zelenioncode/dreambooth_sdxl | Run with an API on Replicate

zelenioncode / dreambooth_sdxl

Cold

Public
155 runs
A100 (80GB)

Iterate in playground

Run with an API

Playground API Examples README Versions

Input

gender

string

Gender of person in training photo ( woman or man )

Default: "woman"

name_model

string

Shift + Return to add a new line

Give name for your .safetensors model

Default: "Dreambooth_sdxl"

send_to_huggingface

boolean

Send folder have .safetensors model direct to your huggingface account

Default: false

token_huggingface

string

Shift + Return to add a new line

If you use huggingface, enter your API TOKEN

Default: "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"

repo_id_huggingface

string

Shift + Return to add a new line

If you use huggingface, enter repo_id to one of your project

Default: "WGlint/SafetensorsFromReplicate"

folder_huggingface

string

Shift + Return to add a new line

Enter a path to download your .safetensors model. Default = ./

Default: ""

input

string

Shift + Return to add a new line

Direct link download with training picture (Only .zip file and picture in 1024px/1024px !)

Default: "https://huggingface.co/WGlint/SafetensorsFromReplicate/resolve/main/input.zip"

repeat_input

integer

(minimum: 1, maximum: 1000)

Repeat of time GPU look training data ( e.g. 15 pictures * 100 repeat = 1500 steps )

Default: 100

use_class_reg

boolean

Use regulat classification picture

Default: false

repeat_class_reg

integer

(minimum: 1, maximum: 1000)

Repeat of time GPU look class reg picture ( e.g. 5000 pictures * 2 repeat = 10000 steps cache latents )

Default: 1

class_reg

string

Shift + Return to add a new line

Direct link download for regular classification picture ( Default = class image of gender you use )

Default: ""

model_sdxl

string

Choice a model pretrained can run for SDXL training with dreambooth

Default: "Stable Diffusion XL"

num_cpu_threads_per_process

integer

(minimum: 1, maximum: 10)

Number CPU thread use with accelerate module

Default: 4

resolution

string

Shift + Return to add a new line

Resolution of your training picture data. WARNING ! Write in this formet : width,height ( e.g. 1024,1024 )

Default: "1024,1024"

vae

string

Shift + Return to add a new line

VAE use for create model training

Default: "stabilityai/sdxl-vae"

lr_scheduler_num_cycles

integer

(minimum: 1, maximum: 1000)

Num learning rate cycles for your training

Default: 1

max_data_loader_n_workers

integer

(minimum: 0, maximum: 100)

Maximun data loader for n workers you set

Default: 0

learning_rate_te1

number

Value for learning_rate te1

Default: 0.00001

learning_rate_te2

number

Value for learning_rate te2

Default: 0.00001

learning_rate

number

Value for learning_rate

Default: 0.00001

lr_scheduler

string

Method use for learning rate scheduler

Default: "constant"

train_batch_size

integer

(minimum: 1, maximum: 64)

Select value for device max train step and speed the generation, WARINING ! High value = High value to have CUDA Memory

Default: 1

max_train_steps

integer

(minimum: 0, maximum: 25000)

Number of step you want for your training, and in average 1000 steps = 10 minutes

Default: 3000

save_every_n_epochs

integer

(minimum: 1, maximum: 64)

Number of epochs model you want

Default: 1

mixed_precision

string

Select if you want to use miwed precision

Default: "fp16"

save_precision

string

Select if you want to use save precision

Default: "fp16"

optimizer_type

string

Select a optimiser type

Default: "AdaFactor"

scale_parameter

boolean

Use scale parameter

Default: false

relative_step

boolean

Use relative step

Default: false

warmup_init

boolean

Use warmup init

Default: false

weight_decay

number

Give a float value for weight decay

Default: 0.01

bucket_reso_steps

integer

(minimum: 1, maximum: 1000)

Give a int value for bucket reso steps

Default: 64

save_every_n_steps

integer

(minimum: 1, maximum: 5)

Number of .safetensors model you want, if you select 2 with 2000 max train steps, you well get 2 .safetensors. 1 with 1000 steps and 1 with 2000 steps

