hardware problem

[ljx761795750]

Hi, thank you very much for your contribution to this project. I have a question to ask: what does your project hardware look like. What is the graphics card memory for the training project? thanks

[605521553]

I alse have the same question,im trying to use 4070ti to finish this job

[zhihao-lin]

Thanks for your interest in our projects. I used NVIDIA TITAN RTX for training and inference, which has 24G memory

[Sylva-Lin]

Thanks for your interest in our projects. I used NVIDIA TITAN RTX for training and inference, which has 24G memory

Hi, thank you for your great work. I found the phenomenon of increasing the Gpu memory during the training stage, but I don't know which link caused it, does this happen when you're running?

[Sylva-Lin]

Thanks for your interest in our projects. I used NVIDIA TITAN RTX for training and inference, which has 24G memory

Hi, thank you for your great work. I found the phenomenon of increasing the Gpu memory during the training stage, but I don't know which link caused it, does this happen when you're running?

I spent a lot of time trying to solve this problem, and finally found that it was caused by lines 107-109, and l commented it out and there was no longer a problem with memory increments.

[Ting-Devin-Han]

Thanks for your interest in our projects. I used NVIDIA TITAN RTX for training and inference, which has 24G memory

Hi, thank you for your great work. I found the phenomenon of increasing the Gpu memory during the training stage, but I don't know which link caused it, does this happen when you're running?

I spent a lot of time trying to solve this problem, and finally found that it was caused by lines 107-109, and l commented it out and there was no longer a problem with memory increments.

Hi, I had the same problem. Could you please provide details of the solution. Thanks for your reply.

[Sylva-Lin]

Thanks for your interest in our projects. I used NVIDIA TITAN RTX for training and inference, which has 24G memory

Hi, thank you for your great work. I found the phenomenon of increasing the Gpu memory during the training stage, but I don't know which link caused it, does this happen when you're running?

I spent a lot of time trying to solve this problem, and finally found that it was caused by lines 107-109, and l commented it out and there was no longer a problem with memory increments.

Hi, I had the same problem. Could you please provide details of the solution. Thanks for your reply.

my solution is , in ” segmentation/manager.py “，Put lines 107-109 into the following if statement, because “out[mask == 0] = torch.min(out)” takes up a lot of GPU memory if not placed this way

[Ting-Devin-Han]

Thanks for your interest in our projects. I used NVIDIA TITAN RTX for training and inference, which has 24G memory

Hi, thank you for your great work. I found the phenomenon of increasing the Gpu memory during the training stage, but I don't know which link caused it, does this happen when you're running?

I spent a lot of time trying to solve this problem, and finally found that it was caused by lines 107-109, and l commented it out and there was no longer a problem with memory increments.

Hi, I had the same problem. Could you please provide details of the solution. Thanks for your reply.

my solution is , in ” segmentation/manager.py “，Put lines 107-109 into the following if statement, because “out[mask == 0] = torch.min(out)” takes up a lot of GPU memory if not placed this way

Your help is greatly appreciated.

[wxycwymds]

Thanks for your interest in our projects. I used NVIDIA TITAN RTX for training and inference, which has 24G memory

Hi, thank you for your great work. I found the phenomenon of increasing the Gpu memory during the training stage, but I don't know which link caused it, does this happen when you're running?

I spent a lot of time trying to solve this problem, and finally found that it was caused by lines 107-109, and l commented it out and there was no longer a problem with memory increments.

Hi, I had the same problem. Could you please provide details of the solution. Thanks for your reply.

my solution is , in ” segmentation/manager.py “，Put lines 107-109 into the following if statement, because “out[mask == 0] = torch.min(out)” takes up a lot of GPU memory if not placed this way

Your help is greatly appreciated.

I modified the code according to the solution and solved the problem of gradually increasing GPU memory. However, after training for a period of time, the GPU memory suddenly increased and exploded. I wonder that whether you met this issue and how to fix it.Thank you very much！