I honestly, have NO knowledge, how it worked.
Did you make sure you pasted the right API key?
It should be "sk-blahblah" like that. The API key was made from the official dashscope site. That's the best I could say.

Let me know how you solved it. I find this problem quite interesting.

not working
regeneration for 1000 sec now
150% generation... lol

not working
regeneration for 1000 sec now
150% generation... lol

I saw the same issue, but didn't pay much attention since it happened once in a blue moon.

Please tell me, Is it happening all the time?

Solution so far:

• Refresh the browser.

Cause of problem(probable):

• Likely, it is an SSR issue perhaps, caused by async tasks.
• Or, you might be giving it Inappropriate Prompts
• Or, Server side problem.

Is the API working?

• Yes. API is showing 100% uptime.

Is the space working?

• Yes. The space was ran over 40+ times, and in all cases, it worked, i.e., it generated a video, in each case.

Thanks for pointing that out!

yes all the time...a prompt is a prompt, an image is an image-not working at all

it's more server ressources...

i'v got same problem, sometime you need wait 900-1000%

lets flag as not working

lets flag as not working

no, it's working, you just need to wait

yes all the time...a prompt is a prompt, an image is an image-not working at all

it's more server ressources...

Solution:

Wait for some more time. The expected time is only a dummy expectation. Not a calculation, as it appears. Your video WILL be generated.
I understand the issue and the frustration.

However, I have figured it out.

The problem is from the Wan2.1's server. It expects generation to be completed in 639 seconds always, no matter what.

Also, I doubt if any generation has failed, as you state, because, from the logs, all I see is successes and 0 failures so far.

Analysis:

The original Wan2.1 space, appears to be heavily used(despite users needing to wait for humongous amount of time). fffiloni's space is good, but it runs the 1B Params model, unlike this which runs the 14B model.

Hence, even if you used the Wan2.1 Space, it would, well, have the same experience except you would need to wait upto 100+% after perhaps, hours of waiting queue. Wouldn't be a good idea to me, but you might try .

@NihalGazi
and here we are again))) "Warning: The number of running tasks is too large, the estimate waiting time is 480 s."

@Marksssssss I have increased the request bucket limit by 100x. Please do let me know, if the issue persists.

@Marksssssss I have increased the request bucket limit by 100x. Please do let me know, if the issue persists.
all is good for now, I think it was little peak moment

worked for the first time after 10 min but...can you increase the parameters to speed up the generation ?

thanks

i cant generate a single video, all the day, i think we have a peak today, .. idk, you guys can generate?

i cant generate a single video, all the day, i think we have a peak today, .. idk, you guys can generate?

Yes it worked. I generated one just now and it took around 1389 seconds, as opposed to 639 seconds that it was showing.

NihalGazi, can you teach how duplicate "Space"?

Thanks to disponibilize the SPACE. You is very smart

NihalGazi, can you teach how duplicate "Space"?

@gusta07
Sure! To duplicate this space:

• Click on the 3 dots near the name of the space.
• Click "Duplicate Space"
• In DASHSCOPE_API_KEY paste your API key
• Done

However, note, that for some reason, some people's API key isn't working, due to reasons unknown to me.

All I can say is, make sure your API key is in format "sk-blah6969...". Perhaps use ChatGPT to debug. I have no knowledge on how mine worked. Perhaps, a miracle.

I create API key in access tokens, and isnt working. How create your?

I dont know about DASHSCOPE_API_KEY...

I create API key in access tokens, and isnt working. How create your?

@gusta0729
To generate a Dashscope API key, you must enable Alibaba Cloud Model Studio and generate an API key within the Model Studio platform.
You'll also require an Alibaba Cloud account and to activate Model Studio.
Once Model Studio has been opened, you can obtain your API key through the platform interface.

And no is free? Oh no hahaha

Hey NihalGazi, the AI is making an infinite loading in your space. Can you help me?

Hi @NihalGazi You did some awesome work and made this work.I have been using your space and created some awesome videos Thank you so much.But when i tried to create my own privtae space i enetered my hf token which didnt worked and then i saw you upper Model Studio reference If anyhow you create it the Alibaba cloud is free trial for sometimes and then we have to recreate this

Hi @NihalGazi You did some awesome work and made this work.I have been using your space and created some awesome videos Thank you so much.But when i tried to create my own privtae space i enetered my hf token which didnt worked and then i saw you upper Model Studio reference If anyhow you create it the Alibaba cloud is free trial for sometimes and then we have to recreate this

@aefsgerrg ,
You're welcome! And I must thank you for such a nice feedback. I appreciate that.

Regarding your problem,

1. Are you pasting the API key properly? It should be something along the lines "sk-blahblah69420..." . Do not miss the "sk-", else it won't work.

I'm still trying to find, as to why mine worked, and how others could make it work too. Once, it's done(along with some other projects), I'll let it know here.

Hey NihalGazi, the AI is making an infinite loading in your space. Can you help me?

@gusta0729 Sure! Refer to this answer's "solution"

so long today....

worked for the first time after 10 min but...can you increase the parameters to speed up the generation ?

thanks

@deeduckme
After considerable testing, it appears that, speed of inference actually depends on the Wan2.1 code provided in DashScope's Backend servers managed by WanAI Team. DashScope API is only a means to handle call requests.

@NihalGazi , your template is with problem and we cant create videos

@NihalGazi , your template is with problem and we cant create videos

@gusta0729
The inconvenience is regretted, and sincerely apologized for. I have a faster no-queue alternative to it, inorder to make it up meanwhile. Please refer to here