cy0307 commited on
Commit
da4fc02
·
verified ·
1 Parent(s): 6d5c3f2

Publish Ropedia Xperience-10M derived artifacts

Browse files
Files changed (34) hide show
  1. PROJECT_README.md +6 -0
  2. README.md +5 -0
  3. assets/charts/cross_modal_retrieval.svg +28 -27
  4. assets/charts/episode_task_scores.svg +52 -51
  5. assets/charts/episode_task_scores_minimal_vs_neural.svg +88 -87
  6. assets/charts/episode_task_scores_neural_mlp.svg +52 -51
  7. assets/charts/feature_blocks.svg +67 -66
  8. assets/charts/model_macro_f1.svg +28 -27
  9. assets/charts/research_direction_coverage.svg +40 -39
  10. assets/charts/research_direction_extension_tasks.svg +64 -64
  11. assets/pipeline_diagram.png +2 -2
  12. assets/pipeline_diagram.svg +64 -62
  13. assets/task_architectures.png +2 -2
  14. assets/task_architectures.svg +221 -219
  15. assets/task_suite_infographic.png +2 -2
  16. docs/assets/charts/cross_modal_retrieval.svg +28 -27
  17. docs/assets/charts/episode_task_scores.svg +52 -51
  18. docs/assets/charts/episode_task_scores_minimal_vs_neural.svg +88 -87
  19. docs/assets/charts/episode_task_scores_neural_mlp.svg +52 -51
  20. docs/assets/charts/feature_blocks.svg +67 -66
  21. docs/assets/charts/model_macro_f1.svg +28 -27
  22. docs/assets/charts/research_direction_coverage.svg +40 -39
  23. docs/assets/charts/research_direction_extension_tasks.svg +64 -64
  24. docs/assets/pipeline_diagram.png +2 -2
  25. docs/assets/pipeline_diagram.svg +64 -62
  26. docs/assets/task_architectures.png +2 -2
  27. docs/assets/task_architectures.svg +221 -219
  28. docs/assets/task_suite_infographic.png +2 -2
  29. docs/index.html +98 -92
  30. scripts/generate_visualizations.py +64 -59
  31. scripts/render_overview_figures.py +74 -74
  32. scripts/render_task_suite_infographic.py +94 -96
  33. scripts/research_direction_extension_tasks.py +19 -19
  34. scripts/research_direction_taxonomy.py +10 -9
PROJECT_README.md CHANGED
@@ -8,6 +8,12 @@
8
  An audit-first embodied-AI learning repo built around one public
9
  Xperience-10M sample episode released by Ropedia.
10
 
 
 
 
 
 
 
11
  The project does one narrow thing carefully: it turns a raw multimodal episode
12
  into:
13
 
 
8
  An audit-first embodied-AI learning repo built around one public
9
  Xperience-10M sample episode released by Ropedia.
10
 
11
+ The public dashboard and generated figures deliberately follow the visual
12
+ language of [ropedia.com](https://ropedia.com/): near-black 4D-world canvas,
13
+ lime-green identity accents, thin green-tinted cards, point-cloud texture, and
14
+ the Inter Tight / Space Grotesk typography pairing. The layout is original to
15
+ this project, but the style stays aligned with Ropedia's own product site.
16
+
17
  The project does one narrow thing carefully: it turns a raw multimodal episode
18
  into:
19
 
README.md CHANGED
@@ -26,6 +26,11 @@ size_categories:
26
 
27
  This dataset repo contains the derived evidence layer for the public Xperience-10M sample episode released by Ropedia: metrics, predictions, manifests, charts, diagrams, notes, reproduction scripts, and the small neural MLP task-head artifacts.
28
 
 
 
 
 
 
29
  It does **not** contain raw Xperience-10M videos or raw `annotation.hdf5`. Download raw data only from the official Ropedia / Hugging Face sources and follow their terms.
30
 
31
  Current scale-up status: the full `ropedia-ai/xperience-10m` Hugging Face
 
26
 
27
  This dataset repo contains the derived evidence layer for the public Xperience-10M sample episode released by Ropedia: metrics, predictions, manifests, charts, diagrams, notes, reproduction scripts, and the small neural MLP task-head artifacts.
28
 
29
+ The dashboard assets follow a Ropedia-inspired visual system: dark 4D-world
30
+ canvas, lime-green accents, point-cloud texture, thin green cards, and
31
+ research-grade typography, while all labels and metrics are script-generated
32
+ from committed result files.
33
+
34
  It does **not** contain raw Xperience-10M videos or raw `annotation.hdf5`. Download raw data only from the official Ropedia / Hugging Face sources and follow their terms.
35
 
36
  Current scale-up status: the full `ropedia-ai/xperience-10m` Hugging Face
assets/charts/cross_modal_retrieval.svg CHANGED
assets/charts/episode_task_scores.svg CHANGED
assets/charts/episode_task_scores_minimal_vs_neural.svg CHANGED
assets/charts/episode_task_scores_neural_mlp.svg CHANGED
assets/charts/feature_blocks.svg CHANGED
assets/charts/model_macro_f1.svg CHANGED
assets/charts/research_direction_coverage.svg CHANGED
assets/charts/research_direction_extension_tasks.svg CHANGED
assets/pipeline_diagram.png CHANGED

Git LFS Details

  • SHA256: a4d5c5393952f8e399d6af3cb92a179a1de52094fa675ec62a48c95734d4c4e5
  • Pointer size: 131 Bytes
  • Size of remote file: 868 kB

Git LFS Details

  • SHA256: 09de5f158927c51d474d991ee90f1e2c5d77616e79ff6fb0cca7eacfa78144a8
  • Pointer size: 131 Bytes
  • Size of remote file: 703 kB
assets/pipeline_diagram.svg CHANGED
assets/task_architectures.png CHANGED

Git LFS Details

  • SHA256: 1d8882e017283413e98a46b92da131d0b7a5f0dd4f1d296ee3e4a698bed02504
  • Pointer size: 131 Bytes
  • Size of remote file: 816 kB

Git LFS Details

  • SHA256: b0fd65c9f4e16661f6cba64745c2fa9a621c5077327415623af2fb081fc758d9
  • Pointer size: 131 Bytes
  • Size of remote file: 737 kB
assets/task_architectures.svg CHANGED
assets/task_suite_infographic.png CHANGED

Git LFS Details

  • SHA256: 6dd90f87b7e6c4aaecf6568fff39cdf2640e0e03c11d6c9664e62a9a9e1daedd
  • Pointer size: 132 Bytes
  • Size of remote file: 1.15 MB

Git LFS Details

  • SHA256: 96a76242c07297df0673a81be2fb7aefb16b86cace8d958e5b94fb6d52c12d2a
  • Pointer size: 132 Bytes
  • Size of remote file: 1.05 MB
docs/assets/charts/cross_modal_retrieval.svg CHANGED
docs/assets/charts/episode_task_scores.svg CHANGED
docs/assets/charts/episode_task_scores_minimal_vs_neural.svg CHANGED
docs/assets/charts/episode_task_scores_neural_mlp.svg CHANGED
docs/assets/charts/feature_blocks.svg CHANGED
docs/assets/charts/model_macro_f1.svg CHANGED
docs/assets/charts/research_direction_coverage.svg CHANGED
docs/assets/charts/research_direction_extension_tasks.svg CHANGED
docs/assets/pipeline_diagram.png CHANGED

Git LFS Details

  • SHA256: a4d5c5393952f8e399d6af3cb92a179a1de52094fa675ec62a48c95734d4c4e5
  • Pointer size: 131 Bytes
  • Size of remote file: 868 kB

Git LFS Details

  • SHA256: 09de5f158927c51d474d991ee90f1e2c5d77616e79ff6fb0cca7eacfa78144a8
  • Pointer size: 131 Bytes
  • Size of remote file: 703 kB
docs/assets/pipeline_diagram.svg CHANGED
docs/assets/task_architectures.png CHANGED

Git LFS Details

  • SHA256: 1d8882e017283413e98a46b92da131d0b7a5f0dd4f1d296ee3e4a698bed02504
  • Pointer size: 131 Bytes
  • Size of remote file: 816 kB

Git LFS Details

  • SHA256: b0fd65c9f4e16661f6cba64745c2fa9a621c5077327415623af2fb081fc758d9
  • Pointer size: 131 Bytes
  • Size of remote file: 737 kB
docs/assets/task_architectures.svg CHANGED
docs/assets/task_suite_infographic.png CHANGED

Git LFS Details

  • SHA256: 6dd90f87b7e6c4aaecf6568fff39cdf2640e0e03c11d6c9664e62a9a9e1daedd
  • Pointer size: 132 Bytes
  • Size of remote file: 1.15 MB

Git LFS Details

  • SHA256: 96a76242c07297df0673a81be2fb7aefb16b86cace8d958e5b94fb6d52c12d2a
  • Pointer size: 132 Bytes
  • Size of remote file: 1.05 MB
docs/index.html CHANGED
@@ -8,27 +8,30 @@
8
  <meta property="og:title" content="Ropedia Xperience-10M Task Suite">
9
  <meta property="og:description" content="One public embodied-AI episode, converted into reproducible multimodal tasks, minimal baselines, neural MLP baselines, metrics, and diagrams.">
10
  <meta property="og:image" content="assets/task_suite_infographic.png?v=xperience10m-nn">
 
 
 
