docs(tutorials): add Runnable Code section to 3 paired tutorials

Bidirectional discoverability between tutorials and the new code/
contributions in docs/tutorials/code/:

- attention_tutorial.md (+ _en) -> mha.py, axial_attention.py
- flow_matching_tutorial.md (+ _en) -> flow_matching.py
- image_generation_systems_tutorial.md (+ _en) -> mmdit_block.py,
  toy_mmdit_t2i_pipeline.py

Each tutorial gains a short "Runnable Code" section at the bottom
pointing to the paired scripts and to code/README.md. HTML
regenerated via tools/render_html.py with original title/eyebrow/
subtitle preserved.

Author: wanshuiyin

docs/tutorials/attention_tutorial.html

@@ -7,8 +7,8 @@
 
 <meta name="generator" content="ARIS render-html (academic, v1)">
 <meta name="aris:source-path" content="docs/tutorials/attention_tutorial.md">
-<meta name="aris:source-sha256" content="8d8d5146c1a9a924cd5e29973b7419d9fe14f5009c392db4b4ad6ff3a12ab1e0">
-<meta name="aris:generated-at" content="2026-05-19 03:40 UTC">
+<meta name="aris:source-sha256" content="970e8a0a0538e0022a8bf1ec44765f3d6a12987ed2d4c322b034e10eb1c3817e">
+<meta name="aris:generated-at" content="2026-05-26 11:27 UTC">
 
 <!-- MathJax 3 -->
 <script>
@@ -380,6 +380,8 @@ <h3>Contents</h3>
 </li>
 <li><a href="#a-附录完整-from-scratch-代码骨架">§A 附录：完整 from-scratch 代码骨架</a>
 </li>
+<li><a href="#runnable-code">📜 Runnable Code</a>
+</li>
 </ol>
 </nav>
 
@@ -388,11 +390,11 @@ <h3>Contents</h3>
   <div class="eyebrow">Interview Prep · Attention / MHA / Cross-Attention</div>
   <h1>Attention 面试 Cheat Sheet</h1>
   <p class="subtitle">公式推导 + From-Scratch 代码 + 25 高频题（L1 必会 · L2 进阶 · L3 顶级 lab）</p>
-  <p class="byline">By <strong>Ruofeng Yang (杨若峰), Shanghai Jiao Tong University</strong></p>
+  
   <div class="meta">
     <span><strong>Source:</strong> <code>docs/tutorials/attention_tutorial.md</code></span>
-    <span><strong>SHA256:</strong> <code>8d8d5146c1a9</code></span>
-    <span><strong>Rendered:</strong> 2026-05-19 03:40 UTC</span>
+    <span><strong>SHA256:</strong> <code>970e8a0a0538</code></span>
+    <span><strong>Rendered:</strong> 2026-05-26 11:27 UTC</span>
 
   </div>
 </header>
@@ -759,12 +761,17 @@ <h2 id="a-附录完整-from-scratch-代码骨架">§A 附录：完整 from-scrat
 [h] d_model not divisible by num_heads -&gt; ValueError  ✓
 [i] return_weights=False -&gt; weights is None  ✓</code></pre>
 <p>代码经独立 reviewer 静态检查 + PyTorch 实跑 sanity check，与 <code>nn.MultiheadAttention</code> diff = 0。</p>
+<hr />
+<h2 id="runnable-code">📜 Runnable Code</h2>
+<p>本 tutorial 的核心概念在 <a href="code/"><code>docs/tutorials/code/</code></a> 里有最小可跑的 PyTorch 实现：</p>
+<ul><li><a href="code/mha.py"><code>mha.py</code></a> — 标准 Multi-Head Self-Attention + causal mask + 跟 <code>nn.MultiheadAttention</code> 数值对齐验证</li><li><a href="code/axial_attention.py"><code>axial_attention.py</code></a> — H/W 轴向 attention + 复杂度对比表 + 感受野隔离测试</li></ul>
+<p>每个脚本默认 CPU 几秒跑完，自带 <code>assert</code> sanity check。完整说明见 <a href="code/README.md"><code>code/README.md</code></a>。</p>
 
