How to continue training LoRA made without llama factory?

[Sehyo]

Reminder

• [X] I have read the README and searched the existing issues.

System Info

I have a lora adapter I have made before with unsloth. I want to do KTO training on it with Llama factory, how can I do this? Setting Custom model and lora adapter path does not seem to work. I would prefer not having to merge the model, as I want to run just the base model in vllm and then load a single lora. How can I accomplish this? Thanks!

Reproduction

(Help)

Expected behavior

No response

Others

No response

[hiyouga]

have you tried --adapter_name_or_path?

[Sehyo]

Hi there! How would I configure that with the webui? Thanks!