i18n(mermaid): 337 个 mermaid 图表标签中文化

120 个 zh.md 文件中的 mermaid 图表标签从英文翻译为中文。
规则：自然语言标签翻中文，专业技术术语（agent、token、embedding、
transformer、softmax、LLM、MCP 等）保留英文。
仅修改 ```mermaid 块内部内容，正文不变。

Author: fancyboi999

phases/00-setup-and-tooling/02-git-and-collaboration/docs/zh.md

@@ -24,10 +24,10 @@ Git 是工具，GitHub 是代码存放的地方。这节课只讲本课程需要
 
 ```mermaid
 sequenceDiagram
-    participant WD as Working Directory
-    participant SA as Staging Area
-    participant LR as Local Repo
-    participant R as Remote (GitHub)
+    participant WD as 工作目录
+    participant SA as 暂存区
+    participant LR as 本地仓库
+    participant R as 远端 (GitHub)
     WD->>SA: git add
     SA->>LR: git commit
     LR->>R: git push

phases/00-setup-and-tooling/04-apis-and-keys/docs/zh.md

@@ -22,10 +22,10 @@
 
 ```mermaid
 sequenceDiagram
-    participant C as Your Code
-    participant S as API Server
-    C->>S: HTTP Request (with API key)
-    S->>C: HTTP Response (JSON)
+    participant C as 你的代码
+    participant S as API 服务器
+    C->>S: HTTP 请求 (携带 API key)
+    S->>C: HTTP 响应 (JSON)
 ```
 
 每一次 API 调用都包含：

phases/00-setup-and-tooling/05-jupyter-notebooks/docs/zh.md

@@ -34,10 +34,10 @@ graph TD
 
 ```mermaid
 graph LR
-    A[Notebook UI] <--> B[Kernel\nPython process]
-    B --> C[Keeps variables in memory]
-    B --> D[Runs cells in whatever order you click]
-    B --> E[Dies when you restart it]
+    A[Notebook UI] <--> B[Kernel\nPython 进程]
+    B --> C[变量保留在内存中]
+    B --> D[你点哪个 cell 就跑哪个]
+    B --> E[重启就没了]
 ```
 
 那个「你点哪个就跑哪个」的特性，既是超能力，也是脚下的地雷。

phases/00-setup-and-tooling/06-python-environments/docs/zh.md

@@ -31,16 +31,16 @@
 
 ```mermaid
 graph TD
-    subgraph without["Without virtual environments"]
-        SP[System Python] --> T24["torch 2.4.0 (CUDA 12.4)\nProject A needs this"]
-        SP --> T21["torch 2.1.0 (CUDA 11.8)\nProject B needs this"]
-        SP --> CONFLICT["CONFLICT: only one\ntorch version can exist"]
+    subgraph without["没有虚拟环境"]
+        SP[系统 Python] --> T24["torch 2.4.0 (CUDA 12.4)\n项目 A 需要这个"]
+        SP --> T21["torch 2.1.0 (CUDA 11.8)\n项目 B 需要这个"]
+        SP --> CONFLICT["冲突：只能有一个\ntorch 版本"]
     end
 
-    subgraph with["With virtual environments"]
-        PA["Project A (.venv/)"] --> PA1["torch 2.4.0 (CUDA 12.4)"]
+    subgraph with["有虚拟环境"]
+        PA["项目 A (.venv/)"] --> PA1["torch 2.4.0 (CUDA 12.4)"]
         PA --> PA2["transformers 4.44"]
-        PB["Project B (.venv/)"] --> PB1["torch 2.1.0 (CUDA 11.8)"]
+        PB["项目 B (.venv/)"] --> PB1["torch 2.1.0 (CUDA 11.8)"]
         PB --> PB2["diffusers 0.28"]
     end
 ```

phases/00-setup-and-tooling/07-docker-for-ai/docs/zh.md

@@ -26,16 +26,16 @@ Docker 把你的代码、运行时、库和系统工具包进一个叫容器（c
 
