Thanks for your feedback! Can u retry the latest version?
Thanks for your feedback! Can u retry the latest version?
Originally posted by @yuhangzang in https://github.com/InternLM/InternLM-XComposer/issues/288#issuecomment-2071359528
你好,4.23日13点更新了代码,微调internlm-xcomposer2-4khd-7b报下面的错误,麻烦再看一下:

Hi, can u post the parameters of your finetune_lora.sh file?
这是我的微调数据,这个数据微调internlm-xcomposer2-vl-7b没有问题。
This issue is due to the current 4khd fine-tuning code does not support multiple images in one query. Please stay tuned.
This issue is due to the current 4khd fine-tuning code does not support multiple images in one query. Please stay tuned.
应该是你说的这个情况,将训练数据中的多图改成一张图后,可以Lora训练完成。
请问Lora训练后,可以合并吗?