11
  <style>
12
  :root {
13
- color-scheme: light;
14
- --ink: #1f2421;
15
- --muted: #6f716c;
16
- --line: #e4ded4;
17
- --soft-line: #eee9e1;
18
- --page: #fbfaf7;
19
- --panel: #f5f1e9;
20
- --surface: #fffefd;
21
- --blue: #1f6c9f;
22
- --cyan: #2e7775;
23
- --green: #346538;
24
- --amber: #956400;
25
- --red: #9f2f2d;
26
- --shadow: 0 24px 60px rgba(68, 55, 38, 0.08);
27
  --radius: 8px;
28
  --max: 1220px;
29
- --font-ui: "Avenir Next", "SF Pro Display", ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif;
30
- --font-copy: "Avenir Next", "SF Pro Text", ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif;
31
- --font-mono: "SF Mono", "JetBrains Mono", ui-monospace, SFMono-Regular, Menlo, monospace;
32
  }
33
 
34
  * { box-sizing: border-box; }
@@ -38,10 +41,11 @@
38
  font-family: var(--font-copy);
39
  color: var(--ink);
40
  background:
41
- linear-gradient(90deg, rgba(68,55,38,0.028) 1px, transparent 1px),
42
- linear-gradient(0deg, rgba(68,55,38,0.022) 1px, transparent 1px),
 
43
  var(--page);
44
- background-size: 56px 56px, 56px 56px, auto;
45
  line-height: 1.5;
46
  text-rendering: optimizeLegibility;
47
  }
@@ -54,7 +58,7 @@
54
  top: 12px;
55
  transform: translateY(-160%);
56
  background: var(--ink);
57
- color: white;
58
  padding: 10px 12px;
59
  border-radius: 6px;
60
  z-index: 40;
@@ -65,7 +69,7 @@
65
  position: sticky;
66
  top: 0;
67
  z-index: 20;
68
- background: rgba(251,250,247,0.88);
69
  backdrop-filter: blur(18px);
70
  border-bottom: 1px solid var(--soft-line);
71
  }
@@ -88,11 +92,11 @@
88
  .mark {
89
  width: 34px;
90
  height: 34px;
91
- border: 1px solid #d7ccbd;
92
  display: grid;
93
  place-items: center;
94
- color: var(--ink);
95
- background: #f1eadf;
96
  font-family: var(--font-mono);
97
  font-size: 13px;
98
  font-weight: 760;
@@ -102,7 +106,7 @@
102
  align-items: center;
103
  gap: 20px;
104
  font-size: 14px;
105
- color: #4e514c;
106
  }
107
  .nav-links a {
108
  text-decoration: none;
@@ -110,13 +114,13 @@
110
  }
111
  .nav-links a:hover { color: var(--ink); transform: translateY(-1px); }
112
  a:focus-visible, button:focus-visible {
113
- outline: 2px solid rgba(31,108,159,0.48);
114
  outline-offset: 3px;
115
  }
116
  .nav-action {
117
  border: 1px solid var(--ink);
118
- background: var(--ink);
119
- color: white;
120
  height: 36px;
121
  padding: 0 14px;
122
  display: inline-flex;
@@ -130,10 +134,10 @@
130
  .hero {
131
  border-bottom: 1px solid var(--soft-line);
132
  background:
133
- linear-gradient(115deg, rgba(255,255,255,0.82), rgba(245,241,233,0.72)),
134
- linear-gradient(90deg, rgba(68,55,38,0.05) 1px, transparent 1px),
135
- linear-gradient(0deg, rgba(68,55,38,0.035) 1px, transparent 1px);
136
- background-size: auto, 54px 54px, 54px 54px;
137
  }
138
  .hero-inner {
139
  min-height: 680px;
@@ -148,7 +152,7 @@
148
  align-items: center;
149
  gap: 9px;
150
  margin-bottom: 22px;
151
- color: #4f524d;
152
  font-family: var(--font-mono);
153
  font-size: 12px;
154
  text-transform: uppercase;
@@ -158,7 +162,7 @@
158
  content: "";
159
  width: 34px;
160
  height: 1px;
161
- background: var(--ink);
162
  }
163
  h1 {
164
  margin: 0;
@@ -168,11 +172,12 @@
168
  letter-spacing: 0;
169
  max-width: 860px;
170
  text-wrap: balance;
 
171
  }
172
  .hero-copy {
173
  margin: 26px 0 0;
174
  max-width: 690px;
175
- color: #4e514c;
176
  font-size: 19px;
177
  line-height: 1.65;
178
  text-wrap: pretty;
@@ -193,15 +198,16 @@
193
  gap: 10px;
194
  text-decoration: none;
195
  font-weight: 720;
196
- background: var(--surface);
197
  transition: transform 240ms cubic-bezier(0.16, 1, 0.3, 1), border-color 240ms cubic-bezier(0.16, 1, 0.3, 1), background 240ms cubic-bezier(0.16, 1, 0.3, 1);
198
  }
199
- .button:hover { transform: translateY(-2px); border-color: #cbbfad; }
200
  .button:active { transform: translateY(0) scale(0.98); }
201
  .button.primary {
202
  background: var(--ink);
203
- color: white;
204
- border-color: var(--ink);
 
205
  }
206
  .hero-stats {
207
  display: grid;
@@ -212,14 +218,14 @@
212
  }
213
  .stat {
214
  border: 1px solid var(--line);
215
- background: rgba(255,254,253,0.86);
216
  padding: 14px 14px 13px;
217
  border-radius: var(--radius);
218
  }
219
  .stat strong { display: block; font-family: var(--font-mono); font-size: 21px; line-height: 1; font-variant-numeric: tabular-nums; }
220
  .stat span { display: block; margin-top: 7px; font-size: 12px; color: var(--muted); }
221
  .hero-panel {
222
- background: rgba(255,254,253,0.92);
223
  border: 1px solid var(--line);
224
  border-radius: var(--radius);
225
  box-shadow: var(--shadow);
@@ -245,8 +251,8 @@
245
  }
246
  .signal:last-child { border-bottom: 0; }
247
  .signal code {
248
- color: #2f3437;
249
- background: #f4f0e8;
250
  border: 1px solid var(--soft-line);
251
  padding: 5px 7px;
252
  border-radius: 5px;
@@ -256,7 +262,7 @@
256
  .track {
257
  position: relative;
258
  height: 12px;
259
- background: #eee8df;
260
  overflow: hidden;
261
  border-radius: 999px;
262
  }
@@ -269,7 +275,7 @@
269
  }
270
  .signal strong { text-align: right; font-family: var(--font-mono); font-size: 13px; font-variant-numeric: tabular-nums; }
271
 
272
- section { padding: 88px 0 96px; border-bottom: 1px solid var(--soft-line); }
273
  #suite { padding: 62px 0 76px; }
274
  .section-head {
275
  display: flex;
@@ -303,7 +309,7 @@
303
  border: 1px solid var(--line);
304
  border-radius: var(--radius);
305
  background: var(--surface);
306
- box-shadow: 0 14px 34px rgba(68, 55, 38, 0.045);
307
  }
308
  .task-suite-image {
309
  display: block;
@@ -328,11 +334,11 @@
328
  border: 1px solid var(--line);
329
  border-radius: var(--radius);
330
  padding: 24px;
331
- background: #f8f2e8;
332
- color: #3f3425;
333
  }
334
  .callout h3, .artifact h3 { margin: 0 0 8px; font-size: 17px; }
335
- .callout p, .artifact p { margin: 0; color: #665d51; line-height: 1.6; }
336
 
337
  .models {
338
  display: grid;
@@ -347,7 +353,7 @@
347
  background: var(--surface);
348
  transition: transform 240ms cubic-bezier(0.16, 1, 0.3, 1), box-shadow 240ms cubic-bezier(0.16, 1, 0.3, 1);
349
  }
350
- .model:hover { transform: translateY(-3px); box-shadow: 0 14px 28px rgba(68, 55, 38, 0.06); }
351
  .model h3 { margin: 0; font-size: 15px; }
352
  .model .score { display: block; margin-top: 18px; font-family: var(--font-mono); font-size: 33px; font-weight: 760; line-height: 1; font-variant-numeric: tabular-nums; }
353
  .model .meta { display: block; margin-top: 8px; color: var(--muted); font-size: 13px; }
@@ -365,12 +371,12 @@
365
  height: 36px;
366
  padding: 0 14px;
367
  font-weight: 650;
368
- color: #354052;
369
  cursor: pointer;
370
  transition: transform 220ms cubic-bezier(0.16, 1, 0.3, 1), background 220ms cubic-bezier(0.16, 1, 0.3, 1);
371
  }
372
  .filter:hover { transform: translateY(-1px); }
373
- .filter.active { color: white; background: var(--ink); border-color: var(--ink); }
374
  .task-grid {
375
  display: grid;
376
  grid-template-columns: repeat(3, minmax(0, 1fr));
@@ -386,7 +392,7 @@
386
  min-height: 184px;
387
  transition: transform 240ms cubic-bezier(0.16, 1, 0.3, 1), border-color 240ms cubic-bezier(0.16, 1, 0.3, 1);
388
  }
389
- .task-card:hover { transform: translateY(-3px); border-color: #cbbfad; }
390
  .task-card.hide { display: none; }
391
  .task-top { display: flex; justify-content: space-between; gap: 14px; align-items: start; }
392
  .task-name { font-family: var(--font-mono); font-size: 13px; font-weight: 760; }
@@ -394,18 +400,18 @@
394
  font-size: 11px;
395
  border-radius: 999px;
396
  padding: 4px 8px;
397
- color: #1f2937;
398
- background: #eef2f7;
399
  white-space: nowrap;
400
  }
401
- .tag.supervised { background: #e1f3fe; color: #1f6c9f; }
402
- .tag.forecast { background: #edf3ec; color: #346538; }
403
- .tag.retrieval { background: #e6f4f1; color: #2e7775; }
404
- .tag.diagnostic { background: #fbf3db; color: #956400; }
405
  .task-card p { margin: 0; color: var(--muted); font-size: 13px; }
406
  .metric-row { display: flex; justify-content: space-between; gap: 12px; font-size: 13px; }
407
  .metric-row strong { font-family: var(--font-mono); font-size: 18px; font-variant-numeric: tabular-nums; }
408
- .mini-bar { height: 7px; background: #eee8df; border-radius: 999px; overflow: hidden; }
409
  .mini-bar span { display: block; height: 100%; width: var(--w); background: var(--c); }
410
 
