[network_arguments]
unet_lr = 0.0003
text_encoder_lr = 6e-5
network_dim = 8
network_alpha = 4
network_module = "networks.lora"
network_train_unet_only = false

[optimizer_arguments]
learning_rate = 0.0003
lr_scheduler = "cosine_with_restarts"
lr_scheduler_num_cycles = 3
lr_warmup_steps = 8
optimizer_type = "AdamW8bit"
optimizer_args = [ "weight_decay=0.1", "betas=[0.9,0.99]",]
loss_type = "l2"
max_grad_norm = 1.0

[training_arguments]
lowram = true
pretrained_model_name_or_path = "/content/sd_xl_base_1.0.safetensors"
vae = "/content/sdxl_vae.safetensors"
max_train_epochs = 10
train_batch_size = 6
seed = 42
max_token_length = 225
xformers = false
sdpa = true
min_snr_gamma = 8.0
no_half_vae = true
gradient_checkpointing = true
gradient_accumulation_steps = 1
max_data_loader_n_workers = 1
persistent_data_loader_workers = true
mixed_precision = "fp16"
full_fp16 = true
full_bf16 = false
cache_latents = true
cache_latents_to_disk = true
cache_text_encoder_outputs = false
min_timestep = 0
max_timestep = 1000
prior_loss_weight = 1.0
multires_noise_iterations = 6
multires_noise_discount = 0.3

[saving_arguments]
save_precision = "fp16"
save_model_as = "safetensors"
save_every_n_epochs = 1
save_last_n_epochs = 10
output_name = "lingospace"
output_dir = "/content/drive/MyDrive/Loras/lingospace/output"
log_prefix = "lingospace"
logging_dir = "/content/drive/MyDrive/Loras/_logs"