sotediffusion-wuerstchen3-alpha3

SoteDiffusion Wuerstchen3

Anime finetune of Würstchen V3.
Still in active training.
No commercial use.

Release Notes

• Switched to OneTrainer.
• Trained more.
• Currently trained on 2,75M images in total
• Used the FP16 fix script made by KBlueLeaf.

<style> .image { float: left; margin-left: 10px; } </style> <table> <img class="image" src="https://cdn-uploads.huggingface.co/production/uploads/6456af6195082f722d178522/lfxdg68qO5jrNg20AMuTf.png"> </table>

UI Guide

SD.Next

URL: https://github.com/vladmandic/automatic/

Go to Models -> Huggingface and type Disty0/sotediffusion-wuerstchen3-alpha3-decoder into the model name and press download.
Load Disty0/sotediffusion-wuerstchen3-alpha3-decoder after the download process is complete.

Prompt:

extremely aesthetic, best quality, newest,

Negative Prompt:

very displeasing, worst quality, monochrome, sketch, realistic, oldest, visual novel cg,

Parameters:
Sampler: Default

Steps: 30 or 40
Refiner Steps: 10

CFG: 8-6
Secondary CFG: 2 or 1

Resolution: 1024x1536, 2048x1152
Anything works as long as it's a multiply of 128.

ComfyUI

Please refer to CivitAI: https://civitai.com/models/353284

Code Example

pip install diffusers

import torch
from diffusers import StableCascadeCombinedPipeline

device = "cuda"
dtype = torch.bfloat16 # or torch.float16
model = "Disty0/sotediffusion-wuerstchen3-alpha3-decoder"

pipe = StableCascadeCombinedPipeline.from_pretrained(model, torch_dtype=dtype)

# send everything to the gpu:
pipe = pipe.to(device, dtype=dtype)
pipe.prior_pipe = pipe.prior_pipe.to(device, dtype=dtype)

# or enable model offload to save vram:
# pipe.enable_model_cpu_offload()



prompt = "1girl, solo, cat ears, straight hair, pink hair, orange eyes, bare shoulders, looking at viewer, smile, newest, extremely aesthetic, best quality, general, black background, wind, petals, feathers, magic, electricity, bioluminescence, depth of field, fireflies, head tilt, fire, space,"
negative_prompt = "very displeasing, worst quality, monochrome, sketch, realistic, oldest, visual novel cg,"
output = pipe(
    width=1024,
    height=1536,
    prompt=prompt,
    negative_prompt=negative_prompt,
    decoder_guidance_scale=2.0,
    prior_guidance_scale=8.0,
    prior_num_inference_steps=40,
    output_type="pil",
    num_inference_steps=10
).images[0]

## do something with the output image

Training Status:

GPU used for training: 1x AMD RX 7900 XTX 24GB
GPU Hours: 500 (Accumulative starting from alpha1)

Note: chunks starts from 0 and there are 8000 images per chunk

Dataset:

GPU used for captioning: 1x Intel ARC A770 16GB
GPU Hours: 350

Model used for captioning: SmilingWolf/wd-swinv2-tagger-v3
Command:

python /mnt/DataSSD/AI/Apps/kohya_ss/sd-scripts/finetune/tag_images_by_wd14_tagger.py --model_dir "/mnt/DataSSD/AI/models/wd14_tagger_model" --repo_id "SmilingWolf/wd-swinv2-tagger-v3" --recursive --remove_underscore --use_rating_tags --character_tags_first --character_tag_expand --append_tags --onnx --caption_separator ", " --general_threshold 0.35 --character_threshold 0.50 --batch_size 4 --caption_extension ".txt" ./

Note:

• Smallest size is 1280x600 | 768.000 pixels
• Deduped based on image similarity using czkawka-cli

Tags:

Model is trained with random tag order but this is the order in the dataset if you are interested:

aesthetic tags, quality tags, date tags, custom tags, rating tags, character, series, rest of the tags

Date:

Aesthetic Tags:

Model used: shadowlilac/aesthetic-shadow-v2

Quality Tags:

Model used: https://huggingface.co/hakurei/waifu-diffusion-v1-4/blob/main/models/aes-B32-v0.pth

Rating Tags

Custom Tags:

Training Parameters:

Software used: OneTrainer
https://github.com/Nerogar/OneTrainer/

Base model: Disty0/sote-diffusion-cascade-alpha2

Config:

{
    "__version": 3,
    "training_method": "FINE_TUNE",
    "model_type": "STABLE_CASCADE_1",
    "debug_mode": false,
    "debug_dir": "debug",
    "workspace_dir": "/mnt/DataSSD/AI/SoteDiffusion/Wuerstchen3/OneTrainer/run",
    "cache_dir": "/mnt/DataSSD/AI/SoteDiffusion/Wuerstchen3/OneTrainer/workspace-cache/run",
    "tensorboard": true,
    "tensorboard_expose": false,
    "continue_last_backup": true,
    "include_train_config": "NONE",
    "base_model_name": "Disty0/sotediffusion-wuerstchen3-alpha2",
    "weight_dtype": "BFLOAT_16",
    "output_dtype": "BFLOAT_16",
    "output_model_format": "SAFETENSORS",
    "output_model_destination": "/mnt/DataSSD/AI/SoteDiffusion/Wuerstchen3/OneTrainer/workspace-cache/models",
    "gradient_checkpointing": true,
    "force_circular_padding": false,
    "concept_file_name": "training_concepts/concepts.json",
    "concepts": [
        {
            "__version": 1,
            "image": {
                "__version": 0,
                "enable_crop_jitter": false,
                "enable_random_flip": false,
                "enable_fixed_flip": false,
                "enable_random_rotate": false,
                "enable_fixed_rotate": false,
                "random_rotate_max_angle": 0.0,
                "enable_random_brightness": false,
                "enable_fixed_brightness": false,
                "random_brightness_max_strength": 0.0,
                "enable_random_contrast": false,
                "enable_fixed_contrast": false,
                "random_contrast_max_strength": 0.0,
                "enable_random_saturation": false,
                "enable_fixed_saturation": false,
                "random_saturation_max_strength": 0.0,
                "enable_random_hue": false,
                "enable_fixed_hue": false,
                "random_hue_max_strength": 0.0,
                "enable_resolution_override": false,
                "resolution_override": "1024"
            },
            "text": {
                "__version": 0,
                "prompt_source": "sample",
                "prompt_path": "",
                "enable_tag_shuffling": true,
                "tag_delimiter": ", ",
                "keep_tags_count": 1
            },
            "name": "",
            "path": "/mnt/DataSSD/AI/anime_image_dataset/best/newest_best",
            "seed": -209204630,
            "enabled": true,
            "include_subdirectories": true,
            "image_variations": 1,
            "text_variations": 1,
            "balancing": 1.0,
            "balancing_strategy": "REPEATS",
            "loss_weight": 1.0
        }
    ],
    "circular_mask_generation": false,
    "random_rotate_and_crop": false,
    "aspect_ratio_bucketing": true,
    "latent_caching": true,
    "clear_cache_before_training": false,
    "learning_rate_scheduler": "CONSTANT",
    "learning_rate": 1e-05,
    "learning_rate_warmup_steps": 200,
    "learning_rate_cycles": 1,
    "epochs": 1,
    "batch_size": 16,
    "gradient_accumulation_steps": 1,
    "ema": "OFF",
    "ema_decay": 0.999,
    "ema_update_step_interval": 5,
    "dataloader_threads": 8,
    "train_device": "cuda",
    "temp_device": "cpu",
    "train_dtype": "FLOAT_16",
    "fallback_train_dtype": "BFLOAT_16",
    "enable_autocast_cache": true,
    "only_cache": false,
    "resolution": "1024",
    "attention_mechanism": "SDP",
    "align_prop": false,
    "align_prop_probability": 0.1,
    "align_prop_loss": "AESTHETIC",
    "align_prop_weight": 0.01,
    "align_prop_steps": 20,
    "align_prop_truncate_steps": 0.5,
    "align_prop_cfg_scale": 7.0,
    "mse_strength": 1.0,
    "mae_strength": 0.0,
    "vb_loss_strength": 1.0,
    "loss_weight_fn": "P2",
    "loss_weight_strength": 1.0,
    "dropout_probability": 0.0,
    "loss_scaler": "NONE",
    "learning_rate_scaler": "NONE",
    "offset_noise_weight": 0.0,
    "perturbation_noise_weight": 0.0,
    "rescale_noise_scheduler_to_zero_terminal_snr": false,
    "force_v_prediction": false,
    "force_epsilon_prediction": false,
    "min_noising_strength": 0.0,