411
  .artifact-grid {
@@ -441,8 +447,8 @@
441
  width: fit-content;
442
  border-radius: 999px;
443
  padding: 4px 9px;
444
- background: #edf2f7;
445
- color: #334155;
446
  font-size: 11px;
447
  font-weight: 740;
448
  }
@@ -451,7 +457,7 @@
451
  grid-template-columns: repeat(3, minmax(0, 1fr));
452
  gap: 8px;
453
  font-size: 12px;
454
- color: #4e514c;
455
  }
456
  .direction-counts strong {
457
  display: block;
@@ -488,7 +494,7 @@
488
  grid-template-columns: repeat(2, minmax(0, 1fr));
489
  gap: 8px;
490
  font-size: 12px;
491
- color: #4e514c;
492
  }
493
  .extension-metrics strong {
494
  display: block;
@@ -521,8 +527,8 @@
521
  font-size: 12px;
522
  }
523
  .walk-flow span {
524
- border: 1px solid #e5dfd5;
525
- background: #faf7f1;
526
  border-radius: 6px;
527
  padding: 9px;
528
  min-height: 58px;
@@ -535,28 +541,28 @@
535
  padding: 18px;
536
  transition: transform 240ms cubic-bezier(0.16, 1, 0.3, 1), border-color 240ms cubic-bezier(0.16, 1, 0.3, 1);
537
  }
538
- .artifact:hover { transform: translateY(-3px); border-color: #cbbfad; }
539
  .artifact a { display: inline-block; margin-top: 14px; font-weight: 740; text-decoration: none; color: var(--blue); }
540
  .artifact a:hover { text-decoration: underline; text-underline-offset: 4px; }
541
 
542
  .code-panel {
543
- background: #1f2421;
544
- color: #f7f2e8;
545
  border-radius: var(--radius);
546
  padding: 18px;
547
  overflow: auto;
548
  font-family: var(--font-mono);
549
  font-size: 13px;
550
  line-height: 1.65;
551
- border: 1px solid #3c403b;
552
  }
553
  .code-panel button {
554
  float: right;
555
  margin-left: 16px;
556
  height: 30px;
557
- border: 1px solid #5a6059;
558
- color: white;
559
- background: #363c36;
560
  border-radius: 5px;
561
  cursor: pointer;
562
  font-weight: 700;
@@ -647,12 +653,12 @@
647
  <span>current feature allocation</span>
648
  <span>window vector</span>
649
  </div>
650
- <div class="signal"><code>mocap</code><div class="track"><span style="--w:25.3%;--c:#1f6c9f"></span></div><strong>2,121</strong></div>
651
- <div class="signal"><code>camera+imu</code><div class="track"><span style="--w:1.5%;--c:#2e7775"></span></div><strong>126</strong></div>
652
- <div class="signal"><code>depth</code><div class="track"><span style="--w:11.7%;--c:#346538"></span></div><strong>980</strong></div>
653
- <div class="signal"><code>video</code><div class="track"><span style="--w:49.1%;--c:#956400"></span></div><strong>4,116</strong></div>
654
- <div class="signal"><code>language</code><div class="track"><span style="--w:10.7%;--c:#6f5b21"></span></div><strong>896</strong></div>
655
- <div class="signal"><code>static</code><div class="track"><span style="--w:1.7%;--c:#6f716c"></span></div><strong>139</strong></div>
656
  </div>
657
  </div>
658
  </header>
@@ -870,18 +876,18 @@
870
  <button class="filter" data-filter="diagnostic">Diagnostic</button>
871
  </div>
872
  <div class="task-grid" id="taskGrid">
873
- <article class="task-card" data-kind="supervised"><div class="task-top"><span class="task-name">timeline_action</span><span class="tag supervised">supervised</span></div><p>All featurized modalities to current action label. Chronological split exposes unseen future actions.</p><div class="metric-row"><span>macro-F1</span><strong>0.0500</strong></div><div class="mini-bar"><span style="--w:5%;--c:#1f6c9f"></span></div></article>
874
- <article class="task-card" data-kind="supervised"><div class="task-top"><span class="task-name">timeline_subtask</span><span class="tag supervised">supervised</span></div><p>All featurized modalities to current subtask label. Useful for segmentation diagnostics.</p><div class="metric-row"><span>macro-F1</span><strong>0.0495</strong></div><div class="mini-bar"><span style="--w:5%;--c:#1f6c9f"></span></div></article>
875
- <article class="task-card" data-kind="diagnostic"><div class="task-top"><span class="task-name">transition_detection</span><span class="tag diagnostic">diagnostic</span></div><p>Predict steady vs action boundary. Highlights task-transition localization quality.</p><div class="metric-row"><span>macro-F1</span><strong>0.6552</strong></div><div class="mini-bar"><span style="--w:65.5%;--c:#956400"></span></div></article>
876
- <article class="task-card" data-kind="supervised"><div class="task-top"><span class="task-name">next_action</span><span class="tag supervised">supervised</span></div><p>Current multimodal window to action 20 frames later. Tests short-horizon task flow.</p><div class="metric-row"><span>macro-F1</span><strong>0.0593</strong></div><div class="mini-bar"><span style="--w:5.9%;--c:#1f6c9f"></span></div></article>
877
- <article class="task-card" data-kind="forecast"><div class="task-top"><span class="task-name">hand_trajectory_forecast</span><span class="tag forecast">forecast</span></div><p>Predict future left/right hand 3D joints. Closer to imitation-learning style signals.</p><div class="metric-row"><span>MPJPE</span><strong>0.8223</strong></div><div class="mini-bar"><span style="--w:48%;--c:#346538"></span></div></article>
878
- <article class="task-card" data-kind="supervised"><div class="task-top"><span class="task-name">contact_prediction</span><span class="tag supervised">supervised</span></div><p>Non-contact modalities to binary contact. Degenerate in this sample because one class dominates.</p><div class="metric-row"><span>accuracy</span><strong>1.0000</strong></div><div class="mini-bar"><span style="--w:100%;--c:#346538"></span></div></article>
879
- <article class="task-card" data-kind="supervised"><div class="task-top"><span class="task-name">object_relevance</span><span class="tag supervised">supervised</span></div><p>Predict relevant object set from non-caption feature blocks.</p><div class="metric-row"><span>micro-F1</span><strong>0.1839</strong></div><div class="mini-bar"><span style="--w:18.4%;--c:#1f6c9f"></span></div></article>
880
- <article class="task-card" data-kind="retrieval"><div class="task-top"><span class="task-name">caption_grounding</span><span class="tag retrieval">retrieval</span></div><p>Caption objects/interaction query to matching sensor window.</p><div class="metric-row"><span>MRR</span><strong>0.0172</strong></div><div class="mini-bar"><span style="--w:1.7%;--c:#2e7775"></span></div></article>
881
- <article class="task-card" data-kind="retrieval"><div class="task-top"><span class="task-name">cross_modal_retrieval</span><span class="tag retrieval">retrieval</span></div><p>Motion/IMU/camera query retrieves matching depth/video window. Strongest single-episode signal.</p><div class="metric-row"><span>top-5</span><strong>0.3764</strong></div><div class="mini-bar"><span style="--w:37.6%;--c:#2e7775"></span></div></article>
882
- <article class="task-card" data-kind="forecast"><div class="task-top"><span class="task-name">modality_reconstruction</span><span class="tag forecast">forecast</span></div><p>Motion/IMU/camera to depth/video feature vector.</p><div class="metric-row"><span>R2</span><strong>-0.0160</strong></div><div class="mini-bar"><span style="--w:2%;--c:#9f2f2d"></span></div></article>
883
- <article class="task-card" data-kind="diagnostic"><div class="task-top"><span class="task-name">temporal_order</span><span class="tag diagnostic">diagnostic</span></div><p>Two adjacent windows to correct vs reversed order.</p><div class="metric-row"><span>F1</span><strong>0.5487</strong></div><div class="mini-bar"><span style="--w:54.9%;--c:#956400"></span></div></article>
884
- <article class="task-card" data-kind="diagnostic"><div class="task-top"><span class="task-name">misalignment_detection</span><span class="tag diagnostic">diagnostic</span></div><p>Motion+visual pair to aligned vs shifted by eight windows.</p><div class="metric-row"><span>F1</span><strong>0.4866</strong></div><div class="mini-bar"><span style="--w:48.7%;--c:#956400"></span></div></article>
885
  </div>
886
  </div>
887
  </section>
 
