Torpy

Yes I can understand that some calls would be more expensive then others. But, compared between two of my billings I don't think anything changed with my system. It is a repeating system that kinda follows the same day-to-day routine. So I don't really understand why the price is so much different. It shouldn't. Maybe something changed but I don't know what and the current information (or am I missing something?) makes it sort of impossible to find it.

Did they maybe included some information about process time / pricing / ... in the received resulting array that I am missing?

Their calculations and lack of clear information. 700 calls wouldnt be worth 1.8$ compared to 17k calls that are worth 1.7$. And if so, they are unclear what makes them such high priced. I cant validate it with this kind of information which makes it highly suspecious.

I am not understanding this thing anymore.
Update from this morning (all keys have been deactivated since mid march):

Well I have a potato computer that I cant use for running models and stuff.

I never have used python or shell scripts. I am not an IT person. I m using different languages and techniques which also works. So my code uses HF to compute and let HF do all the heavy stuff. My code also uses Chat GPT api and such. I dont think I had a real exchange of viewpoints about technical topics. Most of the time I saw lots and lots of other users having questions and no one replying on them. But I must say that this option to post on your profile is a key changer.

Funny that you mention the lon/lat photos since I asked that question too and got a working result 🤣

Do I want to make my own models? Yes. Do I want to use all HF models and stuff? Yes. Do I expect to make worlds best and greatest stuff? No, it is just for fun. Not even for work. Do I want to pay hunderds or thousands euros a month for it? No. So I better pull the plug totally. Like I said: I pull out the plug and see if it would be interesting to plug it back in later. I dont want all the shizzle of trying to find a solution to change it all again next month. The new credit system is still WIP, and I think it is better for me to see what it will result into. Also maybe it is better to find a different provider, I dont know, I am not an IT or A.i. expert.

It might give me some time to reflect and reconsider my project. Maybe I never restart it and decides to do a different project with HF. Who knows ...

Well I am happy that the website is very useful for you. To be honest, I loved to play with the models of other people and discovering things and so on. I was like in a sandbox where you could have fun. I only used the PRO membership for their API services. With the new credit system: Last month everything was fine, but this month I reached the limit in 5 days. When they allowed to exceed the limit I saw the price rising with several cents per minute. Even when I turned off all allowed API services I was getting charged for "empty" API usage while only getting "your api is turned off" response. I also found out that there are many moments that you don't have a 1-1 ratio (api call to successfull result) which isnt helpfull.

Also getting help with my journey on HF was hard. I know I havent reached full potential. E.G.: I have no clue how to make my own text to text model and stuff. There are no usefull applied tutorials (read examples) and stuff... (except for one that got deleted a while ago)

it is not about the $9. I don't know what it is exactly, but I think the uncertainty of the total price and the high fail rate are contributing to it.

bug: If you turn off all interference providers, he will continue counting them in the price you need to pay.

This new system is really bad for the users (us/me). Instead of 200.000 api requests like before I am not even getting 5.000 api requests. Already hitting max usage. If you are planning to give us far less credit then we are actually paying for the membership, I would consider not taking the membership and just pay for the spent credits. It is way cheaper. Now I feel like I am paying over 4 times more then the provided services are worth.

Hey @John6666

I have been playing a bit with your space "safetensors_to_diffusers " and I noticed something remarkable.
I don't know if it has to be like this or not. Maybe you can help?

When I try to use your space it creates the HF models in my name but I noticed that:

• Sometimes they are empty (for flux / SD 3.5)
• When not empty, I have the structure and files like normal but when I try to run the on-site inference api (by using the modelcard) I get the error that the "Model ... does not exist". However when I try to curl the api-inference I found out that most of the time the model is loading. And in a few rare occasions it gives me a generated image. I can't make any HF-space out of them since HF thinks the model does not exists either.

I havent had the chance to try different models and different options from your space. But for the several ones I tried I reproduced these results. Is this something that sounds familiar? I kept the safetensor files beneath the 10GB limit as recommended by HF.

Oh about the python code:
To be honest, me & python arent really friends at this moment so I prefer to use a different language. I tried to run the lines in the terminal, installed all the things and got an error with already "from diffusers" so I am kinda postponing it till I followed a beginners course in python or maybe a beginners course in visual studio 2019. I guess I must be doing or forgetting something very simple/stupid when using this language/software. So I decided to do all the things differently ... almost for 2 years already ^^'