    "max_noising_strength": 1.0,
    "noising_weight": 0.0,
    "noising_bias": 0.5,
    "unet": {
        "__version": 0,
        "model_name": "",
        "train": true,
        "stop_training_after": 0,
        "stop_training_after_unit": "NEVER",
        "learning_rate": null,
        "weight_dtype": "NONE"
    },
    "prior": {
        "__version": 0,
        "model_name": "",
        "train": true,
        "stop_training_after": 0,
        "stop_training_after_unit": "NEVER",
        "learning_rate": null,
        "weight_dtype": "NONE"
    },
    "text_encoder": {
        "__version": 0,
        "model_name": "",
        "train": true,
        "stop_training_after": 0,
        "stop_training_after_unit": "NEVER",
        "learning_rate": null,
        "weight_dtype": "NONE"
    },
    "text_encoder_layer_skip": 0,
    "text_encoder_2": {
        "__version": 0,
        "model_name": "",
        "train": true,
        "stop_training_after": 30,
        "stop_training_after_unit": "EPOCH",
        "learning_rate": null,
        "weight_dtype": "NONE"
    },
    "text_encoder_2_layer_skip": 0,
    "vae": {
        "__version": 0,
        "model_name": "",
        "train": true,
        "stop_training_after": null,
        "stop_training_after_unit": "NEVER",
        "learning_rate": null,
        "weight_dtype": "FLOAT_32"
    },
    "effnet_encoder": {
        "__version": 0,
        "model_name": "/mnt/DataSSD/AI/models/wuerstchen3/effnet_encoder.safetensors",
        "train": true,
        "stop_training_after": null,
        "stop_training_after_unit": "NEVER",
        "learning_rate": null,
        "weight_dtype": "FLOAT_16"
    },
    "decoder": {
        "__version": 0,
        "model_name": "Disty0/sotediffusion-wuerstchen3-alpha2-decoder",
        "train": true,
        "stop_training_after": null,
        "stop_training_after_unit": "NEVER",
        "learning_rate": null,
        "weight_dtype": "FLOAT_16"
    },
    "decoder_text_encoder": {
        "__version": 0,
        "model_name": "",
        "train": true,
        "stop_training_after": null,
        "stop_training_after_unit": "NEVER",
        "learning_rate": null,
        "weight_dtype": "NONE"
    },
    "decoder_vqgan": {
        "__version": 0,
        "model_name": "",
        "train": true,
        "stop_training_after": null,
        "stop_training_after_unit": "NEVER",
        "learning_rate": null,
        "weight_dtype": "FLOAT_16"
    },
    "masked_training": false,
    "unmasked_probability": 0.1,
    "unmasked_weight": 0.1,
    "normalize_masked_area_loss": false,
    "embedding_learning_rate": null,
    "preserve_embedding_norm": false,
    "embedding": {
        "__version": 0,
        "uuid": "bf3a36b4-bd01-4b46-b818-3c6414887497",
        "model_name": "",
        "placeholder": "<embedding>",
        "train": true,
        "stop_training_after": null,
        "stop_training_after_unit": "NEVER",
        "token_count": 1,
        "initial_embedding_text": "*"
    },
    "additional_embeddings": [],
    "embedding_weight_dtype": "FLOAT_32",
    "lora_model_name": "",
    "lora_rank": 16,
    "lora_alpha": 1.0,
    "lora_weight_dtype": "FLOAT_32",
    "optimizer": {
        "__version": 0,
        "optimizer": "ADAFACTOR",
        "adam_w_mode": false,
        "alpha": null,
        "amsgrad": false,
        "beta1": null,
        "beta2": null,
        "beta3": null,
        "bias_correction": false,
        "block_wise": false,
        "capturable": false,
        "centered": false,
        "clip_threshold": 1.0,
        "d0": null,
        "d_coef": null,
        "dampening": null,
        "decay_rate": -0.8,
        "decouple": false,
        "differentiable": false,
        "eps": 1e-30,
        "eps2": 0.001,
        "foreach": false,
        "fsdp_in_use": false,
        "fused": false,
        "fused_back_pass": true,
        "growth_rate": null,
        "initial_accumulator_value": null,
        "is_paged": false,
        "log_every": null,
        "lr_decay": null,
        "max_unorm": null,
        "maximize": false,
        "min_8bit_size": null,
        "momentum": null,
        "nesterov": false,
