thibaud frere commited on
Commit
d35f5b3
·
1 Parent(s): 9d8447d
app/src/components/TableOfContents.astro CHANGED
@@ -251,6 +251,7 @@ const { tableOfContentAutoCollapse = false } = Astro.props as Props;
251
  .table-of-contents {
252
  position: sticky;
253
  top: 32px;
 
254
  }
255
 
256
  .table-of-contents nav {
 
251
  .table-of-contents {
252
  position: sticky;
253
  top: 32px;
254
+ margin-top: 12px;
255
  }
256
 
257
  .table-of-contents nav {
app/src/content/chapters/components.mdx CHANGED
@@ -213,6 +213,8 @@ They exist in the `app/src/content/embeds` folder.
213
 
214
  For researchers who want to stay in **Python** while targeting **D3**, the [d3blocks](https://github.com/d3blocks/d3blocks) library lets you create interactive D3 charts with only a few lines of code. In **2025**, **D3** often provides more flexibility and a more web‑native rendering than **Plotly** for custom visualizations.
215
 
 
 
216
  <HtmlEmbed src="d3-line-example.html" title="This is a chart title" desc="Some chart description" />
217
 
218
  | Prop | Required | Description
@@ -249,7 +251,8 @@ Inside this file, you will find every directives you need to code your own **Htm
249
 
250
  #### Real world examples
251
 
252
- Here are some examples that were vibe coded to inspire you.
 
253
 
254
  ---
255
  <HtmlEmbed src="d3-line.html" title="Training curves by metric" desc="Interactive time series across runs. Choose a metric; hover for values." />
@@ -257,7 +260,7 @@ Here are some examples that were vibe coded to inspire you.
257
  <HtmlEmbed src="d3-bar.html" title="D3 Memory usage with recomputation" desc={`Memory usage with recomputation — <a href="https://huggingface.co/spaces/nanotron/ultrascale-playbook?section=activation_recomputation" target="_blank">from the ultrascale playbook</a>`}/>
258
  ---
259
  <Wide>
260
- <HtmlEmbed src="d3-neural.html" title="D3 Interactive neural network (MNIST-like)" desc="Visualize activations and class probabilities (0–9)." align="center" />
261
  </Wide>
262
  ---
263
  <Wide>
 
213
 
214
  For researchers who want to stay in **Python** while targeting **D3**, the [d3blocks](https://github.com/d3blocks/d3blocks) library lets you create interactive D3 charts with only a few lines of code. In **2025**, **D3** often provides more flexibility and a more web‑native rendering than **Plotly** for custom visualizations.
215
 
216
+ <Note>You can refer to a chart by creating a link to it <a href="#real-world-examples">here</a>. All you need to do is put the id="real-world-examples" in the HtmlEmbed. <a href="#neural-network-mnist-like">Fig 1</a></Note>
217
+
218
  <HtmlEmbed src="d3-line-example.html" title="This is a chart title" desc="Some chart description" />
219
 
220
  | Prop | Required | Description
 
251
 
252
  #### Real world examples
253
 
254
+ Here are some examples that were vibe coded to inspire you.
255
+
256
 
257
  ---
258
  <HtmlEmbed src="d3-line.html" title="Training curves by metric" desc="Interactive time series across runs. Choose a metric; hover for values." />
 
260
  <HtmlEmbed src="d3-bar.html" title="D3 Memory usage with recomputation" desc={`Memory usage with recomputation — <a href="https://huggingface.co/spaces/nanotron/ultrascale-playbook?section=activation_recomputation" target="_blank">from the ultrascale playbook</a>`}/>
261
  ---
262
  <Wide>
263
+ <HtmlEmbed src="d3-neural.html" id="neural-network-mnist-like" title="D3 Interactive neural network (MNIST-like)" desc="Visualize activations and class probabilities (0–9)." align="center" />
264
  </Wide>
265
  ---
266
  <Wide>