Unable to Export LoRA Adapter in safetensors Format After Fine-tuning

I followed the [LoRA Fine-tuning guide](https://gemma-llm.readthedocs.io/en/latest/colab_lora_finetuning.html) to fine-tune a Gemma3-4B model using LoRA. After the fine-tuning completed, I attempted to save the LoRA adapter using the following code:

```python
original, lora = peft.split_params(state.params)
gm.ckpts.save_params(lora, "/content/gemma3-lora")
```

However, this does not seem to save the adapter in the `safetensors` format. I need to export the LoRA adapter in `safetensors` format so that I can run the adapter with vLLM.

Could anyone provide guidance on how to properly save the LoRA adapter in `safetensors` format?
Thank you!