Closed sfchen94 closed 1 year ago
Hi, thanks for your interest in our work. Our codebase is developed with DistributedDataParallel for multi-GPU training. Is that possible to try with DistributedDataParallel instead of DataParallel?
Thanks for your reply. It is difficult to apply DDP with DP in my case currently. But I will try to seek a way to do that. And hope DenseCLIP can also train with DP later.
Hi, when I apply DenseCLIP to torch.nn.DataParallel for multi-GPUs training, the pretrained model can not load correctly, how can I fix this? (Single GPU training work well)![image](https://user-images.githubusercontent.com/113094161/191021194-454dbfc7-b503-4e2e-892b-52c34ac50b78.png)