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.
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.
❯ 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