8
  <meta property="og:title" content="Ropedia Xperience-10M Task Suite">
9
  <meta property="og:description" content="One public embodied-AI episode, converted into reproducible multimodal tasks, minimal baselines, neural MLP baselines, metrics, and diagrams.">
10
  <meta property="og:image" content="assets/task_suite_infographic.png?v=xperience10m-nn">
11
+ <link rel="preconnect" href="https://fonts.googleapis.com">
12
+ <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
13
+ <link href="https://fonts.googleapis.com/css2?family=Inter+Tight:wght@400;500;600;700;800&family=Space+Grotesk:wght@400;500;600;700&display=swap" rel="stylesheet">
14
  <style>
15
  :root {
16
+ color-scheme: dark;
17
+ --ink: #f4f8ef;
18
+ --muted: #a5afa2;
19
+ --line: rgba(164, 242, 127, 0.24);
20
+ --soft-line: rgba(164, 242, 127, 0.14);
21
+ --page: #020502;
22
+ --panel: #061006;
23
+ --surface: #071207;
24
+ --blue: #9bdfff;
25
+ --cyan: #7ae5c3;
26
+ --green: #a7f078;
27
+ --amber: #d8f4a5;
28
+ --red: #ff8f7a;
29
+ --shadow: 0 28px 80px rgba(0, 0, 0, 0.42);
30
  --radius: 8px;
31
  --max: 1220px;
32
+ --font-ui: "Inter Tight", "Avenir Next", "SF Pro Display", ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif;
33
+ --font-copy: "Space Grotesk", "Avenir Next", "SF Pro Text", ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif;
34
+ --font-mono: "Space Grotesk", "SF Mono", "JetBrains Mono", ui-monospace, SFMono-Regular, Menlo, monospace;
35
  }
36
 
37
  * { box-sizing: border-box; }
 
41
  font-family: var(--font-copy);
42
  color: var(--ink);
43
  background:
44
+ radial-gradient(circle at 80% 10%, rgba(164, 242, 127, 0.12), transparent 28%),
45
+ radial-gradient(circle at 18% 24%, rgba(164, 242, 127, 0.06), transparent 22%),
46
+ radial-gradient(circle, rgba(164, 242, 127, 0.11) 1px, transparent 1.4px),
47
  var(--page);
48
+ background-size: auto, auto, 22px 22px, auto;
49
  line-height: 1.5;
50
  text-rendering: optimizeLegibility;
51
  }
 
58
  top: 12px;
59
  transform: translateY(-160%);
60
  background: var(--ink);
61
+ color: #020502;
62
  padding: 10px 12px;
63
  border-radius: 6px;
64
  z-index: 40;
 
69
  position: sticky;
70
  top: 0;
71
  z-index: 20;
72
+ background: rgba(2, 5, 2, 0.86);
73
  backdrop-filter: blur(18px);
74
  border-bottom: 1px solid var(--soft-line);
75
  }
 
92
  .mark {
93
  width: 34px;
94
  height: 34px;
95
+ border: 1px solid rgba(164, 242, 127, 0.48);
96
  display: grid;
97
  place-items: center;
98
+ color: var(--green);
99
+ background: #061006;
100
  font-family: var(--font-mono);
101
  font-size: 13px;
102
  font-weight: 760;
 
106
  align-items: center;
107
  gap: 20px;
108
  font-size: 14px;
109
+ color: #c6d0c2;
110
  }
111
  .nav-links a {
112
  text-decoration: none;
 
114
  }
115
  .nav-links a:hover { color: var(--ink); transform: translateY(-1px); }
116
  a:focus-visible, button:focus-visible {
117
+ outline: 2px solid rgba(164, 242, 127, 0.54);
118
  outline-offset: 3px;
119
  }
120
  .nav-action {
121
  border: 1px solid var(--ink);
122
+ background: var(--green);
123
+ color: #020502;
124
  height: 36px;
125
  padding: 0 14px;
126
  display: inline-flex;
 
134
  .hero {
135
  border-bottom: 1px solid var(--soft-line);
136
  background:
137
+ radial-gradient(circle at 78% 28%, rgba(164, 242, 127, 0.20), transparent 18%),
138
+ radial-gradient(circle at 58% 36%, rgba(255, 255, 255, 0.10) 1px, transparent 2px),
139
+ linear-gradient(180deg, #020502, #020702 68%, #050b05);
140
+ background-size: auto, 16px 16px, auto;
141
  }
142
  .hero-inner {
143
  min-height: 680px;
 
152
  align-items: center;
153
  gap: 9px;
154
  margin-bottom: 22px;
155
+ color: var(--green);
156
  font-family: var(--font-mono);
157
  font-size: 12px;
158
  text-transform: uppercase;
 
162
  content: "";
163
  width: 34px;
164
  height: 1px;
165
+ background: var(--green);
166
  }
167
  h1 {
168
  margin: 0;
 
172
  letter-spacing: 0;
173
  max-width: 860px;
174
  text-wrap: balance;
175
+ color: #f7fff0;
176
  }
177
  .hero-copy {
178
  margin: 26px 0 0;
179
  max-width: 690px;
180
+ color: #c7d1c3;
181
  font-size: 19px;
182
  line-height: 1.65;
183
  text-wrap: pretty;
 
198
  gap: 10px;
199
  text-decoration: none;
200
  font-weight: 720;
201
+ background: rgba(7, 18, 7, 0.82);
202
  transition: transform 240ms cubic-bezier(0.16, 1, 0.3, 1), border-color 240ms cubic-bezier(0.16, 1, 0.3, 1), background 240ms cubic-bezier(0.16, 1, 0.3, 1);
203
  }
204
+ .button:hover { transform: translateY(-2px); border-color: var(--green); background: rgba(164, 242, 127, 0.10); }
205
  .button:active { transform: translateY(0) scale(0.98); }
206
  .button.primary {
207
  background: var(--ink);
208
+ color: #020502;
209
+ border-color: var(--green);
210
+ background: var(--green);
211
  }
212
  .hero-stats {
213
  display: grid;
 
218
  }
219
  .stat {
220
  border: 1px solid var(--line);
221
+ background: rgba(7, 18, 7, 0.82);
222
  padding: 14px 14px 13px;
223
  border-radius: var(--radius);
224
  }
225
  .stat strong { display: block; font-family: var(--font-mono); font-size: 21px; line-height: 1; font-variant-numeric: tabular-nums; }
226
  .stat span { display: block; margin-top: 7px; font-size: 12px; color: var(--muted); }
227
  .hero-panel {
228
+ background: rgba(7, 18, 7, 0.88);
229
  border: 1px solid var(--line);
230
  border-radius: var(--radius);
231
  box-shadow: var(--shadow);
 
251
  }
252
  .signal:last-child { border-bottom: 0; }
253
  .signal code {
254
+ color: #e7f3df;
255
+ background: rgba(164, 242, 127, 0.08);
256
  border: 1px solid var(--soft-line);
257
  padding: 5px 7px;
258
  border-radius: 5px;
 
262
  .track {
263
  position: relative;
264
  height: 12px;
265
+ background: rgba(164, 242, 127, 0.16);
266
  overflow: hidden;
267
  border-radius: 999px;
268
  }
 
275
  }
276
  .signal strong { text-align: right; font-family: var(--font-mono); font-size: 13px; font-variant-numeric: tabular-nums; }
277
 
278
+ section { padding: 88px 0 96px; border-bottom: 1px solid var(--soft-line); background: rgba(2, 5, 2, 0.72); }
279
  #suite { padding: 62px 0 76px; }
280
  .section-head {
281
  display: flex;
 
309
  border: 1px solid var(--line);
310
  border-radius: var(--radius);
311
  background: var(--surface);
312
+ box-shadow: 0 18px 46px rgba(0, 0, 0, 0.36);
313
  }
314
  .task-suite-image {
315
  display: block;
 
334
  border: 1px solid var(--line);
335
  border-radius: var(--radius);
336
  padding: 24px;
337
+ background: rgba(164, 242, 127, 0.06);
338
+ color: #eaf5e5;
339
  }
340
  .callout h3, .artifact h3 { margin: 0 0 8px; font-size: 17px; }
341
+ .callout p, .artifact p { margin: 0; color: #aab5a5; line-height: 1.6; }
342
 
343
  .models {
344
  display: grid;
 
353
  background: var(--surface);
354
  transition: transform 240ms cubic-bezier(0.16, 1, 0.3, 1), box-shadow 240ms cubic-bezier(0.16, 1, 0.3, 1);
355
  }
356
+ .model:hover { transform: translateY(-3px); box-shadow: 0 18px 38px rgba(164, 242, 127, 0.08); }
357
  .model h3 { margin: 0; font-size: 15px; }
358
  .model .score { display: block; margin-top: 18px; font-family: var(--font-mono); font-size: 33px; font-weight: 760; line-height: 1; font-variant-numeric: tabular-nums; }
359
  .model .meta { display: block; margin-top: 8px; color: var(--muted); font-size: 13px; }
 
371
  height: 36px;
372
  padding: 0 14px;
373
  font-weight: 650;
374
+ color: #d7e3d0;
375
  cursor: pointer;
376
  transition: transform 220ms cubic-bezier(0.16, 1, 0.3, 1), background 220ms cubic-bezier(0.16, 1, 0.3, 1);
377
  }
378
  .filter:hover { transform: translateY(-1px); }
379
+ .filter.active { color: #020502; background: var(--green); border-color: var(--green); }
380
  .task-grid {
381
  display: grid;
382
  grid-template-columns: repeat(3, minmax(0, 1fr));
 
392
  min-height: 184px;
393
  transition: transform 240ms cubic-bezier(0.16, 1, 0.3, 1), border-color 240ms cubic-bezier(0.16, 1, 0.3, 1);
394
  }
395
+ .task-card:hover { transform: translateY(-3px); border-color: var(--green); }
396
  .task-card.hide { display: none; }
397
  .task-top { display: flex; justify-content: space-between; gap: 14px; align-items: start; }
398
  .task-name { font-family: var(--font-mono); font-size: 13px; font-weight: 760; }
 
