基于昇思+昇腾，玩转 CogView4生成汉字！

2025年3月4日，智谱开源了首个支持生成汉字的开源文生图模型 CogView4。该模型具有6B权重，支持原生中文输入，支持中文文字绘画。CogView4 在 DPG-Bench 基准测试中的综合评分排名第一，在开源文生图模型中达到 SOTA 水平。

昇思MindSpore 团队快速支持 CogView4 Diffusers 工作流并开源至MindSpore ONE仓库，结合昇腾硬件，为开发者提供简单易用体验。本文将介绍如何基于昇思MindSpore和单机Atlas 800T A2，通过简单命令，单卡玩转 CogView4 推理。

开源链接：

https://github.com/mindspore-lab/mindone/tree/master/examples/cogview

魔乐社区在线体验：

https://modelers.cn/spaces/MindSpore-Lab/cogview4

# 01****效果展示

# 02****快速玩转CogView4

1、环境准备

• CANN下载：
• https://www.hiascend.com/developer/download/community/result
• MindSpore下载：
• https://www.mindspore.cn/install

2、安装依赖

git clone https://github.com/mindspore-lab/mindone

# install mindone
cd mindone
pip install -e .

cd examples/cogview

3、启动推理

无需提前准备模型权重，Diffusers 工作流根据脚本给定的模型名称自动从 huggingface 获取权重！以下单行命令即可快速启动推理！

python inference/cli_demo_cogview4.py --prompt {your prompt}

详细推理命令可参考以下样例。可使用 BF16 推理。

from mindone.diffusers import CogView4Pipeline
import mindspore as ms

pipe = CogView4Pipeline.from_pretrained("THUDM/CogView4-6B", mindspore_dtype=ms.bfloat16)

pipe.vae.enable_slicing()
pipe.vae.enable_tiling()

prompt = "A vibrant cherry red sports car sits proudly under the gleaming sun, its polished exterior smooth and flawless, casting a mirror-like reflection. The car features a low, aerodynamic body, angular headlights that gaze forward like predatory eyes, and a set of black, high-gloss racing rims that contrast starkly with the red. A subtle hint of chrome embellishes the grille and exhaust, while the tinted windows suggest a luxurious and private interior. The scene conveys a sense of speed and elegance, the car appearing as if it's about to burst into a sprint along a coastal road, with the ocean's azure waves crashing in the background."
image = pipe(
    prompt=prompt,
    guidance_scale=3.5,
    num_images_per_prompt=1,
    num_inference_steps=50,
    width=1024,
    height=1024,
)[0][0]

image.save("cogview4.png")

4、性能实测

快来体验吧！