use_dynamic_cfg in cli_demo.py

[JuheonYi]

Hi, I am trying to fine-tune CogVideoX-5B-I2V using LoRA, and run the inference.

I noticed that in the inference/cli_demo.py, use_dynamic_cfg is set as True when calling pipe() but it seems that in the finetuning code (e.g., finetune/models/cogvideox_i2v/lora_trainer.py, validation_step()), use_dynamic_cfg is not specified, and seems to be set as False as default in diffusers code (diffusers/pipelines/cogvideo/pipeline_cogvideox_image2video.py).

May I ask if this would make difference in the inference result?

Thanks!

[zRzRzRzRzRzRzR]

Meeting, but the impact is not great, you can set it according to the actual situation, choose False or True.