What kind of GPU do you use to train svd_xt with 1024x576 resolution and 25 frames.

[chenbinghui1]

I find with this configuration, 80G A100 is not enough for even batch_size=1. Am I right?

[fyang064]

As mentioned in the report "Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets," 8 80GB A100 GPUs were used for training.

[DataAIPlayer]

I also tried training with LoRA, and even with a batch size of 1, there was still memory overflow. It seems that model parallel training is needed.