Softmecha 软妹甲版本v1.0 (ID: 168737)

Trained by Prodigy optiminzer.

Compared with Adam and Lion,

This is indeed the best type.

10_times, 15 epoches,

the best one is the third modle.

model_train_type = "sd-lora" pretrained_model_name_or_path = "./sd-models/dreamshaper_8.safetensors" v2 = false train_data_dir = "./train/softmecha" prior_loss_weight = 1 resolution = "512,768" enable_bucket = true min_bucket_reso = 256 max_bucket_reso = 1_024 bucket_reso_steps = 64 output_name = "softmecha" output_dir = "./output" save_model_as = "safetensors" save_precision = "fp16" save_every_n_epochs = 1 max_train_epochs = 15 train_batch_size = 2 gradient_checkpointing = false network_train_unet_only = false network_train_text_encoder_only = false learning_rate = 1 unet_lr = 1 text_encoder_lr = 1 lr_scheduler = "cosine_with_restarts" lr_warmup_steps = 0 lr_scheduler_num_cycles = 1 optimizer_type = "Prodigy" network_module = "networks.lora" network_dim = 64 network_alpha = 32 sample_prompts = "(masterpiece, best quality:1.2), 1girl, solo, --n lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts,signature, watermark, username, blurry, --w 512 --h 768 --l 7 --s 24 --d 1337" sample_sampler = "euler_a" sample_every_n_epochs = 1 log_with = "tensorboard" logging_dir = "./logs" caption_extension = ".txt" shuffle_caption = true weighted_captions = false keep_tokens = 0 max_token_length = 255 seed = 1_337 clip_skip = 2 mixed_precision = "fp16" full_fp16 = false full_bf16 = false xformers = true lowram = false cache_latents = true cache_latents_to_disk = true cache_text_encoder_outputs = false cache_text_encoder_outputs_to_disk = false persistent_data_loader_workers = true optimizer_args = [ "decouple=True", "weight_decay=0.01", "use_bias_correction=True", "d_coef=2.0" ]

描述:

训练词语: