John's picture

John PRO

Keltezaa

AI & ML interests

None yet

Recent Activity

updated a model 1 day ago
Keltezaa/Inni_and_Ass_pfbkFLUX
updated a model 1 day ago
Keltezaa/female-masturbation-fingering
View all activity

Organizations

None yet

Keltezaa's activity

replied to their post about 12 hours ago
view reply

What are being done about
("ZeroGPU worker error", "GPU task aborted")
gradio.exceptions.Error: 'GPU task aborted'

This happen intermittently but often enough to annoy me.
out of 5 attempts I would get this error 3 times, sometimes after the first attempt the 2nd render would be successful,

but if not I reduce "steps" to compensate, allowing the same seed to generate,

Why did it fail?
Why would it sometimes succeed on the 2nd attempt without changing any settings,
Lora weights, steps, CFG, prompt or seed ?
Can this be fixed or can the GPU time lost because of the error be refunded?

Thank you in advance for your assistance.

posted an update 15 days ago
view post
Post
1125
Why does all the Text-to-image models running on HF Inference API fail and report fail with the error
"Model strangerzonehf/Neon-Impressionism-Flux does not exist"

It used to work last month.
  • 2 replies
ยท
replied to their post 24 days ago
view reply

@not-lain
here is a snip from my current code.
@spaces.GPU(duration=60)
def run_lora(prompt, cfg_scale, steps, randomize_seed, seed, width, height, progress=gr.Progress(track_tqdm=True), **kwargs)

If I go below 45sec I get GPU Fail, and the render is aborted , AND the usage is NOT Refunded.
So I stick with 60sec, becuase If you increase the model weight or the steps the generation time cost more (obvious).
So to limit the amount of failures, I stick to 60.

replied to their post 24 days ago
view reply

@reach-vb , @julien-c -

Something I just noticed since I was paying attention to the ZeroGPU recovery.
Inference Credits. ? .... $0.03 / $2.00 - this used to be something else. something like
Inference Usage xxxx / 2000 Uses .
Please Explain?
Why have this now changed?
Am I going to get billed for this now too?

** Edit **
Never mind I found the blog post.. and reading it.

I notice something else int the blog it is mentioned that Pro users get "20x higher limits" currently when I run out of time I get a popup that says if I subscribe to a pro account I get 5x.

Here is a direct Quote from the blog post. :

"Subscribe to the Hugging Face PRO plan to get access to Inference credits, ZeroGPU, Spaces Dev Mode, 20x higher limits , and more.

huh..? Why is this different, I am confused.

Also, If I disable the other Inference API's It wont be used? So I wont get billed?
I dont have API Keys or accounts with any of the listed providers, so naturally I disabled them.

replied to their post 24 days ago
view reply

@John6666 , To you $9 may not be much but for other countries like myself where the exchange rate is deplorable (ZAR16 to $1) it is expensive. So please take such matters into account. When it comes to Value I pay 16 to 1 more than others. Take any "cheap" item you own, multiply its cost by 16. Will you still buy that Item?

All I ask Is for a better Recovery rate that will make it affordable to Us to use
Sincerely... me

posted an update 25 days ago
view post
Post
1087
Other "Pro" members & staff.
I am seriously considering of canceling my Pro subscription, unless this matter can be addressed.

I subscribed to the Pro package for the use of the ZeroGPU, but the current recovery rate is not worth the money. I am supposed to get 5x more usage and or recovery.

By my rough calculation the current recovery rate for GPU time spend is 18min per every 60sec of GPU usage. One of my "T-T-image" Spaces use about 45-50 seconds for each render. that gives 25-28 images
At the current recover rate that is 2 images per hour, limited to 24 images per day.

I may be wrong here or there may be something in my app.py code that is messing with with the usage time. currently it is set "@spaces.ZeroGPU(Duration=60)" If I reduce this I get an error if my render takes more time. (GPU task aborted, or error)

The second thing that bothers me a bit is that some Errors, or failed image generation do not refund the usage. So if an image fails due to whatever error. It still gets added to the usage and as mentioned before recovers very slow. So at the end of the day If I get 15 usable images I consider myself lucky. As we all know, that you can prompt like a "boss" and the image dont always come out as you imagined or hoped.

Thank you for taking the time to read my *RANT* - But I sincerely hope that We as paying users can get more value for money.

Regrds.
ยท