Видео

It has to be mentioned in the prompt

I can get things to work on generating an error: AssertionError: config must have _class_name=CausalVideoAutoencoder. I have no experience in programming... GPT takes me on this to the Illusiana swamp :(...

There is some issue with the torch cuda installed, did you installed it before the requirements.txt installation

@@Aininja94 Yes, I was looking at github and there are other people with the same problem but I didn't find any solution

@@Aininja94 I've reinstalled it three times, following some suggestions, i found regarding similar issues on GitHub, but I still haven't been able to fix it. There's a discussion there about the same problem, but so far, I haven't found a definitive solution.

@@Aininja94 I've tried reinstalling it three times, following some of the issues reported on GitHub, but none of the solutions have worked so far. There's a thread discussing the same problem, but no one seems to have found a fix yet.

Is it happening in Ruined fooocus Or you just tried in other interface

Stable Diffusion models are used inside fooocus to generate a image. Stable diffusion is good, but flux is better. However flux is new, everyone is still figuring out how to get the best out of it

@@Aininja94dope content thanks for all

It should, I tried limiting the my vram for testing but it's not happening. I am waiting for someone commenting it worked 😅

@@Aininja94 oh ok will try and comment back, hope it doesn't blow up 😂

The switch for low vram if you saw in the video. Make sure to toggle it On. Yeh, it should not blow up. It would be terrible, this is the first time i did something with AI and python, trying to optimize some other project as well for low vram. I may start questing my self if bad happns

@@Aininja94 Update it works normally (a side from the usual gradio bugs and some times it crashes) after switching that toggle and yeah it is very impressive, thx for ur support and the tutorial I really appreciate that !

Great, it works on 6gb then.

You have to download for the first time only

@Aininja94 but i cant do it again can tell me what need to do

u have Instagram or any social media? ​@@Aininja94

I am not active on social media but you can msg on Instagram, you should find by searching ai_ninja94

Didn't get by the filtering content, the project took around 60gb space.

You have to add hugging face cli in the project. Then you have to login into hugging face using cli and you hugging face token. Then you should be able to download the file from hugging face

Hope you cleared the hugging face download issue, I thought I should be explaining in a short video

Not implemented yet, it has been couple of days implemented in swarm ui

Safe means this program has no intension of breaking a computer, should be able to run with 4060 card. People used NF4 model on 6gb cards. I am making a follow-up video on models. It will include info about models to be used for smaller card to bigger card. However you can use the model I showed in this video.

@Aininja94 last time i used forge ui mg ssd got currpted while using forge ui so I'm afraid of it

@@Aininja94thank you for all the effort! ❤

My graphic drivers were got deleted or something. It wasn't working at all. It happnd once, but a reinstall worked

@@Aininja94 thats survivable, i was thinking too for an ssd to get corrupted the chances are the same using any AI UI or DAW or perhaps premier pro or DaVinci resolve, if you write and read high speed it could get eventually corrupted very small chance still

Thank for adding the info. I am make a follow up video for models

If you knew...

netninja.in/

Gguf should work, what gpu do you have? 11gb is an odd one

@@Aininja94 GTX 1080ti

1080ti works with gguf i read on somewhere on reddit. What error did you get? Post it on the issue section so it can be checked.

@@Aininja94 w GGUF ok, but w Dev "can't load model" and Ruined shutdown

Dev model takes more vram. The code might need some optimisation to work. Dev model works on 12gb 3060 card. But even there it's 100% occupied. I think there is no room due to the 1gb with the current code. To test it accurately I think you have to try comfy ui once. For dev model, if it's working on 11gb

You are welcome

Thanks for mentioning, happy creating :)

Working on a video on the same topic

@@Aininja94 Great! I am looking forward to your new video. Thank you. I have subscribed to your channel and become your loyal fan.

