yxuansu/PandaGPT

Video encoding for inference

bitnom opened this issue · 2 comments

bitnom commented

Can you please tell us the optimal video encoder settings in detail? I have tried a few different variations of mp4 (HVENC h.264, x.264, multiple resolutions, and 30/60fps) but I can't get it to recognize my video. On the demo, it always comes up as black but I can play it on my computer.

ghrua commented

Hi @bitnom

Thanks for your feedback. We apologize for this issue. Actually, this is caused by an internal error of pytorchvideo library, which occurs when parsing videos in concurrent requests.

We are actively working on resolving this issue. If it is convenient for you, you could also deploy the demo locally. We didn't observe this error of pytorchvideo when testing on local devices.

We are so sorry for this.

mlhui commented

do you meet this error?:
I use 7B .A decoder-only architecture is being used, but right-padding was detected! For correct generation results, please set padding_side='left' when initializing the tokenizer.

can you tell me how to solve this problem?