why these huggingface spaces not work properly

#7
by sravanK - opened

errors, busy, bla blah blah

sravanK changed discussion title from why this huggingface spaces not work properly to why these huggingface spaces not work properly

This is a popular “space” I have been trying since yesterday and in a few hours 1 movie succeeded,wait(which I doubt that the queue will go down so quickly) or find some API for this model.

3rd day of nothing

Wan-AI org

@Marksssssss I wanted to check if you saw the 'Generate Video' button change to 'Refresh Generation Status'. If so, this means your task has been submitted. You can now click the 'Refresh Generation Status' button to get updates on the task's progress. At the same time, the 'Cost Time', 'Estimated Waiting Time (secs)', and the progress bar will update, showing the task's elapsed time, estimated waiting time, and completion progress, respectively. Please be patient while the video is being generated. Due to a recent surge in traffic, we've had to modify the task submission process. We appreciate your patience and understanding.

image.png

I have tried a few sizes of images and prompts on the "Image to Video" tab. I can confirm that the 'Generate Video' button won't change to 'Refresh Generation Status' after clicking, and the time estimation/progress bar have been always zeros.

@chaojiemao no, for 4 days only ones i saw "Refresh Generation Status", but after hours there was nothing.... since that nothing for all days - zero

nothing works....

same nothing try https://huggingface.co./spaces/fffiloni/Wan2.1

can you do the same for video generation ?

Zrzut ekranu 2025-03-01 170625.png
finally after many time

With the image-to-video option, I get the error message: "Warning: The number of running tasks is too large, the estimate waiting time is 319 s.", and the button does not change into "Refresh Generation Status".
Would it be possible to create another space with the same model where the GPU time is deducted from the user's allotment? I think in that case there should be no queues necessary, only the user's alloted time would be the limit then.

there this alternate
https://huggingface.co./spaces/Dagfinn1962/Wan2.1 (no queue at the moment because is new space)
https://huggingface.co./spaces/fffiloni/Wan2.1
this 2 space using l40s gpu but this is 1.3b version

@Clausss Thank you for the links, but these two spaces both don't support the image-to-video-option, which only seems to be supported by the official (overloaded) space and by the company's Chinese website where everything's written in Chinese, and I probably can't use it. Oh, and by another, older space which seems to have an API key problem, here: https://huggingface.co./spaces/bdsqlsz/WanX2.1

this both space don't support image-video because is 1.3b version of this model(not 14b)so you must have extreme patience or find some api or self host(I have host 1.3b on my rtx 4060 ti)
but with each passing day there will be a rather smaller queue because the hype will subside and api etc will be available

Wan-AI org

@Marksssssss "You'll need to manually click the 'Refresh Generation Status' button to update the task status. It might take up to 10 minutes, so click it after that time to see if generation is complete.

Wan-AI org

We have added concurrency and queue limits today, which may improve the task success rate.

@chaojiemao "To click this button, it first needs to appear, but after 5 days, it still hasn't shown up for me, so there's nothing to click."

finalllyyy!! but at the end I got - Info Warning: task running FAILED, nice

Sign up or log in to comment