How to merge lora model

[xjwhy]

Thanks for the work, I reproduced the training, but merging lora modules got an error.

if q_len % group_size > 0:
raise ValueError("q_len %d should be divisible by group size %d."%(q_len, group_size))

I find modeling_llama.py in LongQLoRA-Llama2-7b-8k you released is different from the configuration in training. So which base model needs to be selected to merge.
Thanks