or should i just rent a runpod with someone with a beast card, train my models, then use them in aut

ZZono50 or should i just rent a runpod with someone with a beast card, train my models, ...

A

ashley•6/6/23, 1:45 PM

Yes

A

ashley•6/6/23, 1:46 PM

You don't need an 80GB beast though, 24GB is fine unless you want to use Adan D-Adapatation Optimizer, then you need at least 32GB

A

ashley•6/6/23, 1:47 PM

W&B can also send email or Slack notifications when your training completes

Tteachtaha Whats the minimum amount of dataset? Also, will it resume if I start it like in ...

F

Furkan Gözükara SECourses•6/6/23, 1:47 PM

i trained with 2 hours speech

Aashley W&B can also send email or Slack notifications when your training completes 😍

F

Furkan Gözükara SECourses•6/6/23, 1:48 PM

really this is 32 gb?

F

Furkan Gözükara SECourses•6/6/23, 1:48 PM

wow

FFurkan Gözükara SECourses wow

A

ashley•6/6/23, 1:49 PM

Only Adan D-Adaptation Optimizer, not the other D-Dadaptation optimizers, the others can all run on 24GB VRAM

A

ashley•6/6/23, 1:51 PM

Maybe its a bug in Dreambooth training for the DB extension, because someone is using the DB extension to train LoRA with Adan D-Adapt with just 8GB: https://civitai.com/articles/52/d8ahazard-dreambooth-extension-guide

Civitai | D8ahazard Dreambooth extension guide

FFurkan Gözükara SECourses sd 21 is really hard to train and get good results

R

Reach•6/6/23, 2:09 PM

So do you advice better to train with 512 x 512 pixel image and SD 1.5 model ?

RReach So do you advice better to train with 512 x 512 pixel image and SD 1.5 model ?

F

Furkan Gözükara SECourses•6/6/23, 2:12 PM

sd 1.5 much easier better

F

Furkan Gözükara SECourses•6/6/23, 2:12 PM

also using public models are even better

A

ashley•6/6/23, 2:12 PM

W&B also helps you compare 4 different trainings to determine the optimal average epoch so you don't overtrain

F

Furkan Gözükara SECourses•6/6/23, 2:12 PM

according to your aim

A

ashley•6/6/23, 2:13 PM

Well can be much more than 4, just I have only done 4 so far

F

Furkan Gözükara SECourses•6/6/23, 2:13 PM

W&B?

A

ashley•6/6/23, 2:13 PM

Weights & Balances (wandb.ai)

F

Furkan Gözükara SECourses•6/6/23, 2:14 PM

i see paid service

A

ashley•6/6/23, 2:14 PM

You can enter your API key into Kohya_ss and EveryDream2trainer, not yet supported by Dreambooth extension but I logged a feature request to add it

FFurkan Gözükara SECourses i see paid service

A

ashley•6/6/23, 2:14 PM

Free for personal and academic use.

FFurkan Gözükara SECourses also using public models are even better

R

Reach•6/6/23, 2:14 PM

okay so can I use your 512 x 512 Reg class man images of your Patreon for training ?

A

ashley•6/6/23, 2:14 PM

I am not paying, I am using it for free

RReach okay so can I use your 512 x 512 Reg class man images of your Patreon for traini...

F

Furkan Gözükara SECourses•6/6/23, 2:15 PM

yes. make sure that you are using the correct model

Aashley I am not paying, I am using it for free

F

Furkan Gözükara SECourses•6/6/23, 2:15 PM

nice

FFurkan Gözükara SECourses yes. make sure that you are using the correct model

R

Reach•6/6/23, 2:16 PM

Yes I will use 512x512 images and SD 1.5 model for training

A

ashley•6/6/23, 2:16 PM

I will maybe pay for it if I decide to use it commercially but just running experiments at the moment, so I have a personal account

RReach Yes I will use 512x512 images and SD 1.5 model for training

F

Furkan Gözükara SECourses•6/6/23, 2:51 PM

yep

F

frandagostino•6/6/23, 3:35 PM

Dear, how are you? Anyone tried training Textual Inversion on an Apple M2 at a reasonable training time?

Z

Zono50OP•6/6/23, 3:36 PM

generated samples using the adan adaption optimizer using learning rate of 1e-3 and 5-e5 and get this

F

frandagostino•6/6/23, 3:36 PM

I've a Macbook M2 16GB, is it viable or should I go with Colab/Runpod/etc?

ZZono50 generated samples using the adan adaption optimizer using learning rate of 1e-3 ...

A

ashley•6/6/23, 3:43 PM

I think the guy on D8ahazard Dreambooth who already answered you is the only person who has experience using this stuff properly

A

ashley•6/6/23, 3:43 PM

The new optimizers are too new, not many people know how to use them yet.

