Only use self.processing_class with transformers v4.46.0+

and use *args, **kwargs in log; seems safer in case of future changes.

src/setfit/trainer.py

@@ -9,11 +9,13 @@
 from sentence_transformers.model_card import ModelCardCallback as STModelCardCallback
 from sentence_transformers.training_args import BatchSamplers
 from sklearn.preprocessing import LabelEncoder
+from packaging.version import parse as parse_version
 from torch import nn
 from transformers.integrations import CodeCarbonCallback
 from transformers.trainer_callback import IntervalStrategy, TrainerCallback
 from transformers.trainer_utils import HPSearchBackend, default_compute_objective, number_of_arguments, set_seed
 from transformers.utils.import_utils import is_in_notebook
+from transformers import __version__ as transformers_version
 
 from setfit.model_card import ModelCardCallback
 
@@ -72,7 +74,7 @@ def overwritten_call_event(self, event, args, state, control, **kwargs):
                     model=self.setfit_model,
                     st_model=self.model,
                     st_args=args,
-                    tokenizer=self.processing_class,
+                    tokenizer=self.processing_class if parse_version(transformers_version) >= parse_version("4.46.0") else self.tokenizer,
                     optimizer=self.optimizer,
                     lr_scheduler=self.lr_scheduler,
                     train_dataloader=self.train_dataloader,
@@ -156,9 +158,9 @@ def _set_logs_prefix(self, logs_prefix: str) -> None:
         """
         self.logs_prefix = logs_prefix
 
-    def log(self, logs: Dict[str, float], start_time: Optional[float] = None) -> None:
+    def log(self, logs: Dict[str, float], *args, **kwargs) -> None:
         logs = {f"{self.logs_prefix}_{k}" if k == "loss" else k: v for k, v in logs.items()}
-        return super().log(logs, start_time)
+        return super().log(logs, *args, **kwargs)
 
     def evaluate(
         self,