Fix quality due to ruff release

examples/flax/language-modeling/run_bart_dlm_flax.py

@@ -319,15 +319,13 @@ class FlaxDataCollatorForBartDenoisingLM:
         sentence_ends = np.argwhere(end_sentence_mask)
         sentence_ends[:, 1] += 1
         example_has_multiple_sentences, num_sentences = np.unique(sentence_ends[:, 0], return_counts=True)
-        num_sentences_map = {sent_idx: count for sent_idx, count in zip(example_has_multiple_sentences, num_sentences)}
+        num_sentences_map = dict(zip(example_has_multiple_sentences, num_sentences))
 
         num_to_permute = np.ceil(num_sentences * self.permute_sentence_ratio).astype(int)
-        num_to_permute_map = {
-            sent_idx: count for sent_idx, count in zip(example_has_multiple_sentences, num_to_permute)
-        }
+        num_to_permute_map = dict(zip(example_has_multiple_sentences, num_to_permute))
 
         sentence_ends = np.split(sentence_ends[:, 1], np.unique(sentence_ends[:, 0], return_index=True)[1][1:])
-        sentence_ends_map = {sent_idx: count for sent_idx, count in zip(example_has_multiple_sentences, sentence_ends)}
+        sentence_ends_map = dict(zip(example_has_multiple_sentences, sentence_ends))
 
         for i in range(input_ids.shape[0]):
             if i not in example_has_multiple_sentences:

examples/legacy/pytorch-lightning/run_glue.py

@@ -124,7 +124,7 @@ class GLUETransformer(BaseTransformer):
 
         results = {**{"val_loss": val_loss_mean}, **compute_metrics(self.hparams.task, preds, out_label_ids)}
 
-        ret = {k: v for k, v in results.items()}
+        ret = dict(results.items())
         ret["log"] = results
         return ret, preds_list, out_label_list
 

examples/legacy/pytorch-lightning/run_ner.py

@@ -122,7 +122,7 @@ class NERTransformer(BaseTransformer):
         preds = np.argmax(preds, axis=2)
         out_label_ids = np.concatenate([x["target"] for x in outputs], axis=0)
 
-        label_map = {i: label for i, label in enumerate(self.labels)}
+        label_map = dict(enumerate(self.labels))
         out_label_list = [[] for _ in range(out_label_ids.shape[0])]
         preds_list = [[] for _ in range(out_label_ids.shape[0])]
 
@@ -140,7 +140,7 @@ class NERTransformer(BaseTransformer):
             "f1": f1_score(out_label_list, preds_list),
         }
 
-        ret = {k: v for k, v in results.items()}
+        ret = dict(results.items())
         ret["log"] = results
         return ret, preds_list, out_label_list
 

examples/legacy/seq2seq/run_eval_search.py

@@ -34,7 +34,7 @@ task_score_names = {
 
 def parse_search_arg(search):
     groups = search.split()
-    entries = {k: vs for k, vs in (g.split("=") for g in groups)}
+    entries = dict((g.split("=") for g in groups))
     entry_names = list(entries.keys())
     sets = [[f"--{k} {v}" for v in vs.split(":")] for k, vs in entries.items()]
     matrix = [list(x) for x in itertools.product(*sets)]
@@ -105,7 +105,7 @@ def run_search():
     col_widths = {col: len(str(col)) for col in col_names}
     results = []
     for r in matrix:
-        hparams = {k: v for k, v in (x.replace("--", "").split() for x in r)}
+        hparams = dict((x.replace("--", "").split() for x in r))
         args_exp = " ".join(r).split()
         args_exp.extend(["--bs", str(args.bs)])  # in case we need to reduce its size due to CUDA OOM
         sys.argv = args_normal + args_exp

examples/legacy/token-classification/run_ner.py

@@ -158,7 +158,7 @@ def main():
 
     # Prepare CONLL-2003 task
     labels = token_classification_task.get_labels(data_args.labels)
-    label_map: Dict[int, str] = {i: label for i, label in enumerate(labels)}
+    label_map: Dict[int, str] = dict(enumerate(labels))
     num_labels = len(labels)
 
     # Load pretrained model and tokenizer

examples/legacy/token-classification/run_tf_ner.py

@@ -144,7 +144,7 @@ def main():
 
     # Prepare Token Classification task
     labels = token_classification_task.get_labels(data_args.labels)
-    label_map: Dict[int, str] = {i: label for i, label in enumerate(labels)}
+    label_map: Dict[int, str] = dict(enumerate(labels))
     num_labels = len(labels)
 
     # Load pretrained model and tokenizer

examples/pytorch/token-classification/run_ner.py

@@ -407,7 +407,7 @@ def main():
 
     # Set the correspondences label/ID inside the model config
     model.config.label2id = {l: i for i, l in enumerate(label_list)}
-    model.config.id2label = {i: l for i, l in enumerate(label_list)}
+    model.config.id2label = dict(enumerate(label_list))
 
     # Map that sends B-Xxx label to its I-Xxx counterpart
     b_to_i_label = []

examples/pytorch/token-classification/run_ner_no_trainer.py

@@ -442,7 +442,7 @@ def main():
 
     # Set the correspondences label/ID inside the model config
     model.config.label2id = {l: i for i, l in enumerate(label_list)}
-    model.config.id2label = {i: l for i, l in enumerate(label_list)}
+    model.config.id2label = dict(enumerate(label_list))
 
     # Map that sends B-Xxx label to its I-Xxx counterpart
     b_to_i_label = []

examples/research_projects/layoutlmv3/run_funsd_cord.py

@@ -294,11 +294,11 @@ def main():
     if isinstance(features[label_column_name].feature, ClassLabel):
         label_list = features[label_column_name].feature.names
         # No need to convert the labels since they are already ints.
-        id2label = {k: v for k, v in enumerate(label_list)}
+        id2label = dict(enumerate(label_list))
         label2id = {v: k for k, v in enumerate(label_list)}
     else:
         label_list = get_label_list(datasets["train"][label_column_name])
-        id2label = {k: v for k, v in enumerate(label_list)}
+        id2label = dict(enumerate(label_list))
         label2id = {v: k for k, v in enumerate(label_list)}
     num_labels = len(label_list)
 

examples/research_projects/rag-end2end-retriever/finetune_rag.py

@@ -360,7 +360,7 @@ class GenerativeQAModule(BaseTransformer):
 
         loss_tensors = self._step(batch)
 
-        logs = {name: loss for name, loss in zip(self.loss_names, loss_tensors)}
+        logs = dict(zip(self.loss_names, loss_tensors))
         # tokens per batch
         tgt_pad_token_id = (
             self.tokenizer.generator.pad_token_id
@@ -434,7 +434,7 @@ class GenerativeQAModule(BaseTransformer):
         target: List[str] = self.ids_to_clean_text(batch["decoder_input_ids"])
         # print(preds,target)
         loss_tensors = self._step(batch)
-        base_metrics = {name: loss for name, loss in zip(self.loss_names, loss_tensors)}
+        base_metrics = dict(zip(self.loss_names, loss_tensors))
         gen_metrics: Dict = self.calc_generative_metrics(preds, target)
 
         summ_len = np.mean(lmap(len, generated_ids))

examples/research_projects/rag/finetune_rag.py

@@ -321,7 +321,7 @@ class GenerativeQAModule(BaseTransformer):
         preds: List[str] = self.ids_to_clean_text(generated_ids)
         target: List[str] = self.ids_to_clean_text(batch["decoder_input_ids"])
         loss_tensors = self._step(batch)
-        base_metrics = {name: loss for name, loss in zip(self.loss_names, loss_tensors)}
+        base_metrics = dict(zip(self.loss_names, loss_tensors))
         gen_metrics: Dict = self.calc_generative_metrics(preds, target)
 
         summ_len = np.mean(lmap(len, generated_ids))

examples/research_projects/seq2seq-distillation/finetune.py

@@ -170,7 +170,7 @@ class SummarizationModule(BaseTransformer):
     def training_step(self, batch, batch_idx) -> Dict:
         loss_tensors = self._step(batch)
 
-        logs = {name: loss for name, loss in zip(self.loss_names, loss_tensors)}
+        logs = dict(zip(self.loss_names, loss_tensors))
         # tokens per batch
         logs["tpb"] = batch["input_ids"].ne(self.pad).sum() + batch["labels"].ne(self.pad).sum()
         logs["bs"] = batch["input_ids"].shape[0]
@@ -225,7 +225,7 @@ class SummarizationModule(BaseTransformer):
         preds: List[str] = self.ids_to_clean_text(generated_ids)
         target: List[str] = self.ids_to_clean_text(batch["labels"])
         loss_tensors = self._step(batch)
-        base_metrics = {name: loss for name, loss in zip(self.loss_names, loss_tensors)}
+        base_metrics = dict(zip(self.loss_names, loss_tensors))
         rouge: Dict = self.calc_generative_metrics(preds, target)
         summ_len = np.mean(lmap(len, generated_ids))
         base_metrics.update(gen_time=gen_time, gen_len=summ_len, preds=preds, target=target, **rouge)

examples/research_projects/zero-shot-distillation/distill_classifier.py

@@ -303,7 +303,7 @@ def main():
         student_args.student_name_or_path, num_labels=len(class_names)
     )
     tokenizer = AutoTokenizer.from_pretrained(student_args.student_name_or_path, use_fast=data_args.use_fast_tokenizer)
-    model.config.id2label = {i: label for i, label in enumerate(class_names)}
+    model.config.id2label = dict(enumerate(class_names))
     model.config.label2id = {label: i for i, label in enumerate(class_names)}
 
     # 4. train student on teacher predictions

src/transformers/benchmark/benchmark_utils.py

@@ -610,7 +610,7 @@ class Benchmark(ABC):
                 model_name: AutoConfig.from_pretrained(model_name) for model_name in self.args.model_names
             }
         else:
-            self.config_dict = {model_name: config for model_name, config in zip(self.args.model_names, configs)}
+            self.config_dict = dict(zip(self.args.model_names, configs))
 
         warnings.warn(
             f"The class {self.__class__} is deprecated. Hugging Face Benchmarking utils"

src/transformers/modelcard.py

@@ -399,9 +399,9 @@ class TrainingSummary:
         dataset_metadata = _listify(self.dataset_metadata)
         if len(dataset_args) < len(dataset_tags):
             dataset_args = dataset_args + [None] * (len(dataset_tags) - len(dataset_args))
-        dataset_mapping = {tag: name for tag, name in zip(dataset_tags, dataset_names)}
-        dataset_arg_mapping = {tag: arg for tag, arg in zip(dataset_tags, dataset_args)}
-        dataset_metadata_mapping = {tag: metadata for tag, metadata in zip(dataset_tags, dataset_metadata)}
+        dataset_mapping = dict(zip(dataset_tags, dataset_names))
+        dataset_arg_mapping = dict(zip(dataset_tags, dataset_args))
+        dataset_metadata_mapping = dict(zip(dataset_tags, dataset_metadata))
 
         task_mapping = {
             task: TASK_TAG_TO_NAME_MAPPING[task] for task in _listify(self.tasks) if task in TASK_TAG_TO_NAME_MAPPING

src/transformers/models/esm/tokenization_esm.py

@@ -57,7 +57,7 @@ class EsmTokenizer(PreTrainedTokenizer):
     def __init__(self, vocab_file, **kwargs):
         super().__init__(**kwargs)
         self.all_tokens = load_vocab_file(vocab_file)
-        self._id_to_token = {ind: tok for ind, tok in enumerate(self.all_tokens)}
+        self._id_to_token = dict(enumerate(self.all_tokens))
         self._token_to_id = {tok: ind for ind, tok in enumerate(self.all_tokens)}
         self.unk_token = "<unk>"
         self.cls_token = "<cls>"

src/transformers/models/maskformer/convert_maskformer_original_pytorch_checkpoint_to_pytorch.py

@@ -111,7 +111,7 @@ class OriginalMaskFormerConfigToOursConverter:
         swin = model.SWIN
 
         dataset_catalog = MetadataCatalog.get(original_config.DATASETS.TEST[0])
-        id2label = {idx: label for idx, label in enumerate(dataset_catalog.stuff_classes)}
+        id2label = dict(enumerate(dataset_catalog.stuff_classes))
         label2id = {label: idx for idx, label in id2label.items()}
 
         config: MaskFormerConfig = MaskFormerConfig(

src/transformers/models/oneformer/convert_to_hf_oneformer.py

@@ -122,7 +122,7 @@ class OriginalOneFormerConfigToOursConverter:
         model = original_config.MODEL
 
         dataset_catalog = MetadataCatalog.get(original_config.DATASETS.TEST_PANOPTIC[0])
-        id2label = {idx: label for idx, label in enumerate(dataset_catalog.stuff_classes)}
+        id2label = dict(enumerate(dataset_catalog.stuff_classes))
         label2id = {label: idx for idx, label in id2label.items()}
 
         if is_swin:

src/transformers/models/speech_encoder_decoder/convert_speech_to_text_wav2vec2_seq2seq_original_to_pytorch.py

@@ -207,7 +207,7 @@ def create_vocab_dict(dict_path):
         "<unk>": 3,
     }
 
-    vocab_dict.update({k: v for k, v in zip(words, range(4, num_words + 4))})
+    vocab_dict.update(dict(zip(words, range(4, num_words + 4))))
     return vocab_dict
 
 

src/transformers/onnx/convert.py

@@ -179,9 +179,7 @@ def export_pytorch(
                         f=output.as_posix(),
                         input_names=list(config.inputs.keys()),
                         output_names=onnx_outputs,
-                        dynamic_axes={
-                            name: axes for name, axes in chain(config.inputs.items(), config.outputs.items())
-                        },
+                        dynamic_axes=dict(chain(config.inputs.items(), config.outputs.items())),
                         do_constant_folding=True,
                         use_external_data_format=config.use_external_data_format(model.num_parameters()),
                         enable_onnx_checker=True,
@@ -208,7 +206,7 @@ def export_pytorch(
                     f=output.as_posix(),
                     input_names=list(config.inputs.keys()),
                     output_names=onnx_outputs,
-                    dynamic_axes={name: axes for name, axes in chain(config.inputs.items(), config.outputs.items())},
+                    dynamic_axes=dict(chain(config.inputs.items(), config.outputs.items())),
                     do_constant_folding=True,
                     opset_version=opset,
                 )