LunaTranslator/docs/zh/sakurallmkagglecolab.md
恍兮惚兮 c706c95035 docs
2024-09-07 11:39:48 +08:00

86 lines
3.6 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

## 部署SakuraLLM到Kaggle/Google Colab
### 1. 注册[ngrok](https://ngrok.com/)以将请求转发给llama.cpp服务
注册后,分别获取[NGROK_TOKEN](https://dashboard.ngrok.com/get-started/your-authtoken)和[NGROK_DOMAIN](https://dashboard.ngrok.com/cloud-edge/domains),以供后面使用。
<details>
<summary><strong>NGROK_TOKEN</strong></summary>
<img src="https://image.lunatranslator.org/zh/sakurallm/ngrok2.png">
</details>
<details>
<summary><strong>NGROK_DOMAIN</strong></summary>
<img src="https://image.lunatranslator.org/zh/sakurallm/ngrok.png">
</details>
>之后在Sakura大模型的设置中将**API接口地址**填写为`https://`加上**NGROK_DOMAIN**即可,该地址不会发生变化。
### 2. 部署到Kaggle/Google Colab
<!-- tabs:start -->
### **Kaggle**
1. 注册<a href="https://kaggle.com/" target="_blank">Kaggle</a>,导入<a href="https://kaggle.com/kernels/welcome?src=https://lunatranslator.org/nginxfile/kaggle_sakurallm.ipynb" target="_blank">ipynb脚本</a>
<details>
<summary>2. 选择GPU运行时打开网络连接。首次使用需要验证手机号</summary>
<img src="https://image.lunatranslator.org/zh/sakurallm/kaggle.2.png">
<img src="https://image.lunatranslator.org/zh/sakurallm/kaggle.3.png">
</details>
<details>
<summary>3. 设置ngrok密钥和域名以及使用的模型</summary>
将注册的ngrok的NGROK_TOKEN和NGROK_DOMAIN填入脚本中。<br>
REPO和MODEL是<code>https://huggingface.co/REPO</code>下的MODEL模型文件名
<img src="https://image.lunatranslator.org/zh/sakurallm/kaggle.png">
</details>
<details>
<summary>4. 运行脚本,稍微等待一分钟左右即可</summary>
llama.cpp是已经预先编译好的省去了编译的时间因此主要是下载模型需要花费一点时间。
<img src="https://image.lunatranslator.org/zh/sakurallm/kagglerun.png">
</details>
### **Google Colab**
<details>
<summary>1. 在Google drive中安装<strong>Colaboratory</strong>应用</summary>
点击<strong>新建</strong>-><strong>更多</strong>-><strong>关联更多应用</strong>
在应用市场中搜索<strong>Colaboratory</strong>安装即可
<img src="https://image.lunatranslator.org/zh/sakurallm/installcolab.png">
<img src="https://image.lunatranslator.org/zh/sakurallm/installcolab2.png">
</details>
<details>
<summary>2. 打开<a href="https://colab.research.google.com/" target="_blank">Colab</a>,下载<a href="https://lunatranslator.org/nginxfile/kaggle_sakurallm.ipynb" target="_blank">ipynb脚本</a>并上传到Colab中。</summary>
<img src="https://image.lunatranslator.org/zh/sakurallm/colab2.png">
<img src="https://image.lunatranslator.org/zh/sakurallm/colab.png">
</details>
<details>
<summary>3. 选择GPU运行时</summary>
默认是使用CPU运行的需要我们手动切换成T4 GPU运行。
<img src="https://image.lunatranslator.org/zh/sakurallm/colab5.png">
<img src="https://image.lunatranslator.org/zh/sakurallm/colab4.png">
</details>
<details>
<summary>4. 设置ngrok密钥和域名以及使用的模型</summary>
将注册的ngrok的NGROK_TOKEN和NGROK_DOMAIN填入脚本中。
REPO和MODEL是<code>https://huggingface.co/REPO</code>下的MODEL模型文件名
<img src="https://image.lunatranslator.org/zh/sakurallm/colab3.png">
</details>
<details>
<summary>5. 运行脚本,稍微等待一分钟左右即可</summary>
llama.cpp是已经预先编译好的省去了编译的时间因此主要是下载模型需要花费一点时间。
<img src="https://image.lunatranslator.org/zh/sakurallm/colab6.png">
</details>
<!-- tabs:end -->