 ```mermaid
 graph TD
-    subgraph without["Without Docker"]
-        A1["Your machine<br/>Python 3.12<br/>CUDA 12.4<br/>PyTorch 2.3"] -->|crashes| X1["???"]
-        A2["Their machine<br/>Python 3.10<br/>CUDA 11.8<br/>PyTorch 2.1"] -->|crashes| X2["???"]
-        A3["Server<br/>Python 3.11<br/>CUDA 12.1<br/>PyTorch 2.2"] -->|crashes| X3["???"]
+    subgraph without["没有 Docker"]
+        A1["你的机器<br/>Python 3.12<br/>CUDA 12.4<br/>PyTorch 2.3"] -->|崩了| X1["???"]
+        A2["同事的机器<br/>Python 3.10<br/>CUDA 11.8<br/>PyTorch 2.1"] -->|崩了| X2["???"]
+        A3["服务器<br/>Python 3.11<br/>CUDA 12.1<br/>PyTorch 2.2"] -->|崩了| X3["???"]
     end
 
-    subgraph with_docker["With Docker — Same image everywhere"]
-        B1["Your machine<br/>Python 3.12 | CUDA 12.4<br/>PyTorch 2.3 | Your code"]
-        B2["Their machine<br/>Python 3.12 | CUDA 12.4<br/>PyTorch 2.3 | Your code"]
-        B3["Server<br/>Python 3.12 | CUDA 12.4<br/>PyTorch 2.3 | Your code"]
+    subgraph with_docker["有 Docker —— 到处同一个镜像"]
+        B1["你的机器<br/>Python 3.12 | CUDA 12.4<br/>PyTorch 2.3 | 你的代码"]
+        B2["同事的机器<br/>Python 3.12 | CUDA 12.4<br/>PyTorch 2.3 | 你的代码"]
+        B3["服务器<br/>Python 3.12 | CUDA 12.4<br/>PyTorch 2.3 | 你的代码"]
     end
 ```
 

phases/00-setup-and-tooling/08-editor-setup/docs/zh.md

@@ -26,11 +26,11 @@
 
 ```mermaid
 graph TD
-    L5["5. Remote Development<br/>SSH into GPU boxes, cloud VMs"] --> L4
-    L4["4. Terminal Integration<br/>Run scripts, debug, monitor GPU"] --> L3
-    L3["3. AI-Specific Settings<br/>Auto-format, type checking, rulers"] --> L2
-    L2["2. Extensions<br/>Python, Jupyter, Pylance, GitLens"] --> L1
-    L1["1. Base Editor<br/>VS Code — free, extensible, universal"]
+    L5["5. 远程开发<br/>SSH 进 GPU 机器、云 VM"] --> L4
+    L4["4. 终端集成<br/>跑脚本、调试、监控 GPU"] --> L3
+    L3["3. AI 专属设置<br/>自动格式化、类型检查、标尺"] --> L2
+    L2["2. 扩展<br/>Python, Jupyter, Pylance, GitLens"] --> L1
+    L1["1. 基础编辑器<br/>VS Code —— 免费、可扩展、通用"]
 ```
 
 ## 动手构建

phases/00-setup-and-tooling/09-data-management/docs/zh.md

@@ -22,12 +22,12 @@
 
 ```mermaid
 graph TD
-    A["Hugging Face Hub"] --> B["datasets library"]
-    B --> C["Load / Stream"]
-    C --> D["Local Cache<br/>~/.cache/huggingface/"]
-    B --> E["Format Conversion<br/>CSV, JSON, Parquet, Arrow"]
-    E --> F["Data Splits<br/>train / val / test"]
-    F --> G["Your Training Pipeline"]
+    A["Hugging Face Hub"] --> B["datasets 库"]
+    B --> C["加载 / 流式读取"]
+    C --> D["本地缓存<br/>~/.cache/huggingface/"]
+    B --> E["格式转换<br/>CSV, JSON, Parquet, Arrow"]
+    E --> F["数据划分<br/>train / val / test"]
+    F --> G["你的训练流水线"]
 ```
 
 Hugging Face 的 `datasets` 库是 AI 工作里加载数据的标准方式。它开箱就处理好下载、缓存、格式转换和流式读取。

phases/00-setup-and-tooling/10-terminal-and-shell/docs/zh.md

@@ -24,12 +24,12 @@
 