        "no_prox": false,
        "optim_bits": null,
        "percentile_clipping": null,
        "r": null,
        "relative_step": false,
        "safeguard_warmup": false,
        "scale_parameter": false,
        "stochastic_rounding": true,
        "use_bias_correction": false,
        "use_triton": false,
        "warmup_init": false,
        "weight_decay": 0.0,
        "weight_lr_power": null
    },
    "optimizer_defaults": {
        "ADAFACTOR": {
            "__version": 0,
            "optimizer": "ADAFACTOR",
            "adam_w_mode": false,
            "alpha": null,
            "amsgrad": false,
            "beta1": null,
            "beta2": null,
            "beta3": null,
            "bias_correction": false,
            "block_wise": false,
            "capturable": false,
            "centered": false,
            "clip_threshold": 1.0,
            "d0": null,
            "d_coef": null,
            "dampening": null,
            "decay_rate": -0.8,
            "decouple": false,
            "differentiable": false,
            "eps": 1e-30,
            "eps2": 0.001,
            "foreach": false,
            "fsdp_in_use": false,
            "fused": false,
            "fused_back_pass": true,
            "growth_rate": null,
            "initial_accumulator_value": null,
            "is_paged": false,
            "log_every": null,
            "lr_decay": null,
            "max_unorm": null,
            "maximize": false,
            "min_8bit_size": null,
            "momentum": null,
            "nesterov": false,
            "no_prox": false,
            "optim_bits": null,
            "percentile_clipping": null,
            "r": null,
            "relative_step": false,
            "safeguard_warmup": false,
            "scale_parameter": false,
            "stochastic_rounding": true,
            "use_bias_correction": false,
            "use_triton": false,
            "warmup_init": false,
            "weight_decay": 0.0,
            "weight_lr_power": null
        }
    },
    "sample_definition_file_name": "training_samples/samples.json",
    "samples": [
        {
            "__version": 0,
            "enabled": true,
            "prompt": "very aesthetic, best quality, newest, sensitive, 1girl, solo, upper body,",
            "negative_prompt": "monochrome, sketch, fat,",
            "height": 1024,
            "width": 1024,
            "seed": 42,
            "random_seed": true,
            "diffusion_steps": 30,
            "cfg_scale": 7.0,
            "noise_scheduler": "EULER_A"
        }
    ],
    "sample_after": 30,
    "sample_after_unit": "MINUTE",
    "sample_image_format": "JPG",
    "samples_to_tensorboard": true,
    "non_ema_sampling": true,
    "backup_after": 30,
    "backup_after_unit": "MINUTE",
    "rolling_backup": true,
    "rolling_backup_count": 10,
    "backup_before_save": true,
    "save_after": 30,
    "save_after_unit": "MINUTE",
    "save_filename_prefix": ""
}

Limitations and Bias

Bias

• This model is intended for anime illustrations.
  Realistic capabilites are not tested at all.

Limitations

• Can fall back to realistic.
  Add "realistic" tag to the negatives when this happens.
• Far shot eyes can be bad.
• Anatomy and hands can be bad.
• Still in active training.

License

SoteDiffusion models falls under Fair AI Public License 1.0-SD license, which is compatible with Stable Diffusion models’ license. Key points:

1. Modification Sharing: If you modify SoteDiffusion models, you must share both your changes and the original license.
2. Source Code Accessibility: If your modified version is network-accessible, provide a way (like a download link) for others to get the source code. This applies to derived models too.
3. Distribution Terms: Any distribution must be under this license or another with similar rules.
4. Compliance: Non-compliance must be fixed within 30 days to avoid license termination, emphasizing transparency and adherence to open-source values.

Notes: Anything not covered by Fair AI license is inherited from Stability AI Non-Commercial license which is named as LICENSE_INHERIT. Meaning, still no commercial use of any kind.