running with corrected black version

Author: cirquit

demo/halomi/reproduce_evaluation.py

@@ -118,10 +118,7 @@ def main(data_root):
     print("Direction-wise mean score for hallucination detection:")
     results = {}
     for fe_name, fe in feature_sets.items():
-        (
-            word_level_preds,
-            direction2auc,
-        ) = evaluate_token_level_features(
+        (word_level_preds, direction2auc,) = evaluate_token_level_features(
             tokens_dataset, fe, word_to_character_maps, word_labels
         )
         auc = np.mean(list(direction2auc.values()))
@@ -152,10 +149,7 @@ def main(data_root):
     results = {}
     print("Direction-wise mean score for omission detection:")
     for fe_name, fe in feature_sets.items():
-        (
-            word_level_preds,
-            direction2auc,
-        ) = evaluate_token_level_features(
+        (word_level_preds, direction2auc,) = evaluate_token_level_features(
             tokens_dataset, fe, word_to_character_maps, word_labels
         )
         auc = np.mean(list(direction2auc.values()))

stopes/core/cache.py

@@ -80,15 +80,17 @@ def save_cache(
         value: tp.Any,
         iteration_value: tp.Optional[tp.Any] = None,
         iteration_index: int = 0,
-    ): ...
+    ):
+        ...
 
     @abstractmethod
     def invalidate_cache(
         self,
         module: "StopesModule",
         iteration_value: tp.Optional[tp.Any] = None,
         iteration_index: int = 0,
-    ): ...
+    ):
+        ...
 
     def invalidate_module_cache(
         self,

stopes/eval/local_prosody/unity2_forced_aligner_f2.py

@@ -40,9 +40,7 @@ class UnitY2F2ForcedAlignerConfig:
     aligner_name: str = "nar_t2u_aligner"
     unit_extractor_name: str = "xlsr2_1b_v2"
     unit_extractor_output_layer_n: int = 35
-    unit_extractor_kmeans_uri: str = (
-        "https://dl.fbaipublicfiles.com/seamlessM4T/models/unit_extraction/kmeans_10k.npy"
-    )
+    unit_extractor_kmeans_uri: str = "https://dl.fbaipublicfiles.com/seamlessM4T/models/unit_extraction/kmeans_10k.npy"
     device: str = "cpu"
 
     fps: float = (

stopes/modules/bitext/mining/calculate_distances.py

@@ -36,9 +36,9 @@ class CalculateDistancesConfig:
     index_other_lang: str = MISSING  # "path/to/index"
     index_other_lang_type: str = MISSING  # type of the index
 
-    output_dir: tp.Optional[str] = (
-        None  # If None, will be set to dist.src-tgt.knn.numprobe.gpu_type
-    )
+    output_dir: tp.Optional[
+        str
+    ] = None  # If None, will be set to dist.src-tgt.knn.numprobe.gpu_type
 
     num_probe: int = 128
     knn: int = 16

stopes/modules/preprocess/sonar_text_embedding.py

@@ -91,7 +91,8 @@ def _apply_on_simple_column(
         self,
         col: tp.Union[pa.Array, pa.ChunkedArray],
         lang_value: str,
-    ) -> tp.Union[pa.Array, pa.ChunkedArray]: ...
+    ) -> tp.Union[pa.Array, pa.ChunkedArray]:
+        ...
 
     def _apply_on_unique_lang_table(
         self, table: pa.Table, config: LangColumnConfig

stopes/modules/preprocess/train_spm.py

@@ -27,9 +27,7 @@ class TrainSpmConfig:
     model_type: str = "unigram"
     shuffle_input_sentence: bool = True
     num_threads: int = 4
-    model_prefix_spm: str = (
-        ""  # optional value; if passed as empty, will be auto set based on train_data_file name
-    )
+    model_prefix_spm: str = ""  # optional value; if passed as empty, will be auto set based on train_data_file name
 
 
 @dataclasses.dataclass()

stopes/modules/speech/denoise.py

@@ -60,9 +60,7 @@ class DenoiserConfig:
     compute_snr: whether to compute signal-to-noise ratio after denoising
     """
 
-    shards: tp.Any = (
-        MISSING  # it is tp.Union[Path, tp.List[Path]], but such annotation is not supported by OmegaConf
-    )
+    shards: tp.Any = MISSING  # it is tp.Union[Path, tp.List[Path]], but such annotation is not supported by OmegaConf
     output_dir: Path = MISSING
     model_name: str = "mdx_extra"
     source_id: tp.Union[str, int] = "vocals"

stopes/modules/speech/video_alignement/video_segmentor.py

@@ -92,9 +92,9 @@ class WhisperSegmentorConfig:
     """
     Filtering option. If not None, it represents a list of languages to restrict on.
     """
-    sibling_ids: tp.Optional[tp.List[str]] = (
-        None  # specific videos ids that we want to segment
-    )
+    sibling_ids: tp.Optional[
+        tp.List[str]
+    ] = None  # specific videos ids that we want to segment
     """
     Filtering option. If not None, it represents a list of ids to restrict on.
     """
@@ -272,10 +272,10 @@ def get_sentence_segments_dataframe(
             & (words_df["capital_case"] | words_df["punctuation_nonstops"])
         ) | words_df["punctuation_stops"]
 
-        words_df["new_segment_index"] = (
-            WhisperSegmentorModule._segments_with_duration_limit(
-                words_df, max_duration_expected
-            )
+        words_df[
+            "new_segment_index"
+        ] = WhisperSegmentorModule._segments_with_duration_limit(
+            words_df, max_duration_expected
         )
 
         grouper_ = words_df.groupby("new_segment_index")

stopes/modules/tests/test_embedding_utils.py

@@ -98,7 +98,9 @@ def test_save_embeddings(
         sample = np.random.choice(
             len(test_data), round(len(test_data) * sample_fraction), replace=False
         )
-        final_result = test_data[sample,]
+        final_result = test_data[
+            sample,
+        ]
     emb.save(copy_path, sample, fp16, mode=read_mode)
     assert np.array_equal(np.load(copy_path), final_result)
 

stopes/modules/tests/test_partitioned_data_mapper.py

@@ -38,7 +38,8 @@ class IdentityPartitionedDataMapper(PartitionedDataMapper):
     def get_batch_mapper(self):
         return lambda batch: batch
 
-    def requirements(self): ...
+    def requirements(self):
+        ...
 
     def get_custom_metadata(self, *args, **kwargs) -> Dict[str, Any]:
         return {}