Hello, after reading the paper, I couldn't find any information about GPU usage during training, including the GPU model and quantity (it seems to be 4 GPUs from the code). When we tried to train the model to reproduce the results, we encountered memory overflow on our 8 L40 GPUs. Therefore, I'm raising this issue in hopes of getting help.
Hello, after reading the paper, I couldn't find any information about GPU usage during training, including the GPU model and quantity (it seems to be 4 GPUs from the code). When we tried to train the model to reproduce the results, we encountered memory overflow on our 8 L40 GPUs. Therefore, I'm raising this issue in hopes of getting help.