You're looking at a specific version of this model. Jump to the model overview.

edenartlab /sdxl-lora-trainer:8ea5fbb5

Playground API

Input schema

The fields you can use to run this model with an API. If you don’t give a value for a field its default value will be used.

Field Type Default value Description
name
string
unnamed
Name of new LORA concept
lora_training_urls
string
Training images for new LORA concept (can be image urls or a .zip file of images)
concept_mode
string
object
'face' / 'style' / 'object' (default)
seed
integer
Random seed for reproducible training. Leave empty to use a random seed
resolution
integer
960
Square pixel resolution which your images will be resized to for training recommended [768-1024]
train_batch_size
integer
4
Batch size (per device) for training
num_train_epochs
integer
10000
Number of epochs to loop through your training dataset
max_train_steps
integer
600
Number of individual training steps. Takes precedence over num_train_epochs
checkpointing_steps
integer
10000
Number of steps between saving checkpoints. Set to very very high number to disable checkpointing, because you don't need one.
is_lora
boolean
True
Whether to use LoRA training. If set to False, will use Full fine tuning
prodigy_d_coef
number
0.8
Multiplier for internal learning rate of Prodigy optimizer
ti_lr
number
0.001
Learning rate for training textual inversion embeddings. Don't alter unless you know what you're doing.
ti_weight_decay
number
0.0003
weight decay for textual inversion embeddings. Don't alter unless you know what you're doing.
lora_weight_decay
number
0.002
weight decay for lora parameters. Don't alter unless you know what you're doing.
l1_penalty
number
0.1
Sparsity penalty for the LoRA matrices, increases merge-ability and maybe generalization
lora_param_scaler
number
0.5
Multiplier for the starting weights of the lora matrices
snr_gamma
number
5
see https://arxiv.org/pdf/2303.09556.pdf, set to None to disable snr training
lora_rank
integer
12
Rank of LoRA embeddings. For faces 5 is good, for complex concepts / styles you can try 8 or 12
caption_prefix
string
Prefix text prepended to automatic captioning. Must contain the 'TOK'. Example is 'a photo of TOK, '. If empty, chatgpt will take care of this automatically
left_right_flip_augmentation
boolean
True
Add left-right flipped version of each img to the training data, recommended for most cases. If you are learning a face, you prob want to disable this
augment_imgs_up_to_n
integer
20
Apply data augmentation (no lr-flipping) until there are n training samples (0 disables augmentation completely)
n_tokens
integer
2
How many new tokens to inject per concept
mask_target_prompts
string
Prompt that describes most important part of the image, will be used for CLIP-segmentation. For example, if you are learning a person 'face' would be a good segmentation prompt
crop_based_on_salience
boolean
True
If you want to crop the image to `target_size` based on the important parts of the image, set this to True. If you want to crop the image based on face detection, set this to False
use_face_detection_instead
boolean
False
If you want to use face detection instead of CLIPSeg for masking. For face applications, we recommend using this option.
clipseg_temperature
number
0.6
How blurry you want the CLIPSeg mask to be. We recommend this value be something between `0.5` to `1.0`. If you want to have more sharp mask (but thus more errorful), you can decrease this value.
verbose
boolean
True
verbose output
run_name
string
1717018285
Subdirectory where all files will be saved
debug
boolean
False
for debugging locally only (dont activate this on replicate)
hard_pivot
boolean
False
Use hard freeze for ti_lr. If set to False, will use soft transition of learning rates
off_ratio_power
number
0.1
How strongly to correct the embedding std vs the avg-std (0=off, 0.05=weak, 0.1=standard)

Output schema

The shape of the response you’ll get when you run this model with an API.

Schema 
{'items': {'properties': {'attributes': {'title': 'Attributes',
                                         'type': 'object'},
                          'files': {'default': [],
                                    'items': {'format': 'uri',
                                              'type': 'string'},
                                    'title': 'Files',
                                    'type': 'array'},
                          'isFinal': {'default': False,
                                      'title': 'Isfinal',
                                      'type': 'boolean'},
                          'name': {'title': 'Name', 'type': 'string'},
                          'progress': {'title': 'Progress', 'type': 'number'},
                          'thumbnails': {'default': [],
                                         'items': {'format': 'uri',
                                                   'type': 'string'},
                                         'title': 'Thumbnails',
                                         'type': 'array'}},
           'title': 'CogOutput',
           'type': 'object'},
 'title': 'Output',
 'type': 'array',
 'x-cog-array-type': 'iterator'}