400
  font-size: 11px;
401
  border-radius: 999px;
402
  padding: 4px 8px;
403
+ color: #d8ead2;
404
+ background: rgba(164, 242, 127, 0.08);
405
  white-space: nowrap;
406
  }
407
+ .tag.supervised { background: rgba(155, 223, 255, 0.12); color: #9bdfff; }
408
+ .tag.forecast { background: rgba(164, 242, 127, 0.12); color: #a7f078; }
409
+ .tag.retrieval { background: rgba(122, 229, 195, 0.12); color: #7ae5c3; }
410
+ .tag.diagnostic { background: rgba(216, 244, 165, 0.12); color: #d8f4a5; }
411
  .task-card p { margin: 0; color: var(--muted); font-size: 13px; }
412
  .metric-row { display: flex; justify-content: space-between; gap: 12px; font-size: 13px; }
413
  .metric-row strong { font-family: var(--font-mono); font-size: 18px; font-variant-numeric: tabular-nums; }
414
+ .mini-bar { height: 7px; background: rgba(164, 242, 127, 0.14); border-radius: 999px; overflow: hidden; }
415
  .mini-bar span { display: block; height: 100%; width: var(--w); background: var(--c); }
416
 
417
  .artifact-grid {
 
447
  width: fit-content;
448
  border-radius: 999px;
449
  padding: 4px 9px;
450
+ background: rgba(164, 242, 127, 0.10);
451
+ color: var(--green);
452
  font-size: 11px;
453
  font-weight: 740;
454
  }
 
457
  grid-template-columns: repeat(3, minmax(0, 1fr));
458
  gap: 8px;
459
  font-size: 12px;
460
+ color: #bcc8b7;
461
  }
462
  .direction-counts strong {
463
  display: block;
 
494
  grid-template-columns: repeat(2, minmax(0, 1fr));
495
  gap: 8px;
496
  font-size: 12px;
497
+ color: var(--muted);
498
  }
499
  .extension-metrics strong {
500
  display: block;
 
527
  font-size: 12px;
528
  }
529
  .walk-flow span {
530
+ border: 1px solid var(--soft-line);
531
+ background: rgba(164, 242, 127, 0.06);
532
  border-radius: 6px;
533
  padding: 9px;
534
  min-height: 58px;
 
541
  padding: 18px;
542
  transition: transform 240ms cubic-bezier(0.16, 1, 0.3, 1), border-color 240ms cubic-bezier(0.16, 1, 0.3, 1);
543
  }
544
+ .artifact:hover { transform: translateY(-3px); border-color: var(--green); }
545
  .artifact a { display: inline-block; margin-top: 14px; font-weight: 740; text-decoration: none; color: var(--blue); }
546
  .artifact a:hover { text-decoration: underline; text-underline-offset: 4px; }
547
 
548
  .code-panel {
549
+ background: #000;
550
+ color: #dff7d4;
551
  border-radius: var(--radius);
552
  padding: 18px;
553
  overflow: auto;
554
  font-family: var(--font-mono);
555
  font-size: 13px;
556
  line-height: 1.65;
557
+ border: 1px solid rgba(164, 242, 127, 0.24);
558
  }
559
  .code-panel button {
560
  float: right;
561
  margin-left: 16px;
562
  height: 30px;
563
+ border: 1px solid rgba(164, 242, 127, 0.36);
564
+ color: #020502;
565
+ background: var(--green);
566
  border-radius: 5px;
567
  cursor: pointer;
568
  font-weight: 700;
 
653
  <span>current feature allocation</span>
654
  <span>window vector</span>
655
  </div>
656
+ <div class="signal"><code>mocap</code><div class="track"><span style="--w:25.3%;--c:#a7f078"></span></div><strong>2,121</strong></div>
657
+ <div class="signal"><code>camera+imu</code><div class="track"><span style="--w:1.5%;--c:#7ae5c3"></span></div><strong>126</strong></div>
658
+ <div class="signal"><code>depth</code><div class="track"><span style="--w:11.7%;--c:#d8f4a5"></span></div><strong>980</strong></div>
659
+ <div class="signal"><code>video</code><div class="track"><span style="--w:49.1%;--c:#9bdfff"></span></div><strong>4,116</strong></div>
660
+ <div class="signal"><code>language</code><div class="track"><span style="--w:10.7%;--c:#f4f8ef"></span></div><strong>896</strong></div>
661
+ <div class="signal"><code>static</code><div class="track"><span style="--w:1.7%;--c:#a5afa2"></span></div><strong>139</strong></div>
662
  </div>
663
  </div>
664
  </header>
 
876
  <button class="filter" data-filter="diagnostic">Diagnostic</button>
877
  </div>
878
  <div class="task-grid" id="taskGrid">
879
+ <article class="task-card" data-kind="supervised"><div class="task-top"><span class="task-name">timeline_action</span><span class="tag supervised">supervised</span></div><p>All featurized modalities to current action label. Chronological split exposes unseen future actions.</p><div class="metric-row"><span>macro-F1</span><strong>0.0500</strong></div><div class="mini-bar"><span style="--w:5%;--c:#9bdfff"></span></div></article>
880
+ <article class="task-card" data-kind="supervised"><div class="task-top"><span class="task-name">timeline_subtask</span><span class="tag supervised">supervised</span></div><p>All featurized modalities to current subtask label. Useful for segmentation diagnostics.</p><div class="metric-row"><span>macro-F1</span><strong>0.0495</strong></div><div class="mini-bar"><span style="--w:5%;--c:#9bdfff"></span></div></article>
881
+ <article class="task-card" data-kind="diagnostic"><div class="task-top"><span class="task-name">transition_detection</span><span class="tag diagnostic">diagnostic</span></div><p>Predict steady vs action boundary. Highlights task-transition localization quality.</p><div class="metric-row"><span>macro-F1</span><strong>0.6552</strong></div><div class="mini-bar"><span style="--w:65.5%;--c:#d8f4a5"></span></div></article>
882
+ <article class="task-card" data-kind="supervised"><div class="task-top"><span class="task-name">next_action</span><span class="tag supervised">supervised</span></div><p>Current multimodal window to action 20 frames later. Tests short-horizon task flow.</p><div class="metric-row"><span>macro-F1</span><strong>0.0593</strong></div><div class="mini-bar"><span style="--w:5.9%;--c:#9bdfff"></span></div></article>
883
+ <article class="task-card" data-kind="forecast"><div class="task-top"><span class="task-name">hand_trajectory_forecast</span><span class="tag forecast">forecast</span></div><p>Predict future left/right hand 3D joints. Closer to imitation-learning style signals.</p><div class="metric-row"><span>MPJPE</span><strong>0.8223</strong></div><div class="mini-bar"><span style="--w:48%;--c:#a7f078"></span></div></article>
884
+ <article class="task-card" data-kind="supervised"><div class="task-top"><span class="task-name">contact_prediction</span><span class="tag supervised">supervised</span></div><p>Non-contact modalities to binary contact. Degenerate in this sample because one class dominates.</p><div class="metric-row"><span>accuracy</span><strong>1.0000</strong></div><div class="mini-bar"><span style="--w:100%;--c:#a7f078"></span></div></article>
885
+ <article class="task-card" data-kind="supervised"><div class="task-top"><span class="task-name">object_relevance</span><span class="tag supervised">supervised</span></div><p>Predict relevant object set from non-caption feature blocks.</p><div class="metric-row"><span>micro-F1</span><strong>0.1839</strong></div><div class="mini-bar"><span style="--w:18.4%;--c:#9bdfff"></span></div></article>
886
+ <article class="task-card" data-kind="retrieval"><div class="task-top"><span class="task-name">caption_grounding</span><span class="tag retrieval">retrieval</span></div><p>Caption objects/interaction query to matching sensor window.</p><div class="metric-row"><span>MRR</span><strong>0.0172</strong></div><div class="mini-bar"><span style="--w:1.7%;--c:#7ae5c3"></span></div></article>
887
+ <article class="task-card" data-kind="retrieval"><div class="task-top"><span class="task-name">cross_modal_retrieval</span><span class="tag retrieval">retrieval</span></div><p>Motion/IMU/camera query retrieves matching depth/video window. Strongest single-episode signal.</p><div class="metric-row"><span>top-5</span><strong>0.3764</strong></div><div class="mini-bar"><span style="--w:37.6%;--c:#7ae5c3"></span></div></article>
888
+ <article class="task-card" data-kind="forecast"><div class="task-top"><span class="task-name">modality_reconstruction</span><span class="tag forecast">forecast</span></div><p>Motion/IMU/camera to depth/video feature vector.</p><div class="metric-row"><span>R2</span><strong>-0.0160</strong></div><div class="mini-bar"><span style="--w:2%;--c:#ff8f7a"></span></div></article>
889
+ <article class="task-card" data-kind="diagnostic"><div class="task-top"><span class="task-name">temporal_order</span><span class="tag diagnostic">diagnostic</span></div><p>Two adjacent windows to correct vs reversed order.</p><div class="metric-row"><span>F1</span><strong>0.5487</strong></div><div class="mini-bar"><span style="--w:54.9%;--c:#d8f4a5"></span></div></article>
890
+ <article class="task-card" data-kind="diagnostic"><div class="task-top"><span class="task-name">misalignment_detection</span><span class="tag diagnostic">diagnostic</span></div><p>Motion+visual pair to aligned vs shifted by eight windows.</p><div class="metric-row"><span>F1</span><strong>0.4866</strong></div><div class="mini-bar"><span style="--w:48.7%;--c:#d8f4a5"></span></div></article>
891
  </div>
892
  </div>
893
  </section>
scripts/generate_visualizations.py CHANGED
@@ -56,25 +56,26 @@ def svg_bar_chart(path: Path, title: str, rows: list[tuple[str, float]], x_label
56
  max_value = max_value if max_value is not None else max([v for _, v in rows] + [1.0])
57
  max_value = max(max_value, 1e-9)
58
  plot_w = width - left - right
59
- colors = ["#2563eb", "#059669", "#ea580c", "#7b5d12", "#0891b2", "#dc2626"]
60
  parts = [
61
  f'<svg xmlns="http://www.w3.org/2000/svg" width="{width}" height="{height}" viewBox="0 0 {width} {height}">',
62
- '<rect width="100%" height="100%" fill="#ffffff"/>',
63
- f'<text x="32" y="42" font-family="Arial, sans-serif" font-size="26" font-weight="700" fill="#111827">{html.escape(title)}</text>',
64
- f'<text x="{left}" y="{height - 24}" font-family="Arial, sans-serif" font-size="13" fill="#6b7280">{html.escape(x_label)}</text>',
 
