[Question] LoRA & QLoRA support in XTuner V1.0 (related to #1381)

[hhhFuture]

Does XTuner V1.0 currently support LoRa and QLora fine-tuning? I currently only see the relevant configurations in the documentation of version 0.2. I couldn't find clear configuration instructions in the documentation of XTuner V1.0. I saw that the related issue was published in October 2025 with the number #1381

If anyone knows, please let me know. Thank you in advance.

[CyCle1024]

Does XTuner V1.0 currently support LoRa and QLora fine-tuning? I currently only see the relevant configurations in the documentation of version 0.2. I couldn't find clear configuration instructions in the documentation of XTuner V1.0. I saw that the related issue was published in October 2025 with the number #1381

If anyone knows, please let me know. Thank you in advance.

@hhhFuture
Thanks for your question. XTuner V1.0 currently does not support LoRa and QLora fine-tuning?

We were working on internal feature development. There were an attempt for lora development, see #1304. But unfortunnately it was disrrupted by other high priority job and the origin developer is still working on that job.

I have some time recently and would try to take over it. I would inform you in this issue if it's ready for using.

[hhhFuture]

Does XTuner V1.0 currently support LoRa and QLora fine-tuning? I currently only see the relevant configurations in the documentation of version 0.2. I couldn't find clear configuration instructions in the documentation of XTuner V1.0. I saw that the related issue was published in October 2025 with the number #1381
If anyone knows, please let me know. Thank you in advance.

@hhhFuture Thanks for your question. XTuner V1.0 currently does not support LoRa and QLora fine-tuning?

We were working on internal feature development. There were an attempt for lora development, see #1304. But unfortunnately it was disrrupted by other high priority job and the origin developer is still working on that job.

I have some time recently and would try to take over it. I would inform you in this issue if it's ready for using.

Thank you very much for your reply. As this is my first time using the Xtuner framework, I have been reading the documentation for Xtuner 1.0.0 and some code on GitHub for the past few days. I found that there are relatively few supported models in Xtuner 0.2.0 version. Currently, I see that there are relatively few supported models, such as the qwen series models. In the configuration, only qwen1 and qwen1.5 series models can be seen. The corresponding link is as follows:
https://github.com/InternLM/xtuner/tree/v0.2.0/xtuner
In the official V1 document, the Qwen3-8B model is supported, and the corresponding link is:
https://xtuner.readthedocs.io/zh-cn/latest/pretrain_sft/tutorial/llm_trainer.html
So I don't know if version 1.0.0 will add support for different models. (Actually, I have a small suggestion personally. I hope to list some supported series of models in the document.)