kaiallll

Has anyone encountered a situation where the program can run normally independently, only occupying 2GB of GPU memory, but when running through Gradio, it reports the error torch.cuda.OutOfMemoryError?

1

details: Traceback (most recent call last): File "/home/liuxuan/.conda/envs/FA/lib/python3.8/site-packages/gradio/queueing.py", line 489, in call_prediction output = await route_utils.call_process_api( File "/home/liuxuan/.conda/envs/FA/lib/python3.8/site-packages/gradio/route_utils.py", line 232, in call_process_api output = await app.get_blocks().process_api( File "/home/liuxuan/.conda/envs/FA/lib/python3.8/site-packages/gradio/blocks.py", line 1561,...

有没有什么加速推理的方法或者可以借鉴的方案

2

本身模型推理一个320帧视频大约花费40s，加上头部贴会原图5s，高清化（gpen）55s，耗时太长了，有没有什么优化方案可以参考