A

ashley•6/6/23, 3:44 PM

Especially when the release notes have incorrect instructions on how to use them

L

LeHeL•6/6/23, 3:55 PM

I want to make a software in PHP to call A1111 (or raw SD) in which to call through the software API.
What it should do is to generate a mix from 2 images. So I'm thinking of an img2img in which I give img as input and net control or swap as the second image.
In fact, what I want is to make some stylized images (like a poster) in which I can swap faces with an uploaded face.
Does anyone have an idea how I could do this through the API?

LLeHeL I want to make a software in PHP to call A1111 (or raw SD) in which to call thro...

A

ashley•6/6/23, 3:57 PM

visit the A1111 URL and add

/docs

/docs

to the end of it to see API documentation.
Also enable

--api

--api

in your launch command in

web-user.*

web-user.*

A

ashley•6/6/23, 3:58 PM

For img2img, the payload looks like this:

    sd_payload = {
        'init_images': [
            image_content
        ],
        'resize_mode': 0,
        'denoising_strength': payload['denoising_strength'],
        'image_cfg_scale': 0,
        'prompt': payload['positive_prompt'],
        'seed': -1,
        "sampler_name": payload['sampler'],
        'batch_size': payload['batch_size'],
        'n_iter': 1,
        'steps': payload['steps'],
        'cfg_scale': payload['cfg_scale'],
        'width': width,
        'height': height,
        'restore_faces': True,
        'tiling': False,
        'do_not_save_samples': False,
        'do_not_save_grid': False,
        'negative_prompt': payload['negative_prompt'],
        'eta': 0,
        's_churn': 0,
        's_tmax': 0,
        's_tmin': 0,
        's_noise': 1,
        'sampler_index': payload['sampler'],
        'include_init_images': False,
        'send_images': True,
        'save_images': False,
    }

    sd_payload = {
        'init_images': [
            image_content
        ],
        'resize_mode': 0,
        'denoising_strength': payload['denoising_strength'],
        'image_cfg_scale': 0,
        'prompt': payload['positive_prompt'],
        'seed': -1,
        "sampler_name": payload['sampler'],
        'batch_size': payload['batch_size'],
        'n_iter': 1,
        'steps': payload['steps'],
        'cfg_scale': payload['cfg_scale'],
        'width': width,
        'height': height,
        'restore_faces': True,
        'tiling': False,
        'do_not_save_samples': False,
        'do_not_save_grid': False,
        'negative_prompt': payload['negative_prompt'],
        'eta': 0,
        's_churn': 0,
        's_tmax': 0,
        's_tmin': 0,
        's_noise': 1,
        'sampler_index': payload['sampler'],
        'include_init_images': False,
        'send_images': True,
        'save_images': False,
    }

    sd_payload = {
        'init_images': [
            image_content
        ],
        'resize_mode': 0,
        'denoising_strength': payload['denoising_strength'],
        'image_cfg_scale': 0,
        'prompt': payload['positive_prompt'],
        'seed': -1,
        "sampler_name": payload['sampler'],
        'batch_size': payload['batch_size'],
        'n_iter': 1,
        'steps': payload['steps'],
        'cfg_scale': payload['cfg_scale'],
        'width': width,
        'height': height,
        'restore_faces': True,
        'tiling': False,
        'do_not_save_samples': False,
        'do_not_save_grid': False,
        'negative_prompt': payload['negative_prompt'],
        'eta': 0,
        's_churn': 0,
        's_tmax': 0,
        's_tmin': 0,
        's_noise': 1,
        'sampler_index': payload['sampler'],
        'include_init_images': False,
        'send_images': True,
        'save_images': False,
    }

    sd_payload = {
        'init_images': [
            image_content
        ],
        'resize_mode': 0,
        'denoising_strength': payload['denoising_strength'],
        'image_cfg_scale': 0,
        'prompt': payload['positive_prompt'],
        'seed': -1,
        "sampler_name": payload['sampler'],
        'batch_size': payload['batch_size'],
        'n_iter': 1,
        'steps': payload['steps'],
        'cfg_scale': payload['cfg_scale'],
        'width': width,
        'height': height,
        'restore_faces': True,
        'tiling': False,
        'do_not_save_samples': False,
        'do_not_save_grid': False,
        'negative_prompt': payload['negative_prompt'],
        'eta': 0,
        's_churn': 0,
        's_tmax': 0,
        's_tmin': 0,
        's_noise': 1,
        'sampler_index': payload['sampler'],
        'include_init_images': False,
        'send_images': True,
        'save_images': False,
    }

L

LeHeL•6/6/23, 3:58 PM

Yeah, I know that but dunno if I can add face swap addon az API call.