 <footer class="aris-footer">
   Generated by <a href="https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep/blob/main/skills/render-html/SKILL.md">ARIS <code>/render-html</code></a> ·
   source path <code>docs/tutorials/attention_tutorial.md</code> ·
-  SHA256 <code>8d8d5146c1a9</code> ·
-  generated at 2026-05-19 03:40 UTC.
+  SHA256 <code>970e8a0a0538</code> ·
+  generated at 2026-05-26 11:27 UTC.
   This is a generated view — edit the source Markdown, then re-render.
 </footer>
 </main>

docs/tutorials/attention_tutorial.md

@@ -810,3 +810,14 @@ FP16 下直接手写 naive attention 不做 fp32 accumulation。
 ```
 
 代码经独立 reviewer 静态检查 + PyTorch 实跑 sanity check，与 `nn.MultiheadAttention` diff = 0。
+
+---
+
+## 📜 Runnable Code
+
+本 tutorial 的核心概念在 [`docs/tutorials/code/`](code/) 里有最小可跑的 PyTorch 实现：
+
+- [`mha.py`](code/mha.py) — 标准 Multi-Head Self-Attention + causal mask + 跟 `nn.MultiheadAttention` 数值对齐验证
+- [`axial_attention.py`](code/axial_attention.py) — H/W 轴向 attention + 复杂度对比表 + 感受野隔离测试
+
+每个脚本默认 CPU 几秒跑完，自带 `assert` sanity check。完整说明见 [`code/README.md`](code/README.md)。

docs/tutorials/attention_tutorial_en.html

@@ -1,14 +1,14 @@
 <!DOCTYPE html>
-<html lang="zh-CN">
+<html lang="en">
 <head>
 <meta charset="UTF-8">
 <meta name="viewport" content="width=device-width, initial-scale=1.0">
 <title>Attention Tutorial En</title>
 
 <meta name="generator" content="ARIS render-html (academic, v1)">
 <meta name="aris:source-path" content="docs/tutorials/attention_tutorial_en.md">
-<meta name="aris:source-sha256" content="ecb6fe924cee8db5b2dd0daa4fb81c3866ec6b47ceadaa73ba0011378eb07e67">
-<meta name="aris:generated-at" content="2026-05-19 18:47 UTC">
+<meta name="aris:source-sha256" content="52fb9664c6793bc81ab2abdec83fbd0978574b6e370057af74ae62525bacf33a">
+<meta name="aris:generated-at" content="2026-05-26 11:27 UTC">
 
 <!-- MathJax 3 -->
 <script>
@@ -380,6 +380,8 @@ <h3>Contents</h3>
 </li>
 <li><a href="#a-appendix-full-from-scratch-code-skeleton">§A Appendix: Full from-scratch code skeleton</a>
 </li>
+<li><a href="#runnable-code">📜 Runnable Code</a>
+</li>
 </ol>
 </nav>
 
@@ -391,8 +393,8 @@ <h1>Attention Tutorial En</h1>
 
   <div class="meta">
     <span><strong>Source:</strong> <code>docs/tutorials/attention_tutorial_en.md</code></span>
-    <span><strong>SHA256:</strong> <code>ecb6fe924cee</code></span>
-    <span><strong>Rendered:</strong> 2026-05-19 18:47 UTC</span>
+    <span><strong>SHA256:</strong> <code>52fb9664c679</code></span>
+    <span><strong>Rendered:</strong> 2026-05-26 11:27 UTC</span>
 
   </div>
 </header>
@@ -759,12 +761,17 @@ <h2 id="a-appendix-full-from-scratch-code-skeleton">§A Appendix: Full from-scra
 [h] d_model not divisible by num_heads -&gt; ValueError  ✓
 [i] return_weights=False -&gt; weights is None  ✓</code></pre>
 <p>Code passed independent reviewer static check + PyTorch sanity-check run, diff vs <code>nn.MultiheadAttention</code> = 0.</p>
+<hr />
+<h2 id="runnable-code">📜 Runnable Code</h2>
+<p>Minimal runnable PyTorch implementations of this tutorial's core concepts live in <a href="code/"><code>docs/tutorials/code/</code></a>:</p>
+<ul><li><a href="code/mha.py"><code>mha.py</code></a> — standard Multi-Head Self-Attention + causal mask + numerical parity check against <code>nn.MultiheadAttention</code></li><li><a href="code/axial_attention.py"><code>axial_attention.py</code></a> — H/W factorized axial attention + complexity comparison table + receptive-field isolation test</li></ul>
+<p>Each script runs on CPU in seconds with built-in <code>assert</code> sanity checks. Full overview in <a href="code/README.md"><code>code/README.md</code></a>.</p>
 
 <footer class="aris-footer">
   Generated by <a href="https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep/blob/main/skills/render-html/SKILL.md">ARIS <code>/render-html</code></a> ·
   source path <code>docs/tutorials/attention_tutorial_en.md</code> ·
-  SHA256 <code>ecb6fe924cee</code> ·
-  generated at 2026-05-19 18:47 UTC.
+  SHA256 <code>52fb9664c679</code> ·
+  generated at 2026-05-26 11:27 UTC.
   This is a generated view — edit the source Markdown, then re-render.
 </footer>
 </main>

docs/tutorials/attention_tutorial_en.md

@@ -810,3 +810,14 @@ Actual sanity-check output (PyTorch 2.x, single-machine GPU):
 ```
 