Default: 1

noise_offset

number

Give a float value for noise offset

Default: 0

max_grad_norm

number

Give a float value for max grad norm

Default: 0

cache_latents_to_disk

boolean

Default: true

cache_latents

boolean

Default: true

mem_eff_attn

boolean

Default: true

gradient_checkpointing

boolean

Default: true

full_fp16

boolean

Default: true

xformers

boolean

Default: true

bucket_no_upscale

boolean

Default: true

no_half_vae

boolean

Default: true

train_text_encoder

boolean

Default: true

learning_rate_te1_bool

number

value for learning rate te1 bool

Default: 0.000003

learning_rate_te2_bool

number

value for learning rate te2 bool

Default: 0

Run this model in Node.js with one line of code:

npx create-replicate --model=zelenioncode/dreambooth_sdxl

or set up a project from scratch

Install Replicate’s Node.js client library:

npm install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import and set up the client:

import Replicate from "replicate";
import fs from "node:fs";

const replicate = new Replicate({
  auth: process.env.REPLICATE_API_TOKEN,
});

Run zelenioncode/dreambooth_sdxl using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "zelenioncode/dreambooth_sdxl:e90767fffa851f984703defc193b1861d3b6d79acbcf1a798a3b9a1dac0bc176",
  {
    input: {
      vae: "stabilityai/sdxl-vae",
      input: "http",
      gender: "woman",
      xformers: true,
      class_reg: "",
      full_fp16: true,
      model_sdxl: "Stable Diffusion XL",
      name_model: "Dreambooth_sdxl",
      resolution: "1024,1024",
      no_half_vae: true,
      warmup_init: false,
      lr_scheduler: "constant",
      mem_eff_attn: true,
      noise_offset: 0,
      repeat_input: 100,
      weight_decay: 0.01,
      cache_latents: true,
      learning_rate: 0.00001,
      max_grad_norm: 0,
      relative_step: false,
      use_class_reg: false,
      optimizer_type: "AdaFactor",
      save_precision: "fp16",
      max_train_steps: 3000,
      mixed_precision: "fp16",
      scale_parameter: false,
      repeat_class_reg: 1,
      train_batch_size: 1,
      bucket_no_upscale: true,
      bucket_reso_steps: 64,
      learning_rate_te1: 0.00001,
      learning_rate_te2: 0.00001,
      token_huggingface: "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",
      folder_huggingface: "",
      save_every_n_steps: 1,
      train_text_encoder: true,
      repo_id_huggingface: "WGlint/SafetensorsFromReplicate",
      save_every_n_epochs: 1,
      send_to_huggingface: false,
      cache_latents_to_disk: true,
      gradient_checkpointing: true,
      learning_rate_te1_bool: 0.000003,
      learning_rate_te2_bool: 0,
      lr_scheduler_num_cycles: 1,
      max_data_loader_n_workers: 0,
      num_cpu_threads_per_process: 4
    }
  }
);

// To access the file URL:
console.log(output.url()); //=> "http://example.com"

// To write the file to disk:
fs.writeFile("my-image.png", output);

To learn more, take a look at the guide on getting started with Node.js.

Install Replicate’s Python client library:

pip install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import the client:

import replicate

Run zelenioncode/dreambooth_sdxl using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "zelenioncode/dreambooth_sdxl:e90767fffa851f984703defc193b1861d3b6d79acbcf1a798a3b9a1dac0bc176",
    input={
        "vae": "stabilityai/sdxl-vae",
        "input": "http",
        "gender": "woman",
        "xformers": True,
        "class_reg": "",
        "full_fp16": True,
        "model_sdxl": "Stable Diffusion XL",
        "name_model": "Dreambooth_sdxl",
        "resolution": "1024,1024",
        "no_half_vae": True,
        "warmup_init": False,
        "lr_scheduler": "constant",
        "mem_eff_attn": True,
        "noise_offset": 0,
        "repeat_input": 100,
        "weight_decay": 0.01,
        "cache_latents": True,
        "learning_rate": 0.00001,
        "max_grad_norm": 0,
        "relative_step": False,
        "use_class_reg": False,
        "optimizer_type": "AdaFactor",
        "save_precision": "fp16",
        "max_train_steps": 3000,
        "mixed_precision": "fp16",
        "scale_parameter": False,
        "repeat_class_reg": 1,
        "train_batch_size": 1,
        "bucket_no_upscale": True,
        "bucket_reso_steps": 64,
        "learning_rate_te1": 0.00001,
        "learning_rate_te2": 0.00001,
        "token_huggingface": "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",
        "folder_huggingface": "",
        "save_every_n_steps": 1,
        "train_text_encoder": True,
        "repo_id_huggingface": "WGlint/SafetensorsFromReplicate",
        "save_every_n_epochs": 1,
        "send_to_huggingface": False,
        "cache_latents_to_disk": True,
        "gradient_checkpointing": True,
        "learning_rate_te1_bool": 0.000003,
        "learning_rate_te2_bool": 0,
        "lr_scheduler_num_cycles": 1,
        "max_data_loader_n_workers": 0,
        "num_cpu_threads_per_process": 4
    }
)
print(output)

To learn more, take a look at the guide on getting started with Python.

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Run zelenioncode/dreambooth_sdxl using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

curl -s -X POST \
  -H "Authorization: Bearer $REPLICATE_API_TOKEN" \
  -H "Content-Type: application/json" \
  -H "Prefer: wait" \
  -d $'{
    "version": "zelenioncode/dreambooth_sdxl:e90767fffa851f984703defc193b1861d3b6d79acbcf1a798a3b9a1dac0bc176",
    "input": {
      "vae": "stabilityai/sdxl-vae",
      "input": "http",
      "gender": "woman",
      "xformers": true,
      "class_reg": "",
      "full_fp16": true,
      "model_sdxl": "Stable Diffusion XL",
      "name_model": "Dreambooth_sdxl",
      "resolution": "1024,1024",
      "no_half_vae": true,
      "warmup_init": false,
      "lr_scheduler": "constant",
      "mem_eff_attn": true,
      "noise_offset": 0,
      "repeat_input": 100,
      "weight_decay": 0.01,
      "cache_latents": true,
      "learning_rate": 0.00001,
      "max_grad_norm": 0,
      "relative_step": false,
      "use_class_reg": false,
      "optimizer_type": "AdaFactor",
      "save_precision": "fp16",
      "max_train_steps": 3000,
      "mixed_precision": "fp16",
      "scale_parameter": false,
      "repeat_class_reg": 1,
      "train_batch_size": 1,
      "bucket_no_upscale": true,
      "bucket_reso_steps": 64,
      "learning_rate_te1": 0.00001,
      "learning_rate_te2": 0.00001,
      "token_huggingface": "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",
      "folder_huggingface": "",
      "save_every_n_steps": 1,
      "train_text_encoder": true,
      "repo_id_huggingface": "WGlint/SafetensorsFromReplicate",
      "save_every_n_epochs": 1,
      "send_to_huggingface": false,
      "cache_latents_to_disk": true,
      "gradient_checkpointing": true,
      "learning_rate_te1_bool": 0.000003,
      "learning_rate_te2_bool": 0,
      "lr_scheduler_num_cycles": 1,
      "max_data_loader_n_workers": 0,
      "num_cpu_threads_per_process": 4
    }
  }' \
  https://api.replicate.com/v1/predictions

To learn more, take a look at Replicate’s HTTP API reference docs.

You can run this model locally using Cog. First, install Cog:

brew install cog

If you don’t have Homebrew, there are other installation options available.

Run this to download the model and run it in your local environment:

