Liang Zichao
Liang Zichao
> Hello, is there a way to train the model on a GPU with 10GB of VRAM like with tiling during inference? Failing to do complete Stage 1 training due...
> 至少需要11GB，对应的输入大小是512。 请问最小的训练显存是多少呢？
> Hey, we've updated for tiled inference to avoid OOM. You can try this new feature to solve the problem. Hi, Can 24G memory train this model?
以及如果是LQ-HQ这种数据集是否可以将dataset_pairs_wRandomSample_Triplet_txt替换成dataset_pairs_wRandomSample
> 可以加联系方式交流一下吗，我这么训完了也没有完整复现，效果上主要是比较容易出现模糊的问题。 请问您用的几张显卡训练的 得多大显存呢
> Hi, I'm having some issues testing the model trained by this code,： the points in the resulting point cloud are very close to each other. > > This problem...
> yes, you need to resize your data to the same size, otherwise there is a bug here If the size is the same, but it's not 512x512, is that...