Updated Ver NOOB V-PRED V1
Trained based on NOOB V-PRED 1.0 with OneTrainer. See about this version for training details. Since NOOB can do trainside concept, the other concepts like train with legs up and train in back view cannot be generated easily. So I made this update.
Pros:
Better artstyle
Better concept generation
Cons:
Frequent distortions of anatomy
Number of characters is difficult to control
Many trials needed
Introduction
A Pony based LyCORIS to enhance the concept Love train. Four trigger words set to control the view. See the examples first.
Pros
Concept enhancement
Cons
Not responsive to trigger words, you need to upweight it or roll for it times to times
Many trials needed
Influence on art style
Training details
See about the version.
Description
{
"LoRA_type": "LyCORIS/LoCon",
"LyCORIS_preset": "full",
"adaptive_noise_scale": 0,
"additional_parameters": "",
"block_alphas": "",
"block_dims": "",
"block_lr_zero_threshold": "",
"bucket_no_upscale": true,
"bucket_reso_steps": 128,
"bypass_mode": false,
"cache_latents": false,
"cache_latents_to_disk": false,
"caption_dropout_every_n_epochs": 0.0,
"caption_dropout_rate": 0,
"caption_extension": ".txt",
"clip_skip": 2,
"color_aug": false,
"constrain": 0.0,
"conv_alpha": 4,
"conv_block_alphas": "",
"conv_block_dims": "",
"conv_dim": 8,
"dataset_config": "",
"debiased_estimation_loss": false,
"decompose_both": false,
"dim_from_weights": false,
"dora_wd": false,
"down_lr_weight": "",
"enable_bucket": true,
"epoch": 10,
"factor": -1,
"flip_aug": false,
"fp8_base": false,
"full_bf16": false,
"full_fp16": false,
"gpu_ids": "",
"gradient_accumulation_steps": "1",
"gradient_checkpointing": true,
"keep_tokens": 1,
"learning_rate": 1.0,
"log_tracker_config": "",
"log_tracker_name": "",
"logging_dir": "/root/autodl-tmp/output",
"lora_network_weights": "",
"lr_scheduler": "cosine",
"lr_scheduler_args": "",
"lr_scheduler_num_cycles": "",
"lr_scheduler_power": "",
"lr_warmup": 0,
"max_bucket_reso": 4096,
"max_data_loader_n_workers": "0",
"max_grad_norm": 1,
"max_resolution": "1024,1024",
"max_timestep": 1000,
"max_token_length": "75",
"max_train_epochs": "",
"max_train_steps": "",
"mem_eff_attn": false,
"mid_lr_weight": "",
"min_bucket_reso": 1024,
"min_snr_gamma": 0,
"min_timestep": 0,
"mixed_precision": "bf16",
"model_list": "custom",
"module_dropout": 0,
"multi_gpu": false,
"multires_noise_discount": 0.1,
"multires_noise_iterations": 6,
"network_alpha": 8,
"network_dim": 16,
"network_dropout": 0,
"noise_offset": 0,
"noise_offset_type": "Multires",
"num_cpu_threads_per_process": 2,
"num_machines": 1,
"num_processes": 1,
"optimizer": "Prodigy",
"optimizer_args": "weight_decay=0.01 d_coef=1 use_bias_correction=True safeguard_warmup=False betas=0.9,0.99 decouple=True ",
"output_dir": "/root/autodl-tmp/output",
"output_name": "Love Train-PonyXL",
"persistent_data_loader_workers": false,
"pretrained_model_name_or_path": "/root/kohya_ss/models/ponyDiffusionV6XL.safetensors",
"prior_loss_weight": 1.0,
"random_crop": false,
"rank_dropout": 0,
"rank_dropout_scale": false,
"reg_data_dir": "",
"rescaled": false,
"resume": "",
"sample_every_n_epochs": 1,
"sample_every_n_steps": 0,
"sample_prompts": "score_9, score_8_up, score_7_up, anime BREAK, trainside, long hair, breasts, blush, short hair, open mouth, bangs, multiple girls, large breasts, black hair, 1boy, 2girls, very long hair, nipples, blue hair, swimsuit, closed eyes, ass, heart, bikini, nude, sweat, censored, penis, barefoot, teeth, tongue, indoors, dark skin, cum, sex, tongue out, armpits, arm up, bar censor, kneeling, white bikini, erection, testicles, trembling, tan, clenched teeth, group sex, anal, ejaculation, multiple penises, handjob, tanlines, futanari, age difference, shota, ejaculating while penetrated, futa with male, futa with futa, penis size difference, futa on male --n source_pony, source_furry, source_cartoon, greyscale, from below --w 1024 --h 1024\n ",
"sample_sampler": "euler_a",
"save_every_n_epochs": 1,
"save_every_n_steps": 0,
"save_last_n_steps": 0,
"save_last_n_steps_state": 0,
"save_model_as": "safetensors",
"save_precision": "bf16",
"save_state": false,
"scale_v_pred_loss_like_noise_pred": false,
"scale_weight_norms": 0,
"sdxl": true,
"sdxl_cache_text_encoder_outputs": false,
"sdxl_no_half_vae": true,
"seed": "31337",
"shuffle_caption": true,
"stop_text_encoder_training_pct": 0,
"text_encoder_lr": 1.0,
"train_batch_size": 5,
"train_data_dir": "/root/autodl-tmp/love_train",
"train_norm": false,
"train_on_input": true,
"training_comment": "",
"unet_lr": 1.0,
"unit": 1,
"up_lr_weight": "",
"use_cp": false,
"use_scalar": false,
"use_tucker": false,
"use_wandb": false,
"v2": false,
"v_parameterization": false,
"v_pred_like_loss": 0,
"vae": "/root/kohya_ss/models/sdxl_vae_fp16fix.safetensors",
"vae_batch_size": 0,
"wandb_api_key": "",
"wandb_run_name": "",
"weighted_captions": false,
"xformers": "xformers"
}FAQ
Comments (5)
is this only for futas or also for girls wearing strap-ons?
Dataset is only about futa. But since they are all tagged with penis. Technically it is feasible to remove penis and replace it with strap in the prompt.
I tried various things because it didn't work well.
When "from side",it seems that you can't "TRAIN" unless the image is landscape.
(For example, W1208 x H1024)
I wasted time trying various things while switching between prompts and checkpoints...sad.
Any chance of making this lora available for Illustrious?
NOOB can do this easily since E-PRED 0.5. See my post for example:https://civitai.com/images/37932090