cog predict r8.im/zelenioncode/dreambooth_sdxl@sha256:e90767fffa851f984703defc193b1861d3b6d79acbcf1a798a3b9a1dac0bc176 \
  -i 'vae="stabilityai/sdxl-vae"' \
  -i 'input="http"' \
  -i 'gender="woman"' \
  -i 'xformers=true' \
  -i 'class_reg=""' \
  -i 'full_fp16=true' \
  -i 'model_sdxl="Stable Diffusion XL"' \
  -i 'name_model="Dreambooth_sdxl"' \
  -i 'resolution="1024,1024"' \
  -i 'no_half_vae=true' \
  -i 'warmup_init=false' \
  -i 'lr_scheduler="constant"' \
  -i 'mem_eff_attn=true' \
  -i 'noise_offset=0' \
  -i 'repeat_input=100' \
  -i 'weight_decay=0.01' \
  -i 'cache_latents=true' \
  -i 'learning_rate=0.00001' \
  -i 'max_grad_norm=0' \
  -i 'relative_step=false' \
  -i 'use_class_reg=false' \
  -i 'optimizer_type="AdaFactor"' \
  -i 'save_precision="fp16"' \
  -i 'max_train_steps=3000' \
  -i 'mixed_precision="fp16"' \
  -i 'scale_parameter=false' \
  -i 'repeat_class_reg=1' \
  -i 'train_batch_size=1' \
  -i 'bucket_no_upscale=true' \
  -i 'bucket_reso_steps=64' \
  -i 'learning_rate_te1=0.00001' \
  -i 'learning_rate_te2=0.00001' \
  -i 'token_huggingface="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"' \
  -i 'folder_huggingface=""' \
  -i 'save_every_n_steps=1' \
  -i 'train_text_encoder=true' \
  -i 'repo_id_huggingface="WGlint/SafetensorsFromReplicate"' \
  -i 'save_every_n_epochs=1' \
  -i 'send_to_huggingface=false' \
  -i 'cache_latents_to_disk=true' \
  -i 'gradient_checkpointing=true' \
  -i 'learning_rate_te1_bool=0.000003' \
  -i 'learning_rate_te2_bool=0' \
  -i 'lr_scheduler_num_cycles=1' \
  -i 'max_data_loader_n_workers=0' \
  -i 'num_cpu_threads_per_process=4'

To learn more, take a look at the Cog documentation.

Run this to download the model and run it in your local environment:

docker run -d -p 5000:5000 --gpus=all r8.im/zelenioncode/dreambooth_sdxl@sha256:e90767fffa851f984703defc193b1861d3b6d79acbcf1a798a3b9a1dac0bc176
curl -s -X POST \
  -H "Content-Type: application/json" \
  -d $'{
    "input": {
      "vae": "stabilityai/sdxl-vae",
      "input": "http",
      "gender": "woman",
      "xformers": true,
      "class_reg": "",
      "full_fp16": true,
      "model_sdxl": "Stable Diffusion XL",
      "name_model": "Dreambooth_sdxl",
      "resolution": "1024,1024",
      "no_half_vae": true,
      "warmup_init": false,
      "lr_scheduler": "constant",
      "mem_eff_attn": true,
      "noise_offset": 0,
      "repeat_input": 100,
      "weight_decay": 0.01,
      "cache_latents": true,
      "learning_rate": 0.00001,
      "max_grad_norm": 0,
      "relative_step": false,
      "use_class_reg": false,
      "optimizer_type": "AdaFactor",
      "save_precision": "fp16",
      "max_train_steps": 3000,
      "mixed_precision": "fp16",
      "scale_parameter": false,
      "repeat_class_reg": 1,
      "train_batch_size": 1,
      "bucket_no_upscale": true,
      "bucket_reso_steps": 64,
      "learning_rate_te1": 0.00001,
      "learning_rate_te2": 0.00001,
      "token_huggingface": "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",
      "folder_huggingface": "",
      "save_every_n_steps": 1,
      "train_text_encoder": true,
      "repo_id_huggingface": "WGlint/SafetensorsFromReplicate",
      "save_every_n_epochs": 1,
      "send_to_huggingface": false,
      "cache_latents_to_disk": true,
      "gradient_checkpointing": true,
      "learning_rate_te1_bool": 0.000003,
      "learning_rate_te2_bool": 0,
      "lr_scheduler_num_cycles": 1,
      "max_data_loader_n_workers": 0,
      "num_cpu_threads_per_process": 4
    }
  }' \
  http://localhost:5000/predictions

To learn more, take a look at the Cog documentation.

Output

No output yet! Press "Submit" to start a prediction.

Run time and cost

This model runs on Nvidia A100 (80GB) GPU hardware. We don't yet have enough runs of this model to provide performance information.

Readme

This model doesn't have a readme.