python gradio/app.py --port 8001 --host 0.0.0.0 --enable-optimization --model-type v1-HQ-16x512x512 一直报torch.cuda.OutOfMemoryError: CUDA out of memory

Open yangjianchun opened this issue 1 year ago • 3 comments

显卡A100 40G python gradio/app.py --port 8001 --host 0.0.0.0 --enable-optimization --model-type v1-HQ-16x512x512 执行后出现torch.cuda.OutOfMemoryError: CUDA out of memory，

但是在命令行下执行CUDA_VISIBLE_DEVICES=1 torchrun --standalone --nproc_per_node 1 scripts/inference.py configs/opensora/inference/16x512x512.py 又没问题，python gradio/app.py 是哪里配置问题吗

Apr 19 '24 07:04 yangjianchun

You may try replacing the Boolean values from this line and this line with args.enable_optimization.

Apr 21 '24 12:04 JThh

the 还是一样，还是显示显存溢出，有没还有其他优化方式？

Apr 22 '24 08:04 yangjianchun

This issue is stale because it has been open for 7 days with no activity.

Apr 30 '24 01:04 github-actions[bot]