This do to dataset size ?
This do to dataset size ?
Fi think it is due to extension bug
Fi opened a thread
Fbut this machine consumes a lot credits
HThis is why I am looking at an alternative to the extension for training.
I am not good enough with code to keep up with fixes
I am not good enough with code to keep up with fixes
Ftrue 
Di have been using stable tuner for on vast or auto1111 on vast for training, currently both repos are broken for my setup
Dcan anyone help me translate the google colab kohya notebook to a jupyter notebook i can run on vast?
HStabletuner is broken? Or cast giving issues?
HI was looking at everydream next possibly on Runpod
Di need to get pull stable tuner local to try the notebook export but dev mentioned in his discord he needed to fix something about notebooks
Di have a working notebook from stable tuner, i wish i could just add in offset noise
Dis that possible to just add a little code to change the noise offset to my notebook?
Fwhy dont you use dreambooth?
F
Fon automatic1111
Fit works great
Dit works good for lora on my 1080, but i want to use vast to not tie up my comp , and also to do normal finetune not lora
Di have been able to run auto1111 on vast up until now i get error
DTraceback (most recent call last):
File "/workspace/sd/launch.py", line 361, in <module>
start()
File "/workspace/sd/launch.py", line 356, in start
webui.webui()
File "/workspace/sd/webui.py", line 240, in webui
app.add_middleware(GZipMiddleware, minimum_size=1000)
File "/workspace/env/lib/python3.10/site-packages/starlette/applications.py", line 135, in add_middleware
raise RuntimeError("Cannot add middleware after an application has started")
RuntimeError: Cannot add middleware after an application has started
[11:57 AM]
trying to run auto1111 on vast , its always worked before , but havent done it in a while ....... it loads the model weights , creates the remote gradio, all looks like usual , but then it adds on this code and kills the port
File "/workspace/sd/launch.py", line 361, in <module>
start()
File "/workspace/sd/launch.py", line 356, in start
webui.webui()
File "/workspace/sd/webui.py", line 240, in webui
app.add_middleware(GZipMiddleware, minimum_size=1000)
File "/workspace/env/lib/python3.10/site-packages/starlette/applications.py", line 135, in add_middleware
raise RuntimeError("Cannot add middleware after an application has started")
RuntimeError: Cannot add middleware after an application has started
[11:57 AM]
trying to run auto1111 on vast , its always worked before , but havent done it in a while ....... it loads the model weights , creates the remote gradio, all looks like usual , but then it adds on this code and kills the port
Di found a thread about this error and people mention changing fast api
Dthe requirements.txt has correct version
Di tried pip reinstall fast api and still get error
HHave you noticed a difference in results using stabletuner vs a1111 for training ?
Dits hard to say , some people in stable tuner mention its better quality, im not sure ...... for me the notebook runs more reliably and quicker setup than installing auto1111 to vast so thats why i used it
Di imagine its both using kohya and should be about same output right?
Di wish i could just learn how to adapt kohya code direct to jupyter notebook
Dshould be as simple as copying the colab code in to notebook right?
Di ran the first cell and got error, i think its because i need to install pytorch or some other depdndency first
Dmaybe i should try running the stabletuner notebook first , to install dependencies, and then switch to kohya copy paste
Dso basically what stable tuner will do is take your dataset and creates a payload zip, and an ipynb notebook, u simply upload it to your instance , unzip the archive and run the notebook cells
Di will say that lora using offset noise i did on my 1080 seems to be really good , its hard to say how good on its own though because i merged it back in to my finetune merge im building (which is a bunch of finetunes ive made merged together)
Djust this one dreambooth with offset noise blended in seems to improve the contrast of the whole model
HOo, I didn’t know stabletuner was using kohya.
I couldn’t get that to work personally yet.
I haven’t had time to really get into finetuning yet but that’s next for me.
I couldn’t get that to work personally yet.
I haven’t had time to really get into finetuning yet but that’s next for me.
HYou are a little farther on your journey
Dive been in to it for a while , 1 step forward 2 steps back it seems ....... the whole space makes me wish i learned basic stuff along time ago
Dpretty sure khoya is the main source of the training code that ST and auto1111 use , someone correct me if im wrong
Di would just use colab but i dont even want to find out what a compute credit is, i imagine its probably 4x as expensive as vast
Din the beginning it seemed like all the SD communties were more active and helpful, i came here from reddit , video post
HYeah I feel that. I think all the Knowlege you’ve acquired will pay off as the tools we use become more stable.
Dr Fs tutorials are the only ones that are thorough enough for me to continue learning. Most vids are just whatever the hot topic is at the time.
Dr Fs tutorials are the only ones that are thorough enough for me to continue learning. Most vids are just whatever the hot topic is at the time.
Di will check out the videos today ....... this new style transfer controlnet looks really cool
Dits so insane what you can do with img2img and controlnet
Dive been doing stuff where i start with my photography as img2img and just get lost in layers of control nets using variations of photos and img2img and just lose track of what even goes in to an image so many different steps were taken
Dstyle transfer looks like a cool way to make use of everything youve done in past to influence stuff in future
TIs anyone having CUDA problems on runpod? I noticed Dr. Gozukara's pod template version in the tutorial videos is web-automatic-1.5.15 but now pods only display the new version web-automatic-2.1.18
TThe latest tutorial isn't working on these, I think
TFreshly updated actually, just a heads up
TOK, got it working now with the latest of Dr. Gozukara's tutorials. It's pretty straightforward actually, just create a new runpod template as follows:
Fi also updated the gist file accordingly to new template