 Code passed independent reviewer static check + PyTorch sanity-check run, diff vs `nn.MultiheadAttention` = 0.
+
+---
+
+## 📜 Runnable Code
+
+Minimal runnable PyTorch implementations of this tutorial's core concepts live in [`docs/tutorials/code/`](code/):
+
+- [`mha.py`](code/mha.py) — standard Multi-Head Self-Attention + causal mask + numerical parity check against `nn.MultiheadAttention`
+- [`axial_attention.py`](code/axial_attention.py) — H/W factorized axial attention + complexity comparison table + receptive-field isolation test
+
+Each script runs on CPU in seconds with built-in `assert` sanity checks. Full overview in [`code/README.md`](code/README.md).

docs/tutorials/flow_matching_tutorial.html

@@ -7,8 +7,8 @@
 
 <meta name="generator" content="ARIS render-html (academic, v1)">
 <meta name="aris:source-path" content="docs/tutorials/flow_matching_tutorial.md">
-<meta name="aris:source-sha256" content="ab4358901d204fec676df6ce3b8559be4e69fec3833f764bcfc937b113eea281">
-<meta name="aris:generated-at" content="2026-05-19 03:40 UTC">
+<meta name="aris:source-sha256" content="435d9a5281b792c9796dffb814c5dd474e15a4173d514405fe3589e22e4b3982">
+<meta name="aris:generated-at" content="2026-05-26 11:27 UTC">
 
 <!-- MathJax 3 -->
 <script>
@@ -367,6 +367,8 @@ <h3>Contents</h3>
 </li>
 <li><a href="#8-完整可运行示例2d-toy">§8 完整可运行示例（2D toy）</a>
 </li>
+<li><a href="#runnable-code">📜 Runnable Code</a>
+</li>
 </ol>
 </nav>
 
@@ -375,11 +377,11 @@ <h3>Contents</h3>
   <div class="eyebrow">Interview Prep · Flow Matching / Diffusion Generative Modeling</div>
   <h1>Flow Matching Quick Reference</h1>
   <p class="subtitle">Conditional Flow Matching · Rectified Flow / VP / VE · 训练 + 采样 + SD3 / FLUX 实战</p>
-  <p class="byline">By <strong>Ruofeng Yang (杨若峰), Shanghai Jiao Tong University</strong></p>
+  
   <div class="meta">
     <span><strong>Source:</strong> <code>docs/tutorials/flow_matching_tutorial.md</code></span>
-    <span><strong>SHA256:</strong> <code>ab4358901d20</code></span>
-    <span><strong>Rendered:</strong> 2026-05-19 03:40 UTC</span>
+    <span><strong>SHA256:</strong> <code>435d9a5281b7</code></span>
+    <span><strong>Rendered:</strong> 2026-05-26 11:27 UTC</span>
 
