
# 光影2

## 模型介绍

本模型依托魔搭社区（ModelScope）AIGC专区[模型训练](https://modelscope.cn/aigc/modelTraining)环境与算力完成训练。

* 模型类型：LoRA
* 基础模型：[black-forest-labs/FLUX.2-klein-base-9B](https://modelscope.cn/models/black-forest-labs/FLUX.2-klein-base-9B)
* 训练代码：[DiffSynth-Studio](https://github.com/modelscope/DiffSynth-Studio)
* 训练数据量：98
* 总训练步数：2000
* 开源协议：Apache-2.0

## 推理代码

安装 [DiffSynth-Studio](https://github.com/modelscope/DiffSynth-Studio)：

```bash
pip install diffsynth
```

开始推理：

```python
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-9B", origin_file_pattern="text_encoder/*.safetensors"),
        ModelConfig(model_id="black-forest-labs/FLUX.2-klein-base-9B", origin_file_pattern="transformer/*.safetensors"),
        ModelConfig(model_id="black-forest-labs/FLUX.2-klein-base-9B", origin_file_pattern="vae/diffusion_pytorch_model.safetensors"),
    ],
    tokenizer_config=ModelConfig(model_id="black-forest-labs/FLUX.2-klein-base-9B", origin_file_pattern="tokenizer/"),
)
pipe.load_lora(pipe.dit, ModelConfig(model_id="Lumos123123/jksw2", origin_file_pattern="jksw2_c1-st2000.safetensors"))
prompt = "a cat"
image = pipe(prompt, num_inference_steps=50, cfg_scale=4)
image.save("image.jpg")

prompt = "Convert the image style to anime style."
image = pipe(prompt, edit_image=[image], num_inference_steps=50, cfg_scale=4)
image.save("image_edited.jpg")
```


Files with SHA256: b5929132e4449eed2a907af9e5cf2a3d26ed906b3960b74d5d76731327ba6c2e