v1 sample images can be found here: https://civitai.com/posts/12553354
https://civitai.com/posts/12553354original images used the following prompts
[{"prompt": "standing by the beach"}, {"prompt": "man working as a police officer"}, {"prompt": "man going to work in an empty train cabin"}]
[[datasets]]
[[datasets.subsets]]
num_repeats = 8
[general]
resolution = 1024
shuffle_caption = true
keep_tokens = 2
flip_aug = true
enable_bucket = true
bucket_reso_steps = ??
bucket_no_upscale = true
min_bucket_reso = 256
max_bucket_reso = 2048
Description
V1 sample images: https://civitai.com/posts/12553354sample images for V1.
the following results in the large file size
[additional_network_arguments]
unet_lr = 1.0
text_encoder_lr = 1.0
network_dim = 128
network_alpha = 16
network_module = "networks.lora"[optimizer_arguments]
learning_rate = 1.0
lr_scheduler = "cosine"
lr_warmup_steps = 0
optimizer_type = "Prodigy"
optimizer_args = [ "decouple=True", "weight_decay=0.5", "betas=0.9,0.99", "use_bias_correction=False",]
[training_arguments]
max_train_steps = 0
max_train_epochs = 52
sample_sampler = "euler_a"
train_batch_size = 5
noise_offset = 0.03
clip_skip = 1
weighted_captions = false
max_token_length = 225
save_precision = "bf16"
mixed_precision = "bf16"
xformers = true
sdpa = true
no_half_vae = true
gradient_checkpointing = true
gradient_accumulation_steps = 1
[advanced_training_config]
multires_noise_iterations = 6
multires_noise_discount = 0.3
min_snr_gamma = 5.0
[model_arguments]
[dreambooth_arguments]
prior_loss_weight = 1.0
[dataset_arguments]
cache_latents = true
FAQ
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.