   </div>
 </header>
@@ -740,12 +742,17 @@ <h2 id="8-完整可运行示例2d-toy">§8 完整可运行示例（2D toy）</h2
 <div class="callout callout-warn"><div class="callout-title">Production 化要补的（教学版未含）</div><p>上线前需补的工程项如下。</p></div>
 <ul><li><strong>EMA scheduler</strong>：训练后期 decay 更接近 1（如 0.9999 → 0.99995）</li><li><strong>Gradient checkpointing</strong>：U-Net / DiT 显存优化</li><li><strong>Mixed precision</strong>：fp16 / bf16 + GradScaler</li><li><strong>Latent space</strong>：高分辨率图像在 VAE latent 里跑 FM（LDM / SD3 / FLUX）</li><li><strong>Conditioning</strong>：text encoder (T5 / CLIP) + cross-attention 或 token concat</li><li><strong>Distributed</strong>：DDP / FSDP for multi-GPU</li><li><strong>Loss weighting</strong>：SD3 用 logit-normal $t$ 已隐式 reweighting；EDM 显式 SNR weighting</li></ul>
 <p><strong>Flow Matching Quick Reference</strong> · 主要参考：Lipman et al. 2023 (Flow Matching), Liu et al. 2022 (Rectified Flow), Esser et al. 2024 (SD3 / MM-DiT)</p>
+<hr />
+<h2 id="runnable-code">📜 Runnable Code</h2>
+<p>本 tutorial 的核心概念在 <a href="code/"><code>docs/tutorials/code/</code></a> 里有最小可跑的 PyTorch 实现：</p>
+<ul><li><a href="code/flow_matching.py"><code>flow_matching.py</code></a> — Rectified Flow on 2D toy moons：训练 <code>(x_1 - x_0)</code> velocity + Euler ODE 采样 + 轨迹可视化</li></ul>
+<p>CPU 30 秒左右跑完（matplotlib 可选），自带 sanity 检查。完整说明见 <a href="code/README.md"><code>code/README.md</code></a>。</p>
 
 <footer class="aris-footer">
   Generated by <a href="https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep/blob/main/skills/render-html/SKILL.md">ARIS <code>/render-html</code></a> ·
   source path <code>docs/tutorials/flow_matching_tutorial.md</code> ·
-  SHA256 <code>ab4358901d20</code> ·
-  generated at 2026-05-19 03:40 UTC.
+  SHA256 <code>435d9a5281b7</code> ·
+  generated at 2026-05-26 11:27 UTC.
   This is a generated view — edit the source Markdown, then re-render.
 </footer>
 </main>

docs/tutorials/flow_matching_tutorial.md

@@ -535,3 +535,13 @@ if __name__ == "__main__":
 - **Loss weighting**：SD3 用 logit-normal $t$ 已隐式 reweighting；EDM 显式 SNR weighting
 
 **Flow Matching Quick Reference** · 主要参考：Lipman et al. 2023 (Flow Matching), Liu et al. 2022 (Rectified Flow), Esser et al. 2024 (SD3 / MM-DiT)
+
+---
+
+## 📜 Runnable Code
+
+本 tutorial 的核心概念在 [`docs/tutorials/code/`](code/) 里有最小可跑的 PyTorch 实现：
+
+- [`flow_matching.py`](code/flow_matching.py) — Rectified Flow on 2D toy moons：训练 `(x_1 - x_0)` velocity + Euler ODE 采样 + 轨迹可视化
+
+CPU 30 秒左右跑完（matplotlib 可选），自带 sanity 检查。完整说明见 [`code/README.md`](code/README.md)。

docs/tutorials/flow_matching_tutorial_en.html

@@ -1,14 +1,14 @@
 <!DOCTYPE html>
-<html lang="zh-CN">
+<html lang="en">
 <head>
 <meta charset="UTF-8">
 <meta name="viewport" content="width=device-width, initial-scale=1.0">
 <title>Flow Matching Tutorial En</title>
 
 <meta name="generator" content="ARIS render-html (academic, v1)">
 <meta name="aris:source-path" content="docs/tutorials/flow_matching_tutorial_en.md">
-<meta name="aris:source-sha256" content="90243d531e3f172dc24c2345e6ada197471f80f92dd45892bc0d002afa86c3f0">
-<meta name="aris:generated-at" content="2026-05-19 18:47 UTC">
+<meta name="aris:source-sha256" content="30fb19ce90837b396223edc71b7d1276adb53fa6ab8b488e27d89a9507b577e7">
+<meta name="aris:generated-at" content="2026-05-26 11:27 UTC">
 
 <!-- MathJax 3 -->
 <script>
@@ -367,6 +367,8 @@ <h3>Contents</h3>
 </li>
 <li><a href="#8-complete-runnable-example-2d-toy">§8 Complete runnable example (2D toy)</a>
 </li>
+<li><a href="#runnable-code">📜 Runnable Code</a>
+</li>
 </ol>
 </nav>
 
