| --- |
| license: apache-2.0 |
| --- |
| ### 我们的网站 |
| ### Our website |
| [turingevo.com](https://turingevo.com) |
|
|
|
|
| 欢迎使用我们的产品——开源免费的多模态AI笔记 |
|
|
| Welcome to use our product - an open-source and free multimodal AI note-taking tool. |
|
|
| ### 本模型使用 [stable-diffusion.cpp](https://github.com/akleine/stable-diffusion.cpp/) 从 [turingevo/tiny-sd-safetensors](https://huggingface.co/turingevo/tiny-sd-safetensors) 转换 |
| ### This model uses [stable-diffusion.cpp](https://github.com/akleine/stable-diffusion.cpp/) to convert from [turingevo/tiny-sd-safetensors](https://huggingface.co/turingevo/tiny-sd-safetensors). |
|
|
| ### stable-diffusion.cpp 直接运行此模型 |
| ### stable-diffusion.cpp can be directly run to execute this model |
|
|
|
|
| ``` |
| #!/bin/bash |
| |
| # 设置模型路径和输出路径 |
| TYPE='q4_K' |
| #TYPE='q8_0' |
| MODEL_DIR="/media/wmx/soft2/huggingface_cache/segmind/tiny-sd-gguf/segmind_tiny-sd-${TYPE}.gguf" |
| OUTPUT_IMAGE="tiny_sd_output-${TYPE}.png" |
| |
| # Tiny-SD 本质是极小化的 SD 1.5 架构 |
| # 尺寸必须是 512x512 |
| # 采样器使用默认的 euler_a (或者 dpm++2m) |
| # CFG 7.0 是 SD 1.5 的标准值 |
| sd-cli \ |
| -m "$MODEL_DIR" \ |
| -W 512 -H 512 \ |
| --cfg-scale 7.0 \ |
| --steps 20 \ |
| --sampling-method euler_a \ |
| --diffusion-fa \ |
| -v \ |
| -o "$OUTPUT_IMAGE" \ |
| -p "a cute car, high quality" |
| |
| echo "" |
| if [ ! -f "$OUTPUT_IMAGE" ]; then |
| echo "[错误] 图像生成失败，请检查上面的错误日志。" |
| exit 1 |
| fi |
| echo "生成完毕！图片保存在: $OUTPUT_IMAGE" |
| ``` |
|
|
|
|
| q4_K: |
|  |
| |
| q8_0: |
|  |
|
|
|
|
|
|
|
|
|
|
