improve KTO impl., replace datasets

Former-commit-id: e56a57ddcf061de6e4acc8679f7dbf0b68364986

src/llamafactory/train/kto/workflow.py

@@ -48,9 +48,9 @@ def run_kto(
         ref_model=ref_model,
         args=training_args,
         finetuning_args=finetuning_args,
-        tokenizer=tokenizer,
         data_collator=data_collator,
         callbacks=callbacks,
+        **tokenizer_module,
         **split_dataset(dataset, data_args, training_args),
     )