File size: 8,645 Bytes
6322336 41ee3b0 6322336 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 | ---
license: apache-2.0
---
# Templates - Content Reference (FLUX.2-klein-base-4B)
This model is one of the Diffusion Templates series models open-sourced by [DiffSynth-Studio](https://github.com/modelscope/DiffSynth-Studio). It can extract visual features from an input reference image and fuse them into the base generation guided by natural language descriptions.
* 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.
| Template | Generated | Template | Generated |
|:---:|:---:|:---:|:---:|
|  |  |  |  |
---
> **Prompt:** A cozy wooden cottage in a lush green valley, white fluffy clouds in the sky, peaceful atmosphere.
| Template | Generated | Template | Generated |
|:---:|:---:|:---:|:---:|
|  |  |  |  |
---
> **Prompt:** A beautiful girl on an outdoor adventure.
| Template | Generated | Template | Generated |
|:---:|:---:|:---:|:---:|
|  |  |  |  |
## 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
from modelscope import dataset_snapshot_download
from PIL import Image
import numpy as np
```
```python
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-ContentRef")],
)
dataset_snapshot_download(
"DiffSynth-Studio/examples_in_diffsynth",
allow_file_pattern=["templates/*"],
local_dir="data/examples",
)
image = template(
pipe,
prompt="A cat is sitting on a stone.",
seed=0, cfg_scale=4, num_inference_steps=50,
template_inputs=[{
"image": Image.open("data/examples/templates/image_style_1.jpg"),
}],
negative_template_inputs=[{
"image": Image.fromarray(np.zeros((1024, 1024, 3), dtype=np.uint8) + 128),
}],
)
image.save("image_ContentRef_1.jpg")
image = template(
pipe,
prompt="A cat is sitting on a stone.",
seed=0, cfg_scale=4, num_inference_steps=50,
template_inputs=[{
"image": Image.open("data/examples/templates/image_style_2.jpg"),
}],
negative_template_inputs=[{
"image": Image.fromarray(np.zeros((1024, 1024, 3), dtype=np.uint8) + 128),
}],
)
image.save("image_ContentRef_2.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
from modelscope import dataset_snapshot_download
from PIL import Image
import numpy as np
```
```python
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-ContentRef")],
lazy_loading=True,
)
dataset_snapshot_download(
"DiffSynth-Studio/examples_in_diffsynth",
allow_file_pattern=["templates/*"],
local_dir="data/examples",
)
image = template(
pipe,
prompt="A cat is sitting on a stone.",
seed=0, cfg_scale=4, num_inference_steps=50,
template_inputs=[{
"image": Image.open("data/examples/templates/image_style_1.jpg"),
}],
negative_template_inputs=[{
"image": Image.fromarray(np.zeros((1024, 1024, 3), dtype=np.uint8) + 128),
}],
)
image.save("image_ContentRef_1.jpg")
image = template(
pipe,
prompt="A cat is sitting on a stone.",
seed=0, cfg_scale=4, num_inference_steps=50,
template_inputs=[{
"image": Image.open("data/examples/templates/image_style_2.jpg"),
}],
negative_template_inputs=[{
"image": Image.fromarray(np.zeros((1024, 1024, 3), dtype=np.uint8) + 128),
}],
)
image.save("image_ContentRef_2.jpg")
```
## Training Code
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-ContentRef/*" --local_dir ./data/diffsynth_example_dataset
accelerate launch examples/flux2/model_training/train.py \
--dataset_base_path data/diffsynth_example_dataset/flux2/Template-KleinBase4B-ContentRef \
--dataset_metadata_path data/diffsynth_example_dataset/flux2/Template-KleinBase4B-ContentRef/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-ContentRef:" \
--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-ContentRef_full" \
--trainable_models "template_model" \
--use_gradient_checkpointing \
--find_unused_parameters \
--enable_lora_hot_loading
``` |