File size: 8,194 Bytes
0c4cb0e
 
 
 
 
 
 
35d8195
 
 
 
 
 
 
 
a389631
0c4cb0e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a389631
0c4cb0e
 
 
 
 
cb54d10
0c4cb0e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a389631
0c4cb0e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
---
license: apache-2.0
---
# Templates - Aesthetic Alignment (FLUX.2-klein-base-4B)

This model is one of the open-source Diffusion Templates series models from [DiffSynth-Studio](https://github.com/modelscope/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](https://github.com/modelscope/DiffSynth-Studio)
* Technical report: [arXiv](https://arxiv.org/abs/2604.24351)
* Project page: [GitHub](https://modelscope.github.io/diffusion-templates-web/)
* Documentation: [English Version](https://diffsynth-studio-doc.readthedocs.io/en/latest/Diffusion_Templates/Introducing_Diffusion_Templates.html)、[中文版](https://diffsynth-studio-doc.readthedocs.io/zh-cn/latest/Diffusion_Templates/Introducing_Diffusion_Templates.html)
* Online demo: [ModelScope](https://modelscope.cn/studios/DiffSynth-Studio/Diffusion-Templates)
* Models: [ModelScope](https://modelscope.cn/collections/DiffSynth-Studio/KleinBase4B-Templates)、[ModelScope International](https://modelscope.ai/collections/DiffSynth-Studio/KleinBase4B-Templates)、[HuggingFace](https://huggingface.co/collections/DiffSynth-Studio/kleinbase4b-templates)
* Datasets: [ModelScope](https://modelscope.cn/collections/DiffSynth-Studio/ImagePulseV2)、[ModelScope International](https://modelscope.cn/collections/DiffSynth-Studio/ImagePulseV2)、[HuggingFace](https://huggingface.co/collections/DiffSynth-Studio/imagepulsev2)

## Results

> **Prompt:** A cat is sitting on a stone.

| base model | scale=1.0 | scale=2.5 |
|:---:|:---:|:---:|
| ![](./assets/cat_base.jpg) | ![](./assets/cat_Aesthetic_1.0.jpg) | ![](./assets/cat_Aesthetic_2.5.jpg) |

---

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

| base model | scale=1.0 | scale=2.5 |
|:---:|:---:|:---:|
| ![](./assets/girl_base.jpg) | ![](./assets/girl_Aesthetic_1.0.jpg) | ![](./assets/girl_Aesthetic_2.5.jpg) |

---

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

| base model | scale=1.0 | scale=2.5 |
|:---:|:---:|:---:|
| ![](./assets/apartment_base.jpg) | ![](./assets/apartment_Aesthetic_1.0.jpg) | ![](./assets/apartment_Aesthetic_2.5.jpg) |

## Inference Code

* Install [DiffSynth-Studio](https://github.com/modelscope/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](https://diffsynth-studio-doc.readthedocs.io/zh-cn/latest/).

```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
```