No but you can try, one of my viewer ran flux model on 4gb vram. It took 20 min to generate a image but it worked

Your welcome

Python seems not there in your computer. If it's there. Then CMD is not able to find python installed on you computer. You have to add executable in the environment variable of windows

@@Aininja94 how do i do that

Add python to path realpython.com/add-python-to-path/

Your welcome, there is a video almost done, if you are looking for AI video application Will be upload in an hour

@@Aininja94 BET

If you are talking about inpaint like fooocus then it is different. The inpaint you will find in the power up section (it's a check box)

It has comfy Ui stuff inside.

@@Aininja94 ))) I was beginning to think that

The code will remain the same for the functionality then why not copy and paste it, be it from comfy ui Or any other application.

I have already tried using the command pip install -e

The command is pip install -e MagivQuill/LLaVA/ Please check the description and copy from there

@@Aininja94 Thanks, I have done this and I'm not getting this error anymore. I am now getting the following error: (base) c:\Users\Dan\MagicQuill>set CUDA_VISIBLE_DEVICES=0 && python gradio_run.py Traceback (most recent call last): File "c:\Users\Dan\MagicQuill\gradio_run.py", line 5, in <module> from gradio_magicquill import MagicQuill ModuleNotFoundError: No module named 'gradio_magicquill'

After keeping the lora in the lora folder, click the refresh button in fooocus. Also, you lora should be a dot safetensor file

If you are talking about inpaint like fooocus then it's different. The inpaint you will find in the power up section (it's a check box)

I was working for marketing agency, few b2b clients

In the right section, click the info tab. You will see the image info of the current generated image. It has the seed id

No docker file was there

@Aininja94 without docker this a nightmare especially for windows users

Well, I have idea about dockerr and the orchestration stuff but never used any. Setting up this is a challenging as there are errors posted in the comments 😖

Hi buddy, can you make for us a video about creating and training a flux lora model but on google colab please?

I am 25% done making video on flux lora but on personal computer. Making it on colab, I think will have to be a seperate video.

@@Aininja94 Alr, we'll wait for ya buddy, I liked your explanations!

Will add a file for colab and let you know

Click this link: colab.research.google.com/drive/1vvXTwDUJoIT4trsiJtdBrPKOjjFu0X74?usp=sharing

@Aininja94 thank you bro 🙏😊

When creating Conda environment, did you added python? conda create -n env python==3.10 Like this.

@@Aininja94 yes look F:\MagicQuill>conda create -n env python==3.10 WARNING: A conda environment already exists at 'C:\Users\Júnior3D\anaconda3\envs\env' Remove existing environment? This will remove ALL directories contained within this specified prefix directory, including any other conda environments. (y/[n])?

I think the first time when you ran this command, python was not included hence pip is not recognised in the env. Try creat a different env Or press y for yes remove the existing environment and have a new environment with python

@@Aininja94 I'll try to redo everything from scratch again to see if it's ok

what's your name on discord

Some supporting modules like ip adapters and control net are not that great in flux yet.

Remove the torch from requirements.txt file. And install the torch from the command I have given in the description.

yup did that and thats what came up everything else worked fine untill i added the torch

I didn't try on colab, should be working on colab as well, just follow the steps properly.

the color palette not there? What browser you are using

thanks for the tip, changed the browser to chrome and the brush became active)

Did you clicked the check box in advance setting?

@Aininja94 yes, but now it worked by just keeping new download control net and deleting old control net in the folder

You can't miss the space in the left as shown in the video. Press 'tab' to leave proper space as shown in the video.

you have not installed the requirements.txt file. This gradio error will be solved by 'pip install gradio'

@@Aininja94 you have not installed the requirements.txt file. --- installed it((( This gradio error will be solved by 'pip install gradio' --- installed it (((

If you followed the video but still the error, then you must not be in a active conda environment.

@Thank you, I’ll track the error with conda

​@@Aininja94 I went through a small circle of hell, starting with the fact that I couldn't log in to (env ) - before reinstalling cuda version 11.8, and even a mountain of small errors, but with the help of gpt4 I managed it)))

Yes, for someone who prefer simple interface.