A

ashley•6/6/23, 3:58 PM

where image_content is base64 encoded binary content of the image

L

LeHeL•6/6/23, 3:59 PM

tx

A

ashley•6/6/23, 3:59 PM

to add ControlNet:

sd_payload['alwayson_scripts'] = {
                'controlnet': {
                    'args': [
                        {
                            'module': 'canny',
                            'model': 'control_sd15_canny [fef5e48e]',
                            'weight': 1,
                            'resize_mode': 'Crop and Resize',
                            'lowvram': False,
                            'processor_res': 512,
                            'threshold_a': 75,
                            'threshold_b': 75,
                            'guidance': 1,
                            'guidance_start': 0,
                            'guidance_end': 1,
                            'control_mode': "Balanced",
                            'pixel_perfect': False
                        }
                    ]
                }
            }

sd_payload['alwayson_scripts'] = {
                'controlnet': {
                    'args': [
                        {
                            'module': 'canny',
                            'model': 'control_sd15_canny [fef5e48e]',
                            'weight': 1,
                            'resize_mode': 'Crop and Resize',
                            'lowvram': False,
                            'processor_res': 512,
                            'threshold_a': 75,
                            'threshold_b': 75,
                            'guidance': 1,
                            'guidance_start': 0,
                            'guidance_end': 1,
                            'control_mode': "Balanced",
                            'pixel_perfect': False
                        }
                    ]
                }
            }

sd_payload['alwayson_scripts'] = {
                'controlnet': {
                    'args': [
                        {
                            'module': 'canny',
                            'model': 'control_sd15_canny [fef5e48e]',
                            'weight': 1,
                            'resize_mode': 'Crop and Resize',
                            'lowvram': False,
                            'processor_res': 512,
                            'threshold_a': 75,
                            'threshold_b': 75,
                            'guidance': 1,
                            'guidance_start': 0,
                            'guidance_end': 1,
                            'control_mode': "Balanced",
                            'pixel_perfect': False
                        }
                    ]
                }
            }

sd_payload['alwayson_scripts'] = {
                'controlnet': {
                    'args': [
                        {
                            'module': 'canny',
                            'model': 'control_sd15_canny [fef5e48e]',
                            'weight': 1,
                            'resize_mode': 'Crop and Resize',
                            'lowvram': False,
                            'processor_res': 512,
                            'threshold_a': 75,
                            'threshold_b': 75,
                            'guidance': 1,
                            'guidance_start': 0,
                            'guidance_end': 1,
                            'control_mode': "Balanced",
                            'pixel_perfect': False
                        }
                    ]
                }
            }

F

Furkan Gözükara SECourses•6/6/23, 4:05 PM

anyone have any idea of this error https://github.com/Mikubill/sd-webui-controlnet/issues/1573

GitHub

FileNotFoundError: [Errno 2] - works but result not displayed in gr...

Is there an existing issue for this? I have searched the existing issues and checked the recent builds/commits of both this extension and the webui What happened? Just made a fresh installation The...

LLeHeL I want to make a software in PHP to call A1111 (or raw SD) in which to call thro...

F

Furkan Gözükara SECourses•6/6/23, 4:06 PM

api is something very less covered. never tried personally but you should be able to make

FFurkan Gözükara SECourses api is something very less covered. never tried personally but you should be abl...

A

ashley•6/6/23, 4:10 PM

Yes he can do it, I have done it, also done for Dreambooth

A

ashley•6/6/23, 4:11 PM

A1111 API is more stable than Dreambooth API, the Dreambooth API to get status is broken

A

ashley•6/6/23, 4:11 PM

But everything Dreambooth is unstable

F

Furkan Gözükara SECourses•6/6/23, 4:12 PM

if i needed it in a business

F

Furkan Gözükara SECourses•6/6/23, 4:12 PM

i would use the script diredtly

F

Furkan Gözükara SECourses•6/6/23, 4:12 PM

not via api of auto1111

A

ashley•6/6/23, 4:15 PM

Yeah I think thats what Fast Stable Diffusion by TheLastBen does, and thats what RunPod use in their ridiculously expensive dreambooth API endpoint.

Aashley Yeah I think thats what Fast Stable Diffusion by TheLastBen does, and thats what...

F

Furkan Gözükara SECourses•6/6/23, 4:19 PM

do you have any idea regarding this : https://github.com/Mikubill/sd-webui-controlnet/issues/1573

GitHub

FileNotFoundError: [Errno 2] - works but result not displayed in gr...

Is there an existing issue for this? I have searched the existing issues and checked the recent builds/commits of both this extension and the webui What happened? Just made a fresh installation The...

F

Furkan Gözükara SECourses•6/6/23, 4:19 PM

this happened on a patreon supporter

or should i just rent a runpod with someone with a beast card, train my models, then use them in aut

Similar Threads