrid.py

@@ -764,7 +764,7 @@ class DeepseekVLHybridImageProcessorFast(DeepseekVLImageProcessorFast):
         if kwargs.get("image_mean") is None:
             background_color = (127, 127, 127)
         else:
-            background_color = tuple([int(x * 255) for x in kwargs.get("image_mean")])
+            background_color = tuple(int(x * 255) for x in kwargs.get("image_mean"))
         if kwargs.get("high_res_image_mean") is None:
             high_res_background_color = (127, 127, 127)
         else:

src/transformers/models/musicgen/modeling_musicgen.py

@@ -551,7 +551,7 @@ class MusicgenDecoder(MusicgenPreTrainedModel):
         past_key_values_length = past_key_values.get_seq_length() if past_key_values is not None else 0
 
         if inputs_embeds is None:
-            inputs_embeds = sum([self.embed_tokens[codebook](input[:, codebook]) for codebook in range(num_codebooks)])
+            inputs_embeds = sum(self.embed_tokens[codebook](input[:, codebook]) for codebook in range(num_codebooks))
 
         attention_mask = self._update_causal_mask(
             attention_mask,

src/transformers/models/oneformer/modeling_oneformer.py

@@ -718,7 +718,7 @@ class OneFormerLoss(nn.Module):
         """
         Computes the average number of target masks across the batch, for normalization purposes.
         """
-        num_masks = sum([len(classes) for classes in class_labels])
+        num_masks = sum(len(classes) for classes in class_labels)
         num_masks = torch.as_tensor([num_masks], dtype=torch.float, device=device)
         world_size = 1
         if is_accelerate_available():

src/transformers/models/ovis2/image_processing_ovis2.py

@@ -184,7 +184,7 @@ def get_min_tile_covering_grid(
     for tile_grid in candidate_tile_grids:
         tile_regions = split_image_into_grid(image_height, image_width, tile_grid)
         tile_covering_ratio = (
-            sum([compute_patch_covering_area(*region, target_patch_size) for region in tile_regions]) / image_area
+            sum(compute_patch_covering_area(*region, target_patch_size) for region in tile_regions) / image_area
         )
 
         evaluated_grids.append((tile_grid, tile_covering_ratio))

src/transformers/models/patchtsmixer/modeling_patchtsmixer.py

@@ -1542,7 +1542,7 @@ class PatchTSMixerForPrediction(PatchTSMixerPreTrainedModel):
                 "normal": NormalOutput,
                 "negative_binomial": NegativeBinomialOutput,
             }
-            output_class = distribution_output_map.get(config.distribution_output, None)
+            output_class = distribution_output_map.get(config.distribution_output)
             if output_class is not None:
                 self.distribution_output = output_class(dim=dim)
             else:

src/transformers/models/phi4_multimodal/image_processing_phi4_multimodal_fast.py

@@ -237,7 +237,7 @@ class Phi4MultimodalImageProcessorFast(BaseImageProcessorFast):
             images_tokens.append(num_img_tokens)
             image_sizes.append([height, width])
             max_crops = hd_image_reshape.size(0)
-        max_crops = max([img.size(0) for img in images_transformed])
+        max_crops = max(img.size(0) for img in images_transformed)
         images_transformed = [self.pad_to_max_num_crops(im, max_crops) for im in images_transformed]
         images_transformed = torch.stack(images_transformed, dim=0)
         masks_transformed = [self.pad_mask_to_max_num_crops(mask, max_crops) for mask in masks_transformed]

src/transformers/models/pop2piano/tokenization_pop2piano.py

@@ -265,7 +265,7 @@ class Pop2PianoTokenizer(PreTrainedTokenizer):
 
         current_idx = start_idx
         current_velocity = 0
-        note_onsets_ready = [None for i in range(sum([k.endswith("NOTE") for k in self.encoder]) + 1)]
+        note_onsets_ready = [None for i in range(sum(k.endswith("NOTE") for k in self.encoder) + 1)]
         notes = []
         for token_type, number in words:
             if token_type == "TOKEN_SPECIAL":

src/transformers/models/sam_hq/processing_samhq.py

@@ -171,7 +171,7 @@ class SamHQProcessor(ProcessorMixin):
         r"""
         The method pads the 2D points and labels to the maximum number of points in the batch.
         """
-        expected_nb_points = max([point.shape[0] for point in input_points])
+        expected_nb_points = max(point.shape[0] for point in input_points)
         processed_input_points = []
         for i, point in enumerate(input_points):
             if point.shape[0] != expected_nb_points:

tests/models/mask2former/test_image_processing_mask2former.py

@@ -552,8 +552,8 @@ class Mask2FormerImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase
                 fuse_targets = [1 for el in el_unfused if el["label_id"] == 1]
                 num_to_fuse = 0 if len(fuse_targets) == 0 else sum(fuse_targets) - 1
                 # Expected number of segments after fusing
-                expected_num_segments = max([el["id"] for el in el_unfused]) - num_to_fuse
+                expected_num_segments = max(el["id"] for el in el_unfused) - num_to_fuse
-                num_segments_fused = max([el["id"] for el in el_fused])
+                num_segments_fused = max(el["id"] for el in el_fused)
                 self.assertEqual(num_segments_fused, expected_num_segments)
 
     def test_slow_fast_equivalence(self):

tests/models/maskformer/test_image_processing_maskformer.py

@@ -540,8 +540,8 @@ class MaskFormerImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase)
                 fuse_targets = [1 for el in el_unfused if el["label_id"] == 1]
                 num_to_fuse = 0 if len(fuse_targets) == 0 else sum(fuse_targets) - 1
                 # Expected number of segments after fusing
-                expected_num_segments = max([el["id"] for el in el_unfused]) - num_to_fuse
+                expected_num_segments = max(el["id"] for el in el_unfused) - num_to_fuse
-                num_segments_fused = max([el["id"] for el in el_fused])
+                num_segments_fused = max(el["id"] for el in el_fused)
                 self.assertEqual(num_segments_fused, expected_num_segments)
 
     def test_slow_fast_equivalence(self):

tests/models/musicgen/test_modeling_musicgen.py

@@ -242,7 +242,7 @@ class MusicgenDecoderTest(ModelTesterMixin, GenerationTesterMixin, PipelineTeste
             input_ids = input_ids.reshape(-1, config.num_codebooks, input_ids.shape[-1])
 
             inputs["inputs_embeds"] = sum(
-                [embed_tokens[codebook](input_ids[:, codebook]) for codebook in range(config.num_codebooks)]
+                embed_tokens[codebook](input_ids[:, codebook]) for codebook in range(config.num_codebooks)
             )
 
             with torch.no_grad():

tests/models/musicgen_melody/test_modeling_musicgen_melody.py

@@ -251,7 +251,7 @@ class MusicgenMelodyDecoderTest(ModelTesterMixin, GenerationTesterMixin, unittes
             input_ids = input_ids.reshape(-1, config.num_codebooks, input_ids.shape[-1])
 
             inputs["inputs_embeds"] = sum(
-                [embed_tokens[codebook](input_ids[:, codebook]) for codebook in range(config.num_codebooks)]
+                embed_tokens[codebook](input_ids[:, codebook]) for codebook in range(config.num_codebooks)
             )
 
             with torch.no_grad():

tests/models/owlvit/test_processing_owlvit.py

@@ -187,7 +187,7 @@ class OwlViTProcessorTest(ProcessorTesterMixin, unittest.TestCase):
 
         seq_length = 16
         batch_size = len(input_texts)
-        num_max_text_queries = max([len(texts) for texts in input_texts])
+        num_max_text_queries = max(len(texts) for texts in input_texts)
 
         self.assertListEqual(list(inputs.keys()), ["input_ids", "attention_mask"])
         self.assertEqual(inputs["input_ids"].shape, (batch_size * num_max_text_queries, seq_length))

tests/utils/test_cache_utils.py

@@ -352,9 +352,7 @@ class CacheHardIntegrationTest(unittest.TestCase):
         decoded = tokenizer.batch_decode(gen_out.sequences, skip_special_tokens=True)
         # sum of the scores for the generated tokens
         input_length = inputs.input_ids.shape[1]
-        score_sum = sum(
+        score_sum = sum(score[0][gen_out.sequences[0][input_length + idx]] for idx, score in enumerate(gen_out.scores))
-            [score[0][gen_out.sequences[0][input_length + idx]] for idx, score in enumerate(gen_out.scores)]
-        )
 
         EXPECTED_GENERATION = (
             "Here's everything I know about cats. Cats are mammals, they have four legs, they have a tail, they have "

utils/check_modular_conversion.py

@@ -179,7 +179,7 @@ if __name__ == "__main__":
     # we start applying modular conversion to each list in parallel, starting from the first list
 
     console.print(f"[bold yellow]Number of dependency levels: {len(ordered_files)}[/bold yellow]")
-    console.print(f"[bold yellow]Files per level: {tuple([len(x) for x in ordered_files])}[/bold yellow]")
+    console.print(f"[bold yellow]Files per level: {tuple(len(x) for x in ordered_files)}[/bold yellow]")
 
     try:
         for dependency_level_files in ordered_files:

utils/tests_fetcher.py

@@ -876,7 +876,7 @@ def create_reverse_dependency_map() -> dict[str, list[str]]:
     # all the modules impacted by that init.
     for m in [f for f in all_modules if f.endswith("__init__.py")]:
         direct_deps = get_module_dependencies(m, cache=cache)
-        deps = sum([reverse_map[d] for d in direct_deps if not d.endswith("__init__.py")], direct_deps)
+        deps = sum((reverse_map[d] for d in direct_deps if not d.endswith("__init__.py")), direct_deps)
         reverse_map[m] = list(set(deps) - {m})
 
     return reverse_map