来源

综合来自以下文本数据集

• poloclub/diffusiondb
• Gustavosta/Stable-Diffusion-Prompts
• bartman081523/stable-diffusion-discord-prompts
• FredZhang7/krea-ai-prompts

有关预处理方法，请参见 Fast GPT2 PromptGen .

Python

下载并将数据集保存到 all_prompts.txt 本地。

pip install datasets

import datasets

dataset = datasets.load_dataset("FredZhang7/stable-diffusion-prompts-2.47M")

train = dataset["train"]
prompts = train["text"]

with open("all_prompts.txt", "w") as f:
    for prompt in prompts:
        f.write(prompt + "\n")