Skip to content

Commit

Permalink
add model-hub docs
Browse files Browse the repository at this point in the history
  • Loading branch information
windsonsea committed Mar 20, 2024
1 parent a7dc33c commit d618a70
Show file tree
Hide file tree
Showing 12 changed files with 103 additions and 1 deletion.
Binary file added docs/zh/docs/dmc/images/inner01.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added docs/zh/docs/dmc/images/inner02.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added docs/zh/docs/dmc/images/inner03.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added docs/zh/docs/dmc/images/tuner01.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
2 changes: 1 addition & 1 deletion docs/zh/docs/dmc/index.md
Original file line number Diff line number Diff line change
Expand Up @@ -7,6 +7,6 @@ hide:

d.run 模型中心内置了 **模型仓库** ,可以一站式管理各类大模型,按需部署。

另外通过模型服务,您可以从本地、在线提供大语言模型服务、向量化模型服务。
另外通过 **模型服务** ,您可以从本地、在线提供大语言模型服务、向量化模型服务。

![model hub](./images/dmc-home.png)
26 changes: 26 additions & 0 deletions docs/zh/docs/dmc/model-hub/built-in.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,26 @@
---
hide:
- toc
---

# 内置模型

内置模型是 d.run 按照规范适配验证后上线到模型中心的,您可以直接部署。

如果想要部署某个模型,

1. 在模型卡片右下角,点击 **部署** 按钮。

![点击部署按钮](../images/inner02.png)

2. 填写各项参数后点击 **确定**

![填写参数](../images/inner03.png)

!!! info "支持国产 GPU"

其中算力类型支持 Nvidia GPU 和 Ascend 等国产 GPU。

3. 屏幕提示创建成功,接下来可以通过部署的模型提供服务。

下一步:[模型服务](../model-service/index.md)
18 changes: 18 additions & 0 deletions docs/zh/docs/dmc/model-hub/index.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,18 @@
---
hide:
- toc
---

# 模型仓库

模型仓库中存放了已经适配好、经验证可以直接使用的大模型。
目前包括 2 种模型:

- [内置模型](./built-in.md)
- [微调模型](./tuner.md)

## 切换集群

在导航栏左上角,可以切换集群/算力中心。

![切换集群](../images/inner01.png)
38 changes: 38 additions & 0 deletions docs/zh/docs/dmc/model-hub/tuner.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,38 @@
---
hide:
- toc
---

# 微调模型

微调模型是指上线到模型仓库后,又通过 [DataTunerX](https://github.com/DataTunerX/datatunerx) 等工具的微调实验得出的模型。

## 部署

1. 点击某个模型右侧的 **...**,在弹出菜单中选择 **部署**

![点击部署按钮](../images/tuner01.png)

2. 填写各项参数后点击 **确定**

![填写参数](../images/inner03.png)

!!! info "支持国产 GPU"

其中算力类型支持 Nvidia GPU 和 Ascend 等国产 GPU。

3. 屏幕提示创建成功,接下来可以通过部署的模型提供服务。

下一步:[模型服务](../model-service/index.md)

## 删除

1. 点击某个模型右侧的 **...**,在弹出菜单中选择 **删除**

![点击删除按钮](../images/inner02.png)

2. 输入模型名称,确认无误后删除。

!!! note

删除操作不可逆,请谨慎操作。
12 changes: 12 additions & 0 deletions docs/zh/docs/dmc/model-service/index.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,12 @@
# 模型服务

模型服务从地理位置可分为:

- 本地模型
- 在线模型

从类型上可划分为:

- 大语言模型:即 Large Language Model (LLM),这是基于大量数据进行预训练的超大型深度学习模型
- 向量化模型:这种模型将数据表示成计算机可识别的实数向量(vector),根据粒度大小不同可将数据特征表示分为字、词、句子或篇章几个层次。
数据向量化的方法主要分为离散表示和分布式表示。也就是说,数据向量化是将原始数据转换为数值向量的过程,以便计算机可以理解和处理数据。
1 change: 1 addition & 0 deletions docs/zh/docs/dmc/model-service/local.md
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
# 本地模型服务
1 change: 1 addition & 0 deletions docs/zh/docs/dmc/model-service/online.md
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
# 在线模型服务
6 changes: 6 additions & 0 deletions docs/zh/navigation.yml
Original file line number Diff line number Diff line change
Expand Up @@ -27,6 +27,12 @@ nav:
- 系统配置: dak/user-guide/system-config.md
- 模型中心:
- 什么是模型中心: dmc/index.md
- 模型仓库:
- 内置模型: dmc/model-hub/built-in.md
- 微调模型: dmc/model-hub/tuner.md
- 模型服务:
- 本地模型服务: dmc/model-service/local.md
- 在线模型服务: dmc/model-service/online.md
- 模型开发:
- 什么是模型开发: baize/index.md
- 模型微调:
Expand Down

0 comments on commit d618a70

Please sign in to comment.