Upload DESIGN.md with huggingface_hub
Browse files
DESIGN.md
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# RESMP.DEV visual system
|
| 2 |
+
|
| 3 |
+
The system combines an editorial serif hierarchy, instrument-panel grids, dark warning boards, and the RESMP palette:
|
| 4 |
+
|
| 5 |
+
| Token | Value | Use |
|
| 6 |
+
| --- | --- | --- |
|
| 7 |
+
| Ink | `#101012` | NERV-dark field and type |
|
| 8 |
+
| Paper | `#f1f0e9` | Warm technical paper |
|
| 9 |
+
| Unit violet | `#6f36d9` | Primary model/candidate |
|
| 10 |
+
| Signal acid | `#b6f238` | Active state and controls |
|
| 11 |
+
| Warning red | `#e2382f` | Hard boundary and warning rail |
|
| 12 |
+
|
| 13 |
+
## Reusable benchmark chart
|
| 14 |
+
|
| 15 |
+
Copy `benchmark-chart.js` and `benchmark-component.css`, then provide JSON inside the element:
|
| 16 |
+
|
| 17 |
+
```html
|
| 18 |
+
<link rel="stylesheet" href="benchmark-component.css">
|
| 19 |
+
<script type="module" src="benchmark-chart.js"></script>
|
| 20 |
+
|
| 21 |
+
<resmp-benchmark-chart>
|
| 22 |
+
<script type="application/json">
|
| 23 |
+
{
|
| 24 |
+
"title": "MODEL EVALUATIONS",
|
| 25 |
+
"higherIsBetter": true,
|
| 26 |
+
"series": [{"id":"candidate","label":"CANDIDATE","color":"#6f36d9"}],
|
| 27 |
+
"groups": [{"name":"STEM","panels":[
|
| 28 |
+
{"name":"MATH","values":{"candidate":62.4}}
|
| 29 |
+
]}]
|
| 30 |
+
}
|
| 31 |
+
</script>
|
| 32 |
+
</resmp-benchmark-chart>
|
| 33 |
+
```
|
| 34 |
+
|
| 35 |
+
Or set `document.querySelector('resmp-benchmark-chart').config` to the same object. The component supports positive and negative values, direct value labels, fixed series ordering, responsive one-column panels, print output, and SVG export.
|
| 36 |
+
|
| 37 |
+
`benchmark.html` is the complete 4 × 3 small-multiples reference implementation. Its values are explicitly labeled demonstration data.
|