RayYoh

> Hi @RayYoh , Yes, i think elastic transform gives like +4 for OneFormer3D and for SPFormer also. No semantic segmentation doesn't have positive impact on instance segmentation metrics. Also...

Yeah, thanks for your suggestion. Additionally, I am confused about the results of the checkpoint in the repo. It seems that the instance segmentation result is better than the original...

> Could you tell me how to load a model instead of training whole process？ I found that although I write the same exp name and there exactly exist an...

>  Hello, Could you please help me understand why these issues are arising and suggest possible solutions? Thank you for your time and assistance. I guess...

Yeah, it's true the order doesn't matter. But I have not understood the distortion parameters, I mean granularity and magnitude. Which have different scales. [[6, 40], [20, 160]] vs [[0.2,...

Alright, thanks for your reply.

Hope this is helpful for you. ``` @HOOKS.register_module() class CustomInformationWriter(InformationWriter): def __init__(self, interval=1, key=("loss", )): super().__init__() self.interval = interval self.logger_key = key def after_step(self): if "model_output_dict" in self.trainer.comm_info.keys(): model_output_dict =...

> Hi, this is an existing issue I will try to solve in a future version (#103), so you can try to remove S3DIS from the joint training list as...

> > > Hi, this is an existing issue I will try to solve in a future version (#103), so you can try to remove S3DIS from the joint training...

Thanks for your prompt response. For the fluctuated GPU utilization, I mean sometimes the utilization will be 0 (just occasional). I guess it may also be caused by complex data...