Trained on anime only.
Workflow included.
How to prompt: https://rentry.co/43506
[network_arguments]
network_dim = 32
network_alpha = 32
network_module = "networks.lora_flux_2"
[optimizer_arguments]
learning_rate = 1.0
optimizer_type = "prodigyplus.prodigy_plus_schedulefree.ProdigyPlusScheduleFree"
optimizer_args = ["betas=(0.9, 0.999)", "use_speed=True", "use_orthograd=True", "use_bias_correction=True"]
[training_arguments]
max_train_epochs = 20 # should be about 10k~ steps.
seed = 42
timestep_sampling = "custom_flux2"
weighting_scheme = "none"
mixed_precision = "bf16"
flash_attn = true
gradient_checkpointing = true
compile = true
max_data_loader_n_workers = 4
persistent_data_loader_workers = true
[sampling]
sample_every_n_epochs = 1
sample_prompts = "anime-stripper-klein9b-v4/prompts.txt"
sample_at_first = true
[saving_arguments]
save_every_n_epochs = 1
wandb_api_key = ""
output_dir = "anime-stripper-klein9b-v4/output"
output_name = "anime-stripper-klein9b-v4"
wandb_run_name = "anime-stripper-klein9b-v4"
log_prefix = "anime-stripper-klein9b-v4"
log_with = "wandb"
[general]
caption_extension = ".txt"
enable_bucket = true
bucket_no_upscale = false
batch_size = 8
[[datasets]]
resolution = 1024
num_repeats = 1
image_jsonl_file = "anime-stripper-klein9b-v4/dataset.jsonl"
cache_directory = "anime-stripper-klein9b-v4/datasetb"
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for Bakanayatsu/klein-9b-anime-clothes-stripper
Base model
black-forest-labs/FLUX.2-klein-base-9B