65
  ]
66
  for tick in range(6):
67
  x = left + plot_w * tick / 5
68
  val = max_value * tick / 5
69
- parts.append(f'<line x1="{x:.1f}" y1="{top - 18}" x2="{x:.1f}" y2="{height - 50}" stroke="#e5e7eb" stroke-width="1"/>')
70
- parts.append(f'<text x="{x:.1f}" y="{height - 30}" text-anchor="middle" font-family="Arial, sans-serif" font-size="12" fill="#6b7280">{val:.2f}</text>')
71
  for i, (label, value) in enumerate(rows):
72
  y = top + i * row_h
73
  bar_w = max(0.0, min(value / max_value, 1.0)) * plot_w
74
  color = colors[i % len(colors)]
75
- parts.append(f'<text x="{left - 14}" y="{y + 21}" text-anchor="end" font-family="Arial, sans-serif" font-size="14" fill="#111827">{html.escape(label)}</text>')
76
  parts.append(f'<rect x="{left}" y="{y + 5}" width="{bar_w:.1f}" height="20" rx="4" fill="{color}"/>')
77
- parts.append(f'<text x="{left + bar_w + 8:.1f}" y="{y + 21}" font-family="Arial, sans-serif" font-size="13" fill="#374151">{value:.4f}</text>')
78
  parts.append("</svg>")
79
  path.write_text("\n".join(parts), encoding="utf-8")
80
 
@@ -94,47 +95,49 @@ def svg_pipeline_diagram(path: Path, summary: dict) -> None:
94
  "annotation.hdf5",
95
  "6 MP4 videos with audio",
96
  f"{suite['num_frames']:,} aligned frames",
97
- ], "#1f63e9"),
98
  (365, 110, 250, 132, "2. HOMIE loader", [
99
  "video, depth, pose",
100
  "mocap, IMU, language",
101
  "audio not featurized",
102
- ], "#008b9a"),
103
  (670, 110, 250, 132, "3. Window builder", [
104
  f"{suite['window_frames']}-frame windows",
105
  f"{suite['stride_frames']}-frame stride",
106
  f"{suite['num_windows']:,} windows",
107
- ], "#0a7f55"),
108
  (975, 110, 300, 132, "4. Feature vector", [
109
  f"{suite['feature_dim']:,} dimensions",
110
  "17 named blocks, no audio block",
111
  "stored manifest",
112
- ], "#b65b04"),
113
  (60, 380, 360, 168, "5. Baseline models", [
114
  "motion-only action/subtask",
115
  "current all-feature action/subtask",
116
  "numpy softmax classifier",
117
  "metrics and predictions",
118
- ], "#1f63e9"),
119
  (520, 380, 360, 168, "6. Ropedia Xperience-10M suite", [
120
  f"{task_count} supervised/self-supervised tasks",
121
  "chronological split",
122
  "retrieval, forecast, alignment",
123
  "per-task artifacts",
124
- ], "#008b9a"),
125
  (980, 380, 300, 168, "7. Published artifacts", [
126
  "results/**/*.json/csv/npz",
127
  "docs/data/summary_metrics.json",
128
  "GitHub Pages dashboard",
129
  "reproducibility audit",
130
- ], "#0a7f55"),
131
  ]
132
  parts = [
133
  f'<svg xmlns="http://www.w3.org/2000/svg" width="{width}" height="{height}" viewBox="0 0 {width} {height}">',
134
- '<rect width="100%" height="100%" fill="#ffffff"/>',
135
- '<rect x="0" y="0" width="1400" height="760" fill="#ffffff"/>',
136
- '<text x="60" y="58" font-family="Arial, sans-serif" font-size="32" font-weight="700" fill="#10141f">Verified Ropedia Xperience-10M Pipeline</text>',
137
- '<text x="60" y="88" font-family="Arial, sans-serif" font-size="16" fill="#5b6475">Generated from committed scripts and metrics; no conceptual placeholder stages.</text>',
 
 
138
  ]
139
  arrows = [
140
  (310, 176, 365, 176),
@@ -146,23 +149,23 @@ def svg_pipeline_diagram(path: Path, summary: dict) -> None:
146
  (880, 464, 980, 464),
147
  ]
148
  for x1, y1, x2, y2 in arrows:
149
- parts.append(f'<line x1="{x1}" y1="{y1}" x2="{x2}" y2="{y2}" stroke="#cbd5e1" stroke-width="3" marker-end="url(#arrow)"/>')
150
- parts.insert(1, '<defs><marker id="arrow" viewBox="0 0 10 10" refX="8" refY="5" markerWidth="7" markerHeight="7" orient="auto-start-reverse"><path d="M 0 0 L 10 5 L 0 10 z" fill="#cbd5e1"/></marker></defs>')
151
  for x, y, w, h, title, lines, color in boxes:
152
- parts.append(f'<rect x="{x}" y="{y}" width="{w}" height="{h}" rx="8" fill="#ffffff" stroke="#dce2ec" stroke-width="2"/>')
153
  parts.append(f'<rect x="{x}" y="{y}" width="8" height="{h}" rx="4" fill="{color}"/>')
154
- parts.append(f'<text x="{x + 24}" y="{y + 34}" font-family="Arial, sans-serif" font-size="18" font-weight="700" fill="#10141f">{html.escape(title)}</text>')
155
  for i, line in enumerate(lines):
156
- parts.append(f'<text x="{x + 24}" y="{y + 66 + i * 22}" font-family="Arial, sans-serif" font-size="14" fill="#394255">{html.escape(line)}</text>')
157
  checks = [
158
  "Audit check: rerunning scripts to /private/tmp reproduced committed metrics exactly.",
159
  "Modality check: sample covers video, AAC audio, depth, pose/SLAM, mocap, IMU, and language annotation.",
160
  "Feature check: current manifest has video/depth/pose/mocap/IMU/language blocks, but no audio block.",
161
  "Scope check: this validates one public sample episode, not cross-episode generalization.",
162
  ]
163
- parts.append('<rect x="60" y="620" width="1220" height="96" rx="8" fill="#f8fafc" stroke="#dce2ec"/>')
164
  for i, line in enumerate(checks):
165
- parts.append(f'<text x="84" y="{650 + i * 24}" font-family="Arial, sans-serif" font-size="15" fill="#273143">{html.escape(line)}</text>')
166
  parts.append("</svg>")
167
  path.write_text("\n".join(parts), encoding="utf-8")
168
 
@@ -209,12 +212,12 @@ def metric_text_with_neural(task_name: str, metrics: dict, neural_tasks: dict) -
209
  return f"min {text}; NN {metric_text(task_name, neural_metrics)}"
210
 
211
 
212
- def draw_text_block(parts: list[str], x: int, y: int, lines: list[str], size: int = 13, color: str = "#394255", weight: str = "500", max_chars: int = 42, line_h: int = 18) -> int:
213
  cursor = y
214
  for line in lines:
215
  wrapped = textwrap.wrap(line, width=max_chars) or [""]
216
  for item in wrapped:
217
- parts.append(f'<text x="{x}" y="{cursor}" font-family="Arial, sans-serif" font-size="{size}" font-weight="{weight}" fill="{color}">{html.escape(item)}</text>')
218
  cursor += line_h
219
  return cursor
220
 
@@ -339,18 +342,20 @@ def svg_task_architectures(path: Path, summary: dict) -> None:
339
  suite = summary["suite"]
340
  rows = task_architecture_rows(summary)
341
  family_colors = {
342
- "softmax": "#1f63e9",
343
- "ridge": "#0a7f55",
344
- "ridge+rank": "#008b9a",
345
- "multilabel": "#b65b04",
346
  }
347
  width, height = 1500, 1840
348
  parts = [
349
  f'<svg xmlns="http://www.w3.org/2000/svg" width="{width}" height="{height}" viewBox="0 0 {width} {height}">',
350
- '<defs><marker id="arrow2" viewBox="0 0 10 10" refX="8" refY="5" markerWidth="7" markerHeight="7" orient="auto-start-reverse"><path d="M 0 0 L 10 5 L 0 10 z" fill="#cbd5e1"/></marker></defs>',
351
- '<rect width="100%" height="100%" fill="#ffffff"/>',
352
- '<text x="60" y="56" font-family="Arial, sans-serif" font-size="34" font-weight="700" fill="#10141f">Minimal Architectures for 12 Ropedia Xperience-10M Tasks</text>',
353
- '<text x="60" y="88" font-family="Arial, sans-serif" font-size="16" fill="#5b6475">Generated from scripts/episode_task_suite.py semantics and committed summary metrics. These are minimal baselines, not deep foundation models.</text>',
 