@@ -378,8 +380,8 @@ <h1>Flow Matching Tutorial En</h1>
 
   <div class="meta">
     <span><strong>Source:</strong> <code>docs/tutorials/flow_matching_tutorial_en.md</code></span>
-    <span><strong>SHA256:</strong> <code>90243d531e3f</code></span>
-    <span><strong>Rendered:</strong> 2026-05-19 18:47 UTC</span>
+    <span><strong>SHA256:</strong> <code>30fb19ce9083</code></span>
+    <span><strong>Rendered:</strong> 2026-05-26 11:27 UTC</span>
 
   </div>
 </header>
@@ -740,12 +742,17 @@ <h2 id="8-complete-runnable-example-2d-toy">§8 Complete runnable example (2D to
 <div class="callout callout-warn"><div class="callout-title">Production additions (not in this pedagogical version)</div><p>engineering items to add before deployment.</p></div>
 <ul><li><strong>EMA scheduler</strong>: decay closer to 1 in late training (e.g. 0.9999 → 0.99995)</li><li><strong>Gradient checkpointing</strong>: U-Net / DiT memory optimization</li><li><strong>Mixed precision</strong>: fp16 / bf16 + GradScaler</li><li><strong>Latent space</strong>: high-resolution images run FM in VAE latent space (LDM / SD3 / FLUX)</li><li><strong>Conditioning</strong>: text encoder (T5 / CLIP) + cross-attention or token concat</li><li><strong>Distributed</strong>: DDP / FSDP for multi-GPU</li><li><strong>Loss weighting</strong>: SD3 implicitly reweights via logit-normal $t$; EDM uses explicit SNR weighting</li></ul>
 <p><strong>Flow Matching Quick Reference</strong> · Main references: Lipman et al. 2023 (Flow Matching), Liu et al. 2022 (Rectified Flow), Esser et al. 2024 (SD3 / MM-DiT)</p>
+<hr />
+<h2 id="runnable-code">📜 Runnable Code</h2>
+<p>Minimal runnable PyTorch implementations of this tutorial's core concepts live in <a href="code/"><code>docs/tutorials/code/</code></a>:</p>
+<ul><li><a href="code/flow_matching.py"><code>flow_matching.py</code></a> — Rectified Flow on 2D toy moons: train to regress <code>(x_1 - x_0)</code> velocity + Euler ODE sampling + trajectory viz</li></ul>
+<p>Runs on CPU in ~30 seconds (matplotlib optional) with built-in sanity checks. Full overview in <a href="code/README.md"><code>code/README.md</code></a>.</p>
 
 <footer class="aris-footer">
   Generated by <a href="https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep/blob/main/skills/render-html/SKILL.md">ARIS <code>/render-html</code></a> ·
   source path <code>docs/tutorials/flow_matching_tutorial_en.md</code> ·
-  SHA256 <code>90243d531e3f</code> ·
-  generated at 2026-05-19 18:47 UTC.
+  SHA256 <code>30fb19ce9083</code> ·
+  generated at 2026-05-26 11:27 UTC.
   This is a generated view — edit the source Markdown, then re-render.
 </footer>
 </main>

docs/tutorials/flow_matching_tutorial_en.md

@@ -535,3 +535,13 @@ if __name__ == "__main__":
 - **Loss weighting**: SD3 implicitly reweights via logit-normal $t$; EDM uses explicit SNR weighting
 
 **Flow Matching Quick Reference** · Main references: Lipman et al. 2023 (Flow Matching), Liu et al. 2022 (Rectified Flow), Esser et al. 2024 (SD3 / MM-DiT)
+
+---
+
+## 📜 Runnable Code
+
+Minimal runnable PyTorch implementations of this tutorial's core concepts live in [`docs/tutorials/code/`](code/):
+
+- [`flow_matching.py`](code/flow_matching.py) — Rectified Flow on 2D toy moons: train to regress `(x_1 - x_0)` velocity + Euler ODE sampling + trajectory viz
+
+Runs on CPU in ~30 seconds (matplotlib optional) with built-in sanity checks. Full overview in [`code/README.md`](code/README.md).