You can deploy this model and start earning money today!

DiffSynth-Studio/
Template-KleinBase4B-Aesthetic

Name: Template-KleinBase4B-Aesthetic
Availability: InStock

safetensors

2604.24351

apache-2.0

Views

No views yet

Templates - Aesthetic Alignment (FLUX.2-klein-base-4B)

This model is one of the open-source Diffusion Templates series models from DiffSynth-Studio. It is an Aesthetic alignment model that allows adjusting the degree of aesthetic alignment in generated images by modifying the scale parameter.

Open-source code: DiffSynth-Studio
Technical report: arXiv
Project page: GitHub
Documentation: English Version、中文版
Online demo: ModelScope
Models: ModelScope、ModelScope International、HuggingFace
Datasets: ModelScope、ModelScope International、HuggingFace

Results

Prompt: A cat is sitting on a stone.

base model	scale=1.0	scale=2.5

Prompt: A cute anime girl with pink hair and cat ears, pastel colors.

base model	scale=1.0	scale=2.5

Prompt: A cyberpunk apartment with a view of neon lights.

base model	scale=1.0	scale=2.5

Inference Code

Install DiffSynth-Studio

git clone https://github.com/modelscope/DiffSynth-Studio.git
cd DiffSynth-Studio
pip install -e .

Direct inference (requires 40G GPU memory)

python

from diffsynth.diffusion.template import TemplatePipeline
from diffsynth.pipelines.flux2_image import Flux2ImagePipeline, ModelConfig
import torch


pipe = Flux2ImagePipeline.from_pretrained(
    torch_dtype=torch.bfloat16,
    device="cuda",
    model_configs=[
        ModelConfig(model_id="black-forest-labs/FLUX.2-klein-base-4B", origin_file_pattern="transformer/*.safetensors"),
        ModelConfig(model_id="black-forest-labs/FLUX.2-klein-4B", origin_file_pattern="text_encoder/*.safetensors"),
        ModelConfig(model_id="black-forest-labs/FLUX.2-klein-4B", origin_file_pattern="vae/diffusion_pytorch_model.safetensors"),
    ],
    tokenizer_config=ModelConfig(model_id="black-forest-labs/FLUX.2-klein-4B", origin_file_pattern="tokenizer/"),
)
pipe.dit = pipe.enable_lora_hot_loading(pipe.dit) # Important!
template = TemplatePipeline.from_pretrained(
    torch_dtype=torch.bfloat16,
    device="cuda",
    model_configs=[ModelConfig(model_id="DiffSynth-Studio/Template-KleinBase4B-Aesthetic")],
)
image = template(
    pipe,
    prompt="A cat is sitting on a stone.",
    seed=0, cfg_scale=4, num_inference_steps=50,
    template_inputs = [{
        "lora_ids": list(range(1, 180, 2)),
        "lora_scales": 1.0,
        "merge_type": "mean",
    }],
    negative_template_inputs = [{
        "lora_ids": list(range(1, 180, 2)),
        "lora_scales": 1.0,
        "merge_type": "mean",
    }],
)
image.save("image_Aesthetic_1.0.jpg")
image = template(
    pipe,
    prompt="A cat is sitting on a stone.",
    seed=0, cfg_scale=4, num_inference_steps=50,
    template_inputs = [{
        "lora_ids": list(range(1, 180, 2)),
        "lora_scales": 2.5,
        "merge_type": "mean",
    }],
    negative_template_inputs = [{
        "lora_ids": list(range(1, 180, 2)),
        "lora_scales": 2.5,
        "merge_type": "mean",
    }],
)
image.save("image_Aesthetic_2.5.jpg")

Enable lazy loading and memory management, requires 24G GPU memory

python

from diffsynth.diffusion.template import TemplatePipeline
from diffsynth.pipelines.flux2_image import Flux2ImagePipeline, ModelConfig
import torch

vram_config = {
    "offload_dtype": "disk",
    "offload_device": "disk",
    "onload_dtype": torch.float8_e4m3fn,
    "onload_device": "cpu",
    "preparing_dtype": torch.float8_e4m3fn,
    "preparing_device": "cuda",
    "computation_dtype": torch.bfloat16,
    "computation_device": "cuda",
}
pipe = Flux2ImagePipeline.from_pretrained(
    torch_dtype=torch.bfloat16,
    device="cuda",
    model_configs=[
        ModelConfig(model_id="black-forest-labs/FLUX.2-klein-base-4B", origin_file_pattern="transformer/*.safetensors", **vram_config),
        ModelConfig(model_id="black-forest-labs/FLUX.2-klein-4B", origin_file_pattern="text_encoder/*.safetensors", **vram_config),
        ModelConfig(model_id="black-forest-labs/FLUX.2-klein-4B", origin_file_pattern="vae/diffusion_pytorch_model.safetensors"),
    ],
    tokenizer_config=ModelConfig(model_id="black-forest-labs/FLUX.2-klein-4B", origin_file_pattern="tokenizer/"),
    vram_limit=torch.cuda.mem_get_info("cuda")[1] / (1024 ** 3) - 0.5,
)
template = TemplatePipeline.from_pretrained(
    torch_dtype=torch.bfloat16,
    device="cuda",
    model_configs=[ModelConfig(model_id="DiffSynth-Studio/Template-KleinBase4B-Aesthetic")],
    lazy_loading=True,
)
image = template(
    pipe,
    prompt="A cat is sitting on a stone.",
    seed=0, cfg_scale=4, num_inference_steps=50,
    template_inputs = [{
        "lora_ids": list(range(1, 180, 2)),
        "lora_scales": 1.0,
        "merge_type": "mean",
    }],
    negative_template_inputs = [{
        "lora_ids": list(range(1, 180, 2)),
        "lora_scales": 1.0,
        "merge_type": "mean",
    }],
)
image.save("image_Aesthetic_1.0.jpg")
image = template(
    pipe,
    prompt="A cat is sitting on a stone.",
    seed=0, cfg_scale=4, num_inference_steps=50,
    template_inputs = [{
        "lora_ids": list(range(1, 180, 2)),
        "lora_scales": 2.5,
        "merge_type": "mean",
    }],
    negative_template_inputs = [{
        "lora_ids": list(range(1, 180, 2)),
        "lora_scales": 2.5,
        "merge_type": "mean",
    }],
)
image.save("image_Aesthetic_2.5.jpg")

Training Script

After installing DiffSynth-Studio, use the following script to start training. For more information, please refer to the DiffSynth-Studio Documentation.

shell

modelscope download --dataset DiffSynth-Studio/diffsynth_example_dataset --include "flux2/Template-KleinBase4B-Aesthetic/*" --local_dir ./data/diffsynth_example_dataset

accelerate launch examples/flux2/model_training/train.py \
  --dataset_base_path data/diffsynth_example_dataset/flux2/Template-KleinBase4B-Aesthetic \
  --dataset_metadata_path data/diffsynth_example_dataset/flux2/Template-KleinBase4B-Aesthetic/metadata.jsonl \
  --extra_inputs "template_inputs" \
  --max_pixels 1048576 \
  --dataset_repeat 50 \
  --model_id_with_origin_paths "black-forest-labs/FLUX.2-klein-4B:text_encoder/*.safetensors,black-forest-labs/FLUX.2-klein-base-4B:transformer/*.safetensors,black-forest-labs/FLUX.2-klein-4B:vae/diffusion_pytorch_model.safetensors" \
  --template_model_id_or_path "DiffSynth-Studio/Template-KleinBase4B-Aesthetic:" \
  --tokenizer_path "black-forest-labs/FLUX.2-klein-4B:tokenizer/" \
  --learning_rate 1e-4 \
  --num_epochs 2 \
  --remove_prefix_in_ckpt "pipe.template_model." \
  --output_path "./models/train/Template-KleinBase4B-Aesthetic_full" \
  --trainable_models "template_model" \
  --use_gradient_checkpointing \
  --find_unused_parameters \
  --enable_lora_hot_loading