 
354
  ]
355
 
356
  setup = [
@@ -358,44 +363,44 @@ def svg_task_architectures(path: Path, summary: dict) -> None:
358
  f"{suite['num_frames']:,} frames -> {suite['num_windows']:,} windows",
359
  f"{suite['window_frames']}-frame window, {suite['stride_frames']}-frame stride",
360
  "chronological 70/30 split",
361
- ], "#1f63e9"),
362
  (410, 122, 310, 110, "Feature vector", [
363
  f"X_all = {suite['feature_dim']:,} dimensions",
364
  "17 named blocks; no audio block",
365
  "mean/std fit on train only",
366
- ], "#008b9a"),
367
  (760, 122, 320, 110, "Reusable heads", [
368
  "linear softmax classifier",
369
  "dual ridge regression/projection",
370
  "multi-label logistic + cosine rank",
371
- ], "#0a7f55"),
372
  (1120, 122, 320, 110, "Artifacts", [
373
  "metrics.json, predictions.csv/npz",
374
  "model.npz with scaler and weights",
375
  "summary_report.json source of numbers",
376
- ], "#b65b04"),
377
  ]
378
  for i in range(len(setup) - 1):
379
  x1 = setup[i][0] + setup[i][2]
380
  x2 = setup[i + 1][0]
381
  y = setup[i][1] + 55
382
- parts.append(f'<line x1="{x1 + 12}" y1="{y}" x2="{x2 - 14}" y2="{y}" stroke="#cbd5e1" stroke-width="3" marker-end="url(#arrow2)"/>')
383
  for x, y, w, h, title, lines, color in setup:
384
- parts.append(f'<rect x="{x}" y="{y}" width="{w}" height="{h}" rx="8" fill="#ffffff" stroke="#dce2ec" stroke-width="2"/>')
385
  parts.append(f'<rect x="{x}" y="{y}" width="8" height="{h}" rx="4" fill="{color}"/>')
386
- parts.append(f'<text x="{x + 24}" y="{y + 31}" font-family="Arial, sans-serif" font-size="18" font-weight="700" fill="#10141f">{html.escape(title)}</text>')
387
- draw_text_block(parts, x + 24, y + 58, lines, size=13, color="#394255", max_chars=34, line_h=18)
388
 
389
  families = [
390
- ("Softmax classifier", "logits = z(X)W + b; CE + L2; class weights for classifiers", "#1f63e9", 60, 270),
391
- ("Ridge regression/projection", "closed-form dual ridge on z(X), z(Y); used for forecast and reconstruction", "#0a7f55", 780, 270),
392
- ("Ridge + cosine ranking", "project one modality into another feature space, then rank candidates by cosine", "#008b9a", 60, 394),
393
- ("Multi-label logistic", "sigmoid heads for object vocabulary; threshold 0.5 with top-1 fallback", "#b65b04", 780, 394),
394
  ]
395
  for title, desc, color, x, y in families:
396
- parts.append(f'<rect x="{x}" y="{y}" width="660" height="100" rx="8" fill="#f8fafc" stroke="#dce2ec"/>')
397
- parts.append(f'<text x="{x + 18}" y="{y + 33}" font-family="Arial, sans-serif" font-size="18" font-weight="700" fill="{color}">{html.escape(title)}</text>')
398
- draw_text_block(parts, x + 18, y + 60, [desc], size=13, color="#394255", max_chars=76, line_h=18)
399
 
400
  card_w, card_h = 440, 248
401
  gap_x, gap_y = 30, 30
@@ -405,24 +410,24 @@ def svg_task_architectures(path: Path, summary: dict) -> None:
405
  x = start_x + col * (card_w + gap_x)
406
  y = start_y + card_row * (card_h + gap_y)
407
  color = family_colors[row["family"]]
408
- parts.append(f'<rect x="{x}" y="{y}" width="{card_w}" height="{card_h}" rx="8" fill="#ffffff" stroke="#dce2ec" stroke-width="2"/>')
409
  parts.append(f'<rect x="{x}" y="{y}" width="8" height="{card_h}" rx="4" fill="{color}"/>')
410
- parts.append(f'<rect x="{x + 20}" y="{y + 18}" width="96" height="24" rx="6" fill="#f8fafc" stroke="{color}"/>')
411
- parts.append(f'<text x="{x + 68}" y="{y + 35}" text-anchor="middle" font-family="Arial, sans-serif" font-size="11" font-weight="700" fill="{color}">{html.escape(row["family"])}</text>')
412
- parts.append(f'<text x="{x + 20}" y="{y + 72}" font-family="Arial, sans-serif" font-size="20" font-weight="700" fill="#10141f">{html.escape(row["task"])}</text>')
413
  cursor = y + 104
414
  for label in ("input", "head", "output", "metric"):
415
- parts.append(f'<text x="{x + 20}" y="{cursor}" font-family="Arial, sans-serif" font-size="12" font-weight="700" fill="{color}">{label.upper()}</text>')
416
- cursor = draw_text_block(parts, x + 92, cursor, [row[label]], size=13, color="#394255", max_chars=41, line_h=17)
417
  cursor += 8
418
 
419
  notes = [
420
  "Interpretation: this suite tests whether each input/output contract is wired correctly before scaling to many episodes.",
421
  "Research-grade claims need held-out episode splits and stronger sequence/vision-language/robot-policy models.",
422
  ]
423
- parts.append('<rect x="60" y="1688" width="1380" height="72" rx="8" fill="#f8fafc" stroke="#dce2ec"/>')
424
  for i, line in enumerate(notes):
425
- parts.append(f'<text x="84" y="{1718 + i * 24}" font-family="Arial, sans-serif" font-size="15" fill="#273143">{html.escape(line)}</text>')
426
  parts.append("</svg>")
427
  path.write_text("\n".join(parts), encoding="utf-8")
428
 
 
56
  max_value = max_value if max_value is not None else max([v for _, v in rows] + [1.0])
57
  max_value = max(max_value, 1e-9)
58
  plot_w = width - left - right
59
+ colors = ["#a7f078", "#ffffff", "#7ae5c3", "#d8f4a5", "#9bdfff", "#ff8f7a"]
60
  parts = [
61
  f'<svg xmlns="http://www.w3.org/2000/svg" width="{width}" height="{height}" viewBox="0 0 {width} {height}">',
62
+ '<rect width="100%" height="100%" fill="#020502"/>',
63
+ '<rect x="18" y="18" width="1064" height="' + str(height - 36) + '" rx="18" fill="#050905" stroke="#a7f078" stroke-opacity="0.25"/>',
64
+ f'<text x="32" y="42" font-family="Inter Tight, Arial, sans-serif" font-size="26" font-weight="800" fill="#f4f8ef">{html.escape(title)}</text>',
65
+ f'<text x="{left}" y="{height - 24}" font-family="Space Grotesk, Arial, sans-serif" font-size="13" fill="#a5afa2">{html.escape(x_label)}</text>',
66
  ]
67
  for tick in range(6):
68
  x = left + plot_w * tick / 5
69
  val = max_value * tick / 5
70
+ parts.append(f'<line x1="{x:.1f}" y1="{top - 18}" x2="{x:.1f}" y2="{height - 50}" stroke="#a7f078" stroke-opacity="0.13" stroke-width="1"/>')
71
+ parts.append(f'<text x="{x:.1f}" y="{height - 30}" text-anchor="middle" font-family="Space Grotesk, Arial, sans-serif" font-size="12" fill="#a5afa2">{val:.2f}</text>')
72
  for i, (label, value) in enumerate(rows):
73
  y = top + i * row_h
74
  bar_w = max(0.0, min(value / max_value, 1.0)) * plot_w
75
  color = colors[i % len(colors)]
76
+ parts.append(f'<text x="{left - 14}" y="{y + 21}" text-anchor="end" font-family="Space Grotesk, Arial, sans-serif" font-size="14" fill="#dce8d7">{html.escape(label)}</text>')
77
  parts.append(f'<rect x="{left}" y="{y + 5}" width="{bar_w:.1f}" height="20" rx="4" fill="{color}"/>')
78
+ parts.append(f'<text x="{left + bar_w + 8:.1f}" y="{y + 21}" font-family="Space Grotesk, Arial, sans-serif" font-size="13" fill="#f4f8ef">{value:.4f}</text>')
79
  parts.append("</svg>")
80
  path.write_text("\n".join(parts), encoding="utf-8")
81
 