 ```mermaid
 graph TD
-    subgraph tmux["tmux session: training"]
-        subgraph top["Top row"]
-            P1["Pane 1: Training run<br/>python train.py<br/>Epoch 12/100 ..."]
-            P2["Pane 2: GPU monitor<br/>watch -n1 nvidia-smi<br/>GPU: 78% | Mem: 14/24G"]
+    subgraph tmux["tmux 会话: training"]
+        subgraph top["上半部分"]
+            P1["窗格 1: 训练任务<br/>python train.py<br/>Epoch 12/100 ..."]
+            P2["窗格 2: GPU 监控<br/>watch -n1 nvidia-smi<br/>GPU: 78% | Mem: 14/24G"]
         end
-        P3["Pane 3: Logs + experiments<br/>tail -f logs/train.log | grep loss"]
+        P3["窗格 3: 日志 + 实验<br/>tail -f logs/train.log | grep loss"]
     end
 ```
 

phases/00-setup-and-tooling/11-linux-for-ai/docs/zh.md

@@ -26,13 +26,13 @@ Linux 把一切都组织在唯一的根 `/` 下面。没有 `C:\`，没有 `/Vol
 
 ```mermaid
 graph TD
-    root["/"] --> home["home/your-username/<br/>Your files — clone repos, run training"]
-    root --> tmp["tmp/<br/>Temporary files, cleared on reboot"]
-    root --> usr["usr/<br/>System programs and libraries"]
-    root --> etc["etc/<br/>Config files"]
-    root --> varlog["var/log/<br/>Logs — check when something breaks"]
-    root --> mnt["mnt/ or /media/<br/>External drives and volumes"]
-    root --> proc["proc/ and /sys/<br/>Virtual files — kernel and hardware info"]
+    root["/"] --> home["home/your-username/<br/>你的文件 —— 克隆仓库、跑训练"]
+    root --> tmp["tmp/<br/>临时文件，重启后清空"]
+    root --> usr["usr/<br/>系统程序和库"]
+    root --> etc["etc/<br/>配置文件"]
+    root --> varlog["var/log/<br/>日志 —— 出问题时来这看"]
+    root --> mnt["mnt/ 或 /media/<br/>外接硬盘和存储卷"]
+    root --> proc["proc/ 和 /sys/<br/>虚拟文件 —— 内核和硬件信息"]
 ```
 
 你的主目录是 `~` 或 `/home/your-username`。你做的几乎一切都发生在这里。

phases/00-setup-and-tooling/12-debugging-and-profiling/docs/zh.md

@@ -26,9 +26,9 @@ AI 调试在三个层级上运作：
 
 ```mermaid
 graph TD
-    L3["3. Training Dynamics<br/>Loss curves, gradient norms, activations"] --> L2
-    L2["2. Tensor Operations<br/>Shapes, dtypes, devices, NaN/Inf values"] --> L1
-    L1["1. Standard Python<br/>Breakpoints, logging, profiling, memory"]
+    L3["3. 训练动态<br/>损失曲线、梯度范数、激活值"] --> L2
+    L2["2. 张量运算<br/>形状、dtype、设备、NaN/Inf 值"] --> L1
+    L1["1. 标准 Python<br/>断点、日志、性能分析、内存"]
 ```
 
 大多数人直接跳到第 3 层（盯着 TensorBoard 看）。但 80% 的 AI bug 住在第 1 层和第 2 层。