THUDM / CogVideo

Text-to-video generation. The repo for ICLR2023 paper "CogVideo: Large-scale Pretraining for Text-to-Video Generation via Transformers"
Apache License 2.0
3.54k stars 378 forks source link

4090显卡CUDA out of memory #36

Open chenyf2022 opened 1 year ago

chenyf2022 commented 1 year ago

torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 4.56 GiB (GPU 0; 23.68 GiB total capacity; 18.96 GiB already allocated; 2.84 GiB free; 18.97 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation.

chenyf2022 commented 1 year ago

您好,在使用4090显卡尝试生成视频的时候出现了这个问题,请问有什么解决的方法吗?

wenyihong commented 10 months ago

生成视频需要25G左右的显存。我们正在修复huggingface的demo,之后可以去Huggingface上尝试生成