 
95
  "annotation.hdf5",
96
  "6 MP4 videos with audio",
97
  f"{suite['num_frames']:,} aligned frames",
98
+ ], "#9bdfff"),
99
  (365, 110, 250, 132, "2. HOMIE loader", [
100
  "video, depth, pose",
101
  "mocap, IMU, language",
102
  "audio not featurized",
103
+ ], "#7ae5c3"),
104
  (670, 110, 250, 132, "3. Window builder", [
105
  f"{suite['window_frames']}-frame windows",
106
  f"{suite['stride_frames']}-frame stride",
107
  f"{suite['num_windows']:,} windows",
108
+ ], "#a7f078"),
109
  (975, 110, 300, 132, "4. Feature vector", [
110
  f"{suite['feature_dim']:,} dimensions",
111
  "17 named blocks, no audio block",
112
  "stored manifest",
113
+ ], "#d8f4a5"),
114
  (60, 380, 360, 168, "5. Baseline models", [
115
  "motion-only action/subtask",
116
  "current all-feature action/subtask",
117
  "numpy softmax classifier",
118
  "metrics and predictions",
119
+ ], "#9bdfff"),
120
  (520, 380, 360, 168, "6. Ropedia Xperience-10M suite", [
121
  f"{task_count} supervised/self-supervised tasks",
122
  "chronological split",
123
  "retrieval, forecast, alignment",
124
  "per-task artifacts",
125
+ ], "#7ae5c3"),
126
  (980, 380, 300, 168, "7. Published artifacts", [
127
  "results/**/*.json/csv/npz",
128
  "docs/data/summary_metrics.json",
129
  "GitHub Pages dashboard",
130
  "reproducibility audit",
131
+ ], "#a7f078"),
132
  ]
133
  parts = [
134
  f'<svg xmlns="http://www.w3.org/2000/svg" width="{width}" height="{height}" viewBox="0 0 {width} {height}">',
135
+ '<rect width="100%" height="100%" fill="#020502"/>',
136
+ '<rect x="0" y="0" width="1400" height="760" fill="#020502"/>',
137
+ '<rect x="0" y="0" width="1400" height="760" fill="url(#dotgrid)" opacity="0.55"/>',
138
+ '<circle cx="1120" cy="132" r="170" fill="#a7f078" opacity="0.10"/>',
139
+ '<text x="60" y="58" font-family="Inter Tight, Arial, sans-serif" font-size="32" font-weight="800" fill="#f4f8ef">Verified Ropedia Xperience-10M Pipeline</text>',
140
+ '<text x="60" y="88" font-family="Space Grotesk, Arial, sans-serif" font-size="16" fill="#a5afa2">Generated from committed scripts and metrics; no conceptual placeholder stages.</text>',
141
  ]
142
  arrows = [
143
  (310, 176, 365, 176),
 
149
  (880, 464, 980, 464),
150
  ]
151
  for x1, y1, x2, y2 in arrows:
152
+ parts.append(f'<line x1="{x1}" y1="{y1}" x2="{x2}" y2="{y2}" stroke="#a7f078" stroke-opacity="0.54" stroke-width="3" marker-end="url(#arrow)"/>')
153
+ parts.insert(1, '<defs><pattern id="dotgrid" width="18" height="18" patternUnits="userSpaceOnUse"><circle cx="2" cy="2" r="1.2" fill="#a7f078" opacity="0.20"/></pattern><marker id="arrow" viewBox="0 0 10 10" refX="8" refY="5" markerWidth="7" markerHeight="7" orient="auto-start-reverse"><path d="M 0 0 L 10 5 L 0 10 z" fill="#a7f078" fill-opacity="0.72"/></marker></defs>')
154
  for x, y, w, h, title, lines, color in boxes:
155
+ parts.append(f'<rect x="{x}" y="{y}" width="{w}" height="{h}" rx="8" fill="#061006" stroke="#a7f078" stroke-opacity="0.26" stroke-width="2"/>')
156
  parts.append(f'<rect x="{x}" y="{y}" width="8" height="{h}" rx="4" fill="{color}"/>')
157
+ parts.append(f'<text x="{x + 24}" y="{y + 34}" font-family="Inter Tight, Arial, sans-serif" font-size="18" font-weight="800" fill="#f4f8ef">{html.escape(title)}</text>')
158
  for i, line in enumerate(lines):
159
+ parts.append(f'<text x="{x + 24}" y="{y + 66 + i * 22}" font-family="Space Grotesk, Arial, sans-serif" font-size="14" fill="#dce8d7">{html.escape(line)}</text>')
160
  checks = [
161
  "Audit check: rerunning scripts to /private/tmp reproduced committed metrics exactly.",
162
  "Modality check: sample covers video, AAC audio, depth, pose/SLAM, mocap, IMU, and language annotation.",
163
  "Feature check: current manifest has video/depth/pose/mocap/IMU/language blocks, but no audio block.",
164
  "Scope check: this validates one public sample episode, not cross-episode generalization.",
165
  ]
166
+ parts.append('<rect x="60" y="620" width="1220" height="96" rx="8" fill="#071207" stroke="#a7f078" stroke-opacity="0.24"/>')
167
  for i, line in enumerate(checks):
168
+ parts.append(f'<text x="84" y="{650 + i * 24}" font-family="Space Grotesk, Arial, sans-serif" font-size="15" fill="#dce8d7">{html.escape(line)}</text>')
169
  parts.append("</svg>")
170
  path.write_text("\n".join(parts), encoding="utf-8")
171
 
 
212
  return f"min {text}; NN {metric_text(task_name, neural_metrics)}"
213
 
214
 
215
+ def draw_text_block(parts: list[str], x: int, y: int, lines: list[str], size: int = 13, color: str = "#dce8d7", weight: str = "500", max_chars: int = 42, line_h: int = 18) -> int:
216
  cursor = y
217
  for line in lines:
218
  wrapped = textwrap.wrap(line, width=max_chars) or [""]
219
  for item in wrapped:
220
+ parts.append(f'<text x="{x}" y="{cursor}" font-family="Space Grotesk, Arial, sans-serif" font-size="{size}" font-weight="{weight}" fill="{color}">{html.escape(item)}</text>')
221
  cursor += line_h
222
  return cursor
223
 
 
342
  suite = summary["suite"]
343
  rows = task_architecture_rows(summary)
344
  family_colors = {
345
+ "softmax": "#9bdfff",
346
+ "ridge": "#a7f078",
347
+ "ridge+rank": "#7ae5c3",
348
+ "multilabel": "#d8f4a5",
349
  }
350
  width, height = 1500, 1840
351
  parts = [
352
  f'<svg xmlns="http://www.w3.org/2000/svg" width="{width}" height="{height}" viewBox="0 0 {width} {height}">',
353
+ '<defs><pattern id="dotgrid2" width="18" height="18" patternUnits="userSpaceOnUse"><circle cx="2" cy="2" r="1.2" fill="#a7f078" opacity="0.18"/></pattern><marker id="arrow2" viewBox="0 0 10 10" refX="8" refY="5" markerWidth="7" markerHeight="7" orient="auto-start-reverse"><path d="M 0 0 L 10 5 L 0 10 z" fill="#a7f078" fill-opacity="0.72"/></marker></defs>',
354
+ '<rect width="100%" height="100%" fill="#020502"/>',
355
+ '<rect width="100%" height="100%" fill="url(#dotgrid2)" opacity="0.58"/>',
356
+ '<circle cx="1190" cy="150" r="210" fill="#a7f078" opacity="0.08"/>',
357
+ '<text x="60" y="56" font-family="Inter Tight, Arial, sans-serif" font-size="34" font-weight="800" fill="#f4f8ef">Minimal Architectures for 12 Ropedia Xperience-10M Tasks</text>',
358
+ '<text x="60" y="88" font-family="Space Grotesk, Arial, sans-serif" font-size="16" fill="#a5afa2">Generated from scripts/episode_task_suite.py semantics and committed summary metrics. These are minimal baselines, not deep foundation models.</text>',
359
  ]
360
 
361
  setup = [
 
363
  f"{suite['num_frames']:,} frames -> {suite['num_windows']:,} windows",
364
  f"{suite['window_frames']}-frame window, {suite['stride_frames']}-frame stride",
365
  "chronological 70/30 split",
366
+ ], "#9bdfff"),
367
  (410, 122, 310, 110, "Feature vector", [
368
  f"X_all = {suite['feature_dim']:,} dimensions",
369
  "17 named blocks; no audio block",
370
  "mean/std fit on train only",
371
+ ], "#7ae5c3"),
372
  (760, 122, 320, 110, "Reusable heads", [
373
  "linear softmax classifier",
374
  "dual ridge regression/projection",
375
  "multi-label logistic + cosine rank",
376
+ ], "#a7f078"),
377
  (1120, 122, 320, 110, "Artifacts", [
378
  "metrics.json, predictions.csv/npz",
379
  "model.npz with scaler and weights",
380
  "summary_report.json source of numbers",
381
+ ], "#d8f4a5"),
382
  ]
383
  for i in range(len(setup) - 1):
384
  x1 = setup[i][0] + setup[i][2]
385
  x2 = setup[i + 1][0]
386
  y = setup[i][1] + 55
387
+ parts.append(f'<line x1="{x1 + 12}" y1="{y}" x2="{x2 - 14}" y2="{y}" stroke="#a7f078" stroke-opacity="0.54" stroke-width="3" marker-end="url(#arrow2)"/>')
388
  for x, y, w, h, title, lines, color in setup:
389
+ parts.append(f'<rect x="{x}" y="{y}" width="{w}" height="{h}" rx="8" fill="#061006" stroke="#a7f078" stroke-opacity="0.26" stroke-width="2"/>')
390
  parts.append(f'<rect x="{x}" y="{y}" width="8" height="{h}" rx="4" fill="{color}"/>')
391
+ parts.append(f'<text x="{x + 24}" y="{y + 31}" font-family="Inter Tight, Arial, sans-serif" font-size="18" font-weight="800" fill="#f4f8ef">{html.escape(title)}</text>')
392
+ draw_text_block(parts, x + 24, y + 58, lines, size=13, color="#dce8d7", max_chars=34, line_h=18)
393
 
394
  families = [
395
+ ("Softmax classifier", "logits = z(X)W + b; CE + L2; class weights for classifiers", "#9bdfff", 60, 270),
396
+ ("Ridