run code formatter

Author: voorhs

src/autointent/_dump_tools/unit_dumpers.py

@@ -217,7 +217,7 @@ def dump(obj: "PeftModel", path: Path, exists_ok: bool) -> None:
             # strategy to save lora models: merge adapters and save as usual hugging face model
             lora_path = path / "lora"
             lora_path.mkdir(parents=True, exist_ok=exists_ok)
-            merged_model: "PreTrainedModel" = obj.merge_and_unload()
+            merged_model: PreTrainedModel = obj.merge_and_unload()
             merged_model.save_pretrained(lora_path)
 
     @staticmethod

src/autointent/_wrappers/embedder/sentence_transformers.py

@@ -257,7 +257,7 @@ def train(self, utterances: list[str], labels: ListOfLabels, config: EmbedderFin
                 eval_strategy="epoch",
                 greater_is_better=False,
             )
-            callbacks: list["TrainerCallback"] = [
+            callbacks: list[TrainerCallback] = [
                 transformers.EarlyStoppingCallback(
                     early_stopping_patience=config.early_stopping_patience,
                     early_stopping_threshold=config.early_stopping_threshold,

src/autointent/context/data_handler/_stratification.py

@@ -156,7 +156,7 @@ def _split_multilabel(self, dataset: HFDataset, test_size: float) -> Sequence[np
             A sequence containing indices for train and test splits.
         """
         if self.random_seed is not None:
-            # Set all seeds for reproducibility (workaround for buggy nature of IterativeStratification from skmultilearn)
+            # Set all seeds for reproducibility (workaround for bugs in IterativeStratification from skmultilearn)
             random.seed(self.random_seed)
         splitter = IterativeStratification(
             n_splits=2,

src/autointent/modules/scoring/_bert.py

@@ -201,7 +201,7 @@ def _train(self, tokenized_dataset: DatasetDict) -> None:
             trainer.train()
 
     def _get_trainer_callbacks(self) -> list["TrainerCallback"]:
-        res: list["TrainerCallback"] = []
+        res: list[TrainerCallback] = []
         if self.early_stopping_config.metric is not None:
             res.append(
                 self._EarlyStoppingCallback(

src/autointent/modules/scoring/_gcn/gcn_model.py

@@ -1,6 +1,5 @@
 import json
 from pathlib import Path
-from typing import cast
 
 import torch
 from pydantic import BaseModel

tests/modules/scoring/test_catboost.py

@@ -10,6 +10,7 @@
 
 pytest.importorskip("catboost")
 
+
 def test_catboost_scorer_dump_load(dataset):
     """Test that CatBoostScorer can be saved and loaded while preserving predictions."""
     data_handler = DataHandler(dataset)