You are viewing main version, which requires installation from source. If you'd like
regular pip install, checkout the latest stable version (v0.7.69).
DreamBooth
DreamBooth is a method to personalize text-to-image models like Stable Diffusion given just a few (3-5) images of a subject. It allows the model to generate contextualized images of the subject in different scenes, poses, and views.
Data Preparation
The data format for DreamBooth training is simple. All you need is images of a concept (e.g. a person) and a concept token.
To train a dreambooth model, please select an appropriate model from the hub. When choosing a model from the hub, please make sure you select the correct image size compatible with the model.
Your concept token is prompt in parameters section.
Parameters
❯ autotrain dreambooth --help
usage: autotrain <command> [<args>] dreambooth [-h] [--train] [--deploy] [--inference] [--username USERNAME]
[--backend {local-cli,spaces-a10gl,spaces-a10gs,spaces-a100,spaces-t4m,spaces-t4s,spaces-cpu,spaces-cpuf}]
[--token TOKEN] [--push-to-hub] --model MODEL --project-name PROJECT_NAME [--data-path DATA_PATH]
[--train-split TRAIN_SPLIT] [--valid-split VALID_SPLIT] [--batch-size BATCH_SIZE] [--seed SEED]
[--epochs EPOCHS] [--gradient_accumulation GRADIENT_ACCUMULATION] [--disable_gradient_checkpointing]
[--lr LR] [--log {none,wandb,tensorboard}] [--revision REVISION] [--tokenizer TOKENIZER] --image-path
IMAGE_PATH [--class-image-path CLASS_IMAGE_PATH] --prompt PROMPT [--class-prompt CLASS_PROMPT]
[--num-class-images NUM_CLASS_IMAGES] [--class-labels-conditioning CLASS_LABELS_CONDITIONING]
[--prior-preservation] [--prior-loss-weight PRIOR_LOSS_WEIGHT] --resolution RESOLUTION
[--center-crop] [--train-text-encoder] [--sample-batch-size SAMPLE_BATCH_SIZE]
[--num-steps NUM_STEPS] [--checkpointing-steps CHECKPOINTING_STEPS]
[--resume-from-checkpoint RESUME_FROM_CHECKPOINT] [--scale-lr] [--scheduler SCHEDULER]
[--warmup-steps WARMUP_STEPS] [--num-cycles NUM_CYCLES] [--lr-power LR_POWER]
[--dataloader-num-workers DATALOADER_NUM_WORKERS] [--use-8bit-adam] [--adam-beta1 ADAM_BETA1]
[--adam-beta2 ADAM_BETA2] [--adam-weight-decay ADAM_WEIGHT_DECAY] [--adam-epsilon ADAM_EPSILON]
[--max-grad-norm MAX_GRAD_NORM] [--allow-tf32]
[--prior-generation-precision PRIOR_GENERATION_PRECISION] [--local-rank LOCAL_RANK] [--xformers]
[--pre-compute-text-embeddings] [--tokenizer-max-length TOKENIZER_MAX_LENGTH]
[--text-encoder-use-attention-mask] [--rank RANK] [--xl] [--mixed-precision MIXED_PRECISION]
[--validation-prompt VALIDATION_PROMPT] [--num-validation-images NUM_VALIDATION_IMAGES]
[--validation-epochs VALIDATION_EPOCHS] [--checkpoints-total-limit CHECKPOINTS_TOTAL_LIMIT]
[--validation-images VALIDATION_IMAGES] [--logging]
✨ Run AutoTrain DreamBooth Training
options:
-h, --help show this help message and exit
--train Command to train the model
--deploy Command to deploy the model (limited availability)
--inference Command to run inference (limited availability)
--username USERNAME Hugging Face Hub Username
--backend {local-cli,spaces-a10gl,spaces-a10gs,spaces-a100,spaces-t4m,spaces-t4s,spaces-cpu,spaces-cpuf}
Backend to use: default or spaces. Spaces backend requires push_to_hub & username. Advanced users only.
--token TOKEN Your Hugging Face API token. Token must have write access to the model hub.
--push-to-hub Push to hub after training will push the trained model to the Hugging Face model hub.
--model MODEL Base model to use for training
--project-name PROJECT_NAME
Output directory / repo id for trained model (must be unique on hub)
--data-path DATA_PATH
Train dataset to use. When using cli, this should be a directory path containing training and validation data in appropriate
formats
--train-split TRAIN_SPLIT
Train dataset split to use
--valid-split VALID_SPLIT
Validation dataset split to use
--batch-size BATCH_SIZE
Training batch size to use
--seed SEED Random seed for reproducibility
--epochs EPOCHS Number of training epochs
--gradient_accumulation GRADIENT_ACCUMULATION
Gradient accumulation steps
--disable_gradient_checkpointing
Disable gradient checkpointing
--lr LR Learning rate
--log {none,wandb,tensorboard}
Use experiment tracking
--revision REVISION Model revision to use for training
--tokenizer TOKENIZER
Tokenizer to use for training
--image-path IMAGE_PATH
Path to the images
--class-image-path CLASS_IMAGE_PATH
Path to the class images
--prompt PROMPT Instance prompt
--class-prompt CLASS_PROMPT
Class prompt
--num-class-images NUM_CLASS_IMAGES
Number of class images
--class-labels-conditioning CLASS_LABELS_CONDITIONING
Class labels conditioning
--prior-preservation With prior preservation
--prior-loss-weight PRIOR_LOSS_WEIGHT
Prior loss weight
--resolution RESOLUTION
Resolution
--center-crop Center crop
--train-text-encoder Train text encoder
--sample-batch-size SAMPLE_BATCH_SIZE
Sample batch size
--num-steps NUM_STEPS
Max train steps
--checkpointing-steps CHECKPOINTING_STEPS
Checkpointing steps
--resume-from-checkpoint RESUME_FROM_CHECKPOINT
Resume from checkpoint
--scale-lr Scale learning rate
--scheduler SCHEDULER
Learning rate scheduler
--warmup-steps WARMUP_STEPS
Learning rate warmup steps
--num-cycles NUM_CYCLES
Learning rate num cycles
--lr-power LR_POWER Learning rate power
--dataloader-num-workers DATALOADER_NUM_WORKERS
Dataloader num workers
--use-8bit-adam Use 8bit adam
--adam-beta1 ADAM_BETA1
Adam beta 1
--adam-beta2 ADAM_BETA2
Adam beta 2
--adam-weight-decay ADAM_WEIGHT_DECAY
Adam weight decay
--adam-epsilon ADAM_EPSILON
Adam epsilon
--max-grad-norm MAX_GRAD_NORM
Max grad norm
--allow-tf32 Allow TF32
--prior-generation-precision PRIOR_GENERATION_PRECISION
Prior generation precision
--local-rank LOCAL_RANK
Local rank
--xformers Enable xformers memory efficient attention
--pre-compute-text-embeddings
Pre compute text embeddings
--tokenizer-max-length TOKENIZER_MAX_LENGTH
Tokenizer max length
--text-encoder-use-attention-mask
Text encoder use attention mask
--rank RANK Rank
--xl XL
--mixed-precision MIXED_PRECISION
mixed precision, fp16, bf16, none
--validation-prompt VALIDATION_PROMPT
Validation prompt
--num-validation-images NUM_VALIDATION_IMAGES
Number of validation images
--validation-epochs VALIDATION_EPOCHS
Validation epochs
--checkpoints-total-limit CHECKPOINTS_TOTAL_LIMIT
Checkpoints total limit
--validation-images VALIDATION_IMAGES
Validation images
--logging Logging using tensorboard