Jinghui Lu

RE Instructions used in OFA-CN

5

Hi, thanks again for contributing such good work. Just wondering have you revealed prompts(i.e., instructions) for several multi-modality tasks used in OFA-CN, especially for visual grounding task? thanks.

[BUG/Help] 使用deepspeed做全量finetune的时候，开启warmup_steps，保存下来的模型参数特别大。

6

### Is there an existing issue for this? - [X] I have searched the existing issues ### Current Behavior 使用deepspeed做全量finetune的时候，开启warmup_steps，保存下来的模型参数特别大。 保存下来3块模型，每块23g如图  但是加入我没有打开warmup_steps，就只保存下来两块模型，每个12g  请问有大佬遇到过吗 ### Expected Behavior _No response_...