Fix pylint generator warnings

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

Author: cyyever

examples/legacy/run_chinese_ref.py

@@ -55,7 +55,7 @@ def get_chinese_word(tokens: list[str]):
 def add_sub_symbol(bert_tokens: list[str], chinese_word_set: set()):
     if not chinese_word_set:
         return bert_tokens
-    max_word_len = max([len(w) for w in chinese_word_set])
+    max_word_len = max(len(w) for w in chinese_word_set)
 
     bert_word = bert_tokens
     start, end = 0, len(bert_word)

examples/pytorch/question-answering/run_qa_no_trainer.py

@@ -950,7 +950,7 @@ def create_and_fill_np_array(start_or_end_logits, dataset, max_len):
             all_start_logits.append(accelerator.gather_for_metrics(start_logits).cpu().numpy())
             all_end_logits.append(accelerator.gather_for_metrics(end_logits).cpu().numpy())
 
-    max_len = max([x.shape[1] for x in all_start_logits])  # Get the max_length of the tensor
+    max_len = max(x.shape[1] for x in all_start_logits)  # Get the max_length of the tensor
 
     # concatenate the numpy array
     start_logits_concat = create_and_fill_np_array(all_start_logits, eval_dataset, max_len)
@@ -989,7 +989,7 @@ def create_and_fill_np_array(start_or_end_logits, dataset, max_len):
                 all_start_logits.append(accelerator.gather_for_metrics(start_logits).cpu().numpy())
                 all_end_logits.append(accelerator.gather_for_metrics(end_logits).cpu().numpy())
 
-        max_len = max([x.shape[1] for x in all_start_logits])  # Get the max_length of the tensor
+        max_len = max(x.shape[1] for x in all_start_logits)  # Get the max_length of the tensor
         # concatenate the numpy array
         start_logits_concat = create_and_fill_np_array(all_start_logits, predict_dataset, max_len)
         end_logits_concat = create_and_fill_np_array(all_end_logits, predict_dataset, max_len)

src/transformers/generation/beam_constraints.py

@@ -218,7 +218,7 @@ def __init__(self, nested_token_ids: list[list[int]], no_subsets=True):
         r"""
         A helper class that builds a trie with the words represented in `nested_token_ids`.
         """
-        self.max_height = max([len(one) for one in nested_token_ids])
+        self.max_height = max(len(one) for one in nested_token_ids)
 
         root = {}
         for token_ids in nested_token_ids:
@@ -260,7 +260,7 @@ def count_leaves(self, root):
         if len(next_nodes) == 0:
             return 1
         else:
-            return sum([self.count_leaves(nn) for nn in next_nodes])
+            return sum(self.count_leaves(nn) for nn in next_nodes)
 
     def has_subsets(self, trie, nested_token_ids):
         """
@@ -373,7 +373,7 @@ def __init__(self, constraints: list[Constraint]):
         self.constraints = constraints
 
         # max # of steps required to fulfill a given constraint
-        self.max_seqlen = max([c.seqlen for c in constraints])
+        self.max_seqlen = max(c.seqlen for c in constraints)
         self.n_constraints = len(constraints)
         self.completed = False
 

src/transformers/generation/stopping_criteria.py

@@ -249,7 +249,7 @@ def __init__(self, tokenizer: PreTrainedTokenizerBase, stop_strings: Union[str,
             token_list, token_indices, tokenizer
         )
 
-        self.maximum_token_len = max([len(stop_string) for stop_string in self.stop_strings])
+        self.maximum_token_len = max(len(stop_string) for stop_string in self.stop_strings)
         self.num_stop_strings = len(self.stop_strings)
         self.target_lens = torch.tensor([len(stop_string) for stop_string in stop_strings], dtype=torch.int32)
 

src/transformers/modeling_utils.py

@@ -4098,9 +4098,9 @@ def get_memory_footprint(self, return_buffers=True):
                 are tensors that do not require gradients and not registered as parameters. E.g. mean and std in batch
                 norm layers. Please see: https://discuss.pytorch.org/t/what-pytorch-means-by-buffers/120266/2
         """
-        mem = sum([param.nelement() * param.element_size() for param in self.parameters()])
+        mem = sum(param.nelement() * param.element_size() for param in self.parameters())
         if return_buffers:
-            mem_bufs = sum([buf.nelement() * buf.element_size() for buf in self.buffers()])
+            mem_bufs = sum(buf.nelement() * buf.element_size() for buf in self.buffers())
             mem = mem + mem_bufs
         return mem
 

src/transformers/models/deprecated/mctct/modeling_mctct.py

@@ -96,7 +96,7 @@ def __init__(self, config):
     def forward(self, input_features):
         # NOTE: in reference to the NOTE in __init__, right now it just calculates padding as if
         # there will be just one conv layer.
-        padding = sum([size // 2 for size in self.kernel_size])  # (7, 7) -> (3, 3)
+        padding = sum(size // 2 for size in self.kernel_size)  # (7, 7) -> (3, 3)
 
         input_features = torch.nn.functional.pad(input_features, (0, 0, padding, padding), "constant", 0)
         hidden_states = input_features.transpose(1, 2).contiguous()  # -> Batch x Frame x Time

src/transformers/models/deprecated/transfo_xl/modeling_transfo_xl.py

@@ -425,8 +425,8 @@ def _get_new_num_tokens_layer(self, new_num_tokens, layer):
 
         new_num_tokens_layer = (
             new_num_tokens
-            - sum([emb.weight.shape[0] for emb in embeddings.emb_layers[:layer]])
-            - sum([emb.weight.shape[0] for emb in embeddings.emb_layers[layer + 1 :]])
+            - sum(emb.weight.shape[0] for emb in embeddings.emb_layers[:layer])
+            - sum(emb.weight.shape[0] for emb in embeddings.emb_layers[layer + 1 :])
         )
         return new_num_tokens_layer, layer
 

src/transformers/models/deprecated/tvlt/feature_extraction_tvlt.py

@@ -202,7 +202,7 @@ def __call__(
 
         # Create audio attention mask
         max_patch_len = max(
-            [ceil(feature.shape[0] / self.patch_size[0]) * self.freq_len for feature in audio_features]
+            ceil(feature.shape[0] / self.patch_size[0]) * self.freq_len for feature in audio_features
         )  # The maximum number of audio patches in a batch
         if return_attention_mask:
             audio_mask = [

src/transformers/models/deprecated/tvlt/image_processing_tvlt.py

@@ -392,7 +392,7 @@ def preprocess(
                     f"number of frames must not be greater than the maximum frames of the model {self.num_frames}."
                 )
 
-        max_num_frames = max([len(video) for video in videos])
+        max_num_frames = max(len(video) for video in videos)
         num_patches_per_image = (size["shortest_edge"] // patch_size[0]) ** 2
         video_masks = np.array(
             [

src/transformers/models/emu3/image_processing_emu3.py

@@ -266,8 +266,8 @@ def _pad_for_batching(
         """
 
         max_shape = (
-            max([size[0] for size in image_sizes]),
-            max([size[1] for size in image_sizes]),
+            max(size[0] for size in image_sizes),
+            max(size[1] for size in image_sizes),
         )
         pixel_values = [
             pad(