Thank you SO much! I’ve been trying to run some LoRA models on Colab, but I couldn’t find any notebooks specifically for my use case. You have saved my life 🙏
Asked about how to get this running on Mac in the last video, and you deliver the next day. Insane value. Been having fun with this, you've earned my sub and bell notification
Wow, this is amazing! I've always wanted to try Stable Diffusion XL, but I don't have a powerful PC or GPU. Thanks to this tutorial, I can now use it on Kaggle for free, and even with ControlNet and LoRA models. This is a game-changer for image generation AI. I can't wait to see what I can create with this. Great job, SECourses! 👏👏👏
Hello. This video was also very good, just like the previous one where you taught how to train the model. Now if you can make a video of a collab that merges the models, this way we can merge the model created with another model from civitai, for example. Thank you, congratulations on the work.
Thank you very much for your work, I stopped the creation of my visual novel due to lack of conditions to use stable diffusion on a pc and now I can go back... However, I would like to know if there is any way to keep the models folder in the drive so I don't have to upload my models and loras every time?
Super cool video! It's a ton of fun to experiment with Focus on Colab. I'm running into the issue that the 12 GB of RAM are filling up pretty quickly on the virtual machine. Is there any way around this except for upgrading to a paid version of Colab?
why that didn't create your image? as far as im concerned, in the lora training video, you set the prompt to "tom cruise man" in order to generate images of yourself!
It's been stuck on "Error checking version for torchsde: No package metadata was found for torchsde Installing requirements" for about 10 minutes... what's up with that? should I just be patient?
Is it possible to run this via android or iPhone? And all settings, loras and whatnot are lost every time? Or is it possible to reload presets, like the local version?
your guides are fantastic. but i have an error to tackle. when i open the radio page and prompt to produce an image, it loads for a few seconds, and then writes "Error".
Is there a way to directly import LoRA from HF or other sources into the Colab environment w/o downloading it first on my local machine and then uploading it to Colab?
Maybe…. I suppose if you have the hugging face model url you might be able to use the terminal in the colab to download the model directly (I haven’t tested this yet)
oh, I have no idea actually. It would depend on the number of images you generate and download I suppose. The use of the system shouldn't be too intensive though since most of the processing is remote.
@@allyourtechai lora is low size so just have to take sometime...but if is checkpoint at least the size is 2gb left...it is impossible to upload for this🤣
@@allyourtechai I have the same issue. And I notice that the one in this tutorial looks exactly like Tom Cruise and nothing really like you. However when you did this locally in your other video, it looked exactly like you
✨ Support my work on Patreon: www.patreon.com/allyourtech
💻My Stable Diffusion PC: kit.co/AllYourTech/stable-diffusion-build
Thank you SO much! I’ve been trying to run some LoRA models on Colab, but I couldn’t find any notebooks specifically for my use case. You have saved my life 🙏
You're so welcome!
Asked about how to get this running on Mac in the last video, and you deliver the next day. Insane value. Been having fun with this, you've earned my sub and bell notification
Thank you so much for the support! Definitely let me know how I can help in the future !
Wow, this is amazing! I've always wanted to try Stable Diffusion XL, but I don't have a powerful PC or GPU. Thanks to this tutorial, I can now use it on Kaggle for free, and even with ControlNet and LoRA models. This is a game-changer for image generation AI. I can't wait to see what I can create with this. Great job, SECourses! 👏👏👏
Awesome to hear, thank you so much for watching and leaving a comment. Let me know if I can help out with anything!
It looks like it was exactly what I asked for in the comment on your previous video. Thanks boy.
You got it!
Absolutely fantastic, subscribed.
best tutorial so far
Hello. This video was also very good, just like the previous one where you taught how to train the model. Now if you can make a video of a collab that merges the models, this way we can merge the model created with another model from civitai, for example. Thank you, congratulations on the work.
I’ll see what I can do!
But Google Colab doens't like any kind of UI going on..either its always out of memory or you get disconnected out of GPU backend
Thank you very much for your work, I stopped the creation of my visual novel due to lack of conditions to use stable diffusion on a pc and now I can go back... However, I would like to know if there is any way to keep the models folder in the drive so I don't have to upload my models and loras every time?
Super cool video! It's a ton of fun to experiment with Focus on Colab. I'm running into the issue that the 12 GB of RAM are filling up pretty quickly on the virtual machine. Is there any way around this except for upgrading to a paid version of Colab?
Hi, thank you very much for the video! I just have a question, can those generated images be sold?
In most cases, yes that isn't a problem.
@@allyourtechai Thank you very much, friend!
why that didn't create your image?
as far as im concerned, in the lora training video, you set the prompt to "tom cruise man" in order to generate images of yourself!
It's been stuck on "Error checking version for torchsde: No package metadata was found for torchsde
Installing requirements" for about 10 minutes... what's up with that? should I just be patient?
Is it possible to run this via android or iPhone? And all settings, loras and whatnot are lost every time? Or is it possible to reload presets, like the local version?
your guides are fantastic. but i have an error to tackle. when i open the radio page and prompt to produce an image, it loads for a few seconds, and then writes "Error".
Have you adjusted any advanced settings or anything by chance? Can you verify that the t4 colab instance still shows as running?
@@allyourtechai there is a chance it has run out of GPU allowance. Do you have the pro version?
Thank u very very much for this video
Is there a way to directly import LoRA from HF or other sources into the Colab environment w/o downloading it first on my local machine and then uploading it to Colab?
Maybe…. I suppose if you have the hugging face model url you might be able to use the terminal in the colab to download the model directly (I haven’t tested this yet)
There's no Fooocus file folder where I can upload my lora. Why is this so and what can I do to fix this?
There should be a models directory which then has a lora folder. Let me know if you aren't able to find it and i'll find the exact path.
I'm on a data capped internet plan, how much data do you think is used in a daily 4 hour session in Collab?
oh, I have no idea actually. It would depend on the number of images you generate and download I suppose. The use of the system shouldn't be too intensive though since most of the processing is remote.
Hello, is it normal to see an error saying no Nvidia found when I try to run it? I am not using the local link but the other
When you launch the colab, did it show you using a t4 instance in the upper right hand corner?
when i try to upload other models like sdxl it stays loading in red circle. Help?
Same here
Same
hii thanks seems I have ran out of GPU, is there anyway to resolve this issue? thanks!
Unfortunately not. Although, I am building an easy LoRA trainer at PixelDojo to take care of this issue.
@@allyourtechai thanks for the advice. At least I know it’s not my fault 😅
where is the lora file?
I downloaded Lora and writing Lora file is corrupted
this only have one bad....is cannot download the lora on link in civitai...need to manual upload have to take so much time...
It does add extra time for sure
@@allyourtechai lora is low size so just have to take sometime...but if is checkpoint at least the size is 2gb left...it is impossible to upload for this🤣
Great, I don't have to bother with my Mac Mini 👍
I tried it, but the trigger keyword makes the image look exactly like the celebrity...not like my trained lora
how many training images did you use?
@@allyourtechai about 30-50
@@allyourtechai I have the same issue. And I notice that the one in this tutorial looks exactly like Tom Cruise and nothing really like you. However when you did this locally in your other video, it looked exactly like you
@@allyourtechai The loras don't work for me, could it be because fooocus only accepts LORAS SDLX AND NOT SD?......how can I put LORAS SD IN FOOOCUS
@@LeoWestonVFX guess it has something to do with the base model. in the video you are referring to, the base model was SDXL 1.0.
Colab has limited time.
can you do nsfw with this?
Yes you can
Is there a safety option to be turned off?
can you call it via python and an api