---
title: Qwen-Image-LangAlign-LoRA
canonical_url: "https://www.modelscope.cn/models/DiffSynth-Studio/Qwen-Image-LangAlign-LoRA"
md_url: "https://www.modelscope.cn/models/DiffSynth-Studio/Qwen-Image-LangAlign-LoRA.md"
repository: DiffSynth-Studio/Qwen-Image-LangAlign-LoRA
chinese_name: "Qwen-Image 多语言对齐 LoRA"
last_updated: 2025-11-11
license: "Apache License 2.0"
pipeline_tag: text-to-image-synthesis
tasks:
  - text-to-image-synthesis
base_model:
  - Qwen/Qwen-Image
base_model_relation: adapter
parameters: 235.9M
tensor_type:
  - BF16
library_name:
  - pytorch
  - lora
  - safetensors
supports_inference: txt2img
downloads: 144
stars: 2
tags:
  - LoRA
---

# Qwen-Image-LangAlign-LoRA

> Qwen-Image-LangAlign-LoRA - DiffSynth-Studio 在 ModelScope 开源的模型。Qwen-Image 多语言对齐 LoRA

DiffSynth-Studio/Qwen-Image-LangAlign-LoRA 是 ModelScope 魔搭社区上的 235.9M 参数text-to-image-synthesis模型，采用 Apache License 2.0 许可，基于 Qwen/Qwen-Image 构建，并支持在线推理（txt2img）。

- **Repository**: DiffSynth-Studio/Qwen-Image-LangAlign-LoRA
- **License**: Apache License 2.0
- **Tasks**: text-to-image-synthesis
- **Parameters**: 235.9M
- **Base model**: Qwen/Qwen-Image
- **Online inference**: txt2img
- **Tags**: LoRA
- **Downloads**: 144
- **Stars**: 2
- **Last updated**: 2025-11-11

Source: https://www.modelscope.cn/models/DiffSynth-Studio/Qwen-Image-LangAlign-LoRA

---

# Qwen-Image 多语言对齐 LoRA

## 模型介绍

本模型能够提升 [Qwen-Image](https://modelscope.cn/models/Qwen/Qwen-Image) 的多语言理解能力，使多种语言提示词的生成效果趋近统一。模型采用数据集 [DiffSynth-Studio/Qwen-Image-Self-Generated-Dataset](https://modelscope.cn/datasets/DiffSynth-Studio/Qwen-Image-Self-Generated-Dataset) 及其标注的 150 万多语言提示词进行训练，训练代码基于 [DiffSynth-Studio](https://github.com/modelscope/DiffSynth-Studio/tree/main)。模型支持的语言包括中文、英语、韩语、日语、西班牙语、法语、葡萄牙语、俄语、德语、意大利语。

## 效果展示

中文："一只可爱的柯基小狗"

|||||||
|-|-|-|-|-|-|
|英语："A cute Corgi puppy"|![](./assets/example_1/image_1_lora.jpg)|韩语："귀여운 코기 강아지"|![](./assets/example_1/image_2_lora.jpg)|日语："かわいいコーギーの子犬"|![](./assets/example_1/image_3_lora.jpg)|
|西班牙语："Un cachorro de Corgi adorable"|![](./assets/example_1/image_4_lora.jpg)|法语："Un adorable chiot Corgi"|![](./assets/example_1/image_5_lora.jpg)|葡萄牙语："Um filhote de Corgi fofo"|![](./assets/example_1/image_6_lora.jpg)|
|俄语："Милый щенок корги"|![](./assets/example_1/image_7_lora.jpg)|德语："Ein niedlicher Corgi-Welpe"|![](./assets/example_1/image_8_lora.jpg)|意大利语："Un cucciolo di Corgi carino"|![](./assets/example_1/image_9_lora.jpg)|

中文："一只白色的猫趴在沙发上"

|||||||
|-|-|-|-|-|-|
|英语："A white cat is lying on the sofa."|![](./assets/example_2/image_1_lora.jpg)|韩语："하얀 고양이가 소파 위에 누워 있다."|![](./assets/example_2/image_2_lora.jpg)|日语："白い猫がソファの上で横になっている。"|![](./assets/example_2/image_3_lora.jpg)|
|西班牙语："Un gato blanco está tumbado en el sofá."|![](./assets/example_2/image_4_lora.jpg)|法语："Un chat blanc est allongé sur le canapé."|![](./assets/example_2/image_5_lora.jpg)|葡萄牙语："Um gato branco está deitado no sofá."|![](./assets/example_2/image_6_lora.jpg)|
|俄语："Белый кот лежит на диване."|![](./assets/example_2/image_7_lora.jpg)|德语："Eine weiße Katze liegt auf dem Sofa."|![](./assets/example_2/image_8_lora.jpg)|意大利语："Un gatto bianco è sdraiato sul divano."|![](./assets/example_2/image_9_lora.jpg)|

## 推理代码

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

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

推理代码：

```python
from diffsynth.pipelines.qwen_image import QwenImagePipeline, ModelConfig
import torch

pipe = QwenImagePipeline.from_pretrained(
    torch_dtype=torch.bfloat16,
    device="cuda",
    model_configs=[
        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="transformer/diffusion_pytorch_model*.safetensors"),
        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="text_encoder/model*.safetensors"),
        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="vae/diffusion_pytorch_model.safetensors"),
    ],
    tokenizer_config=ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="tokenizer/"),
)
lora = ModelConfig(model_id="DiffSynth-Studio/Qwen-Image-LangAlign-LoRA", origin_file_pattern="model.safetensors")
pipe.load_lora(pipe.dit, lora, alpha=1)
prompt = "하얀 고양이가 소파 위에 누워 있다."
image = pipe(prompt, seed=0, num_inference_steps=50, height=1024, width=1024)
image.save("image.jpg")
```