Set push to hub as private by default

src/axolotl/utils/trainer.py

@@ -181,6 +181,7 @@ def setup_trainer(cfg, train_dataset, eval_dataset, model, tokenizer):
     if cfg.hub_model_id:
         training_arguments_kwargs["hub_model_id"] = cfg.hub_model_id
         training_arguments_kwargs["push_to_hub"] = True
+        training_arguments_kwargs["hub_private_repo"] = True
 
     training_args = AxolotlTrainingArguments(
         per_device_train_batch_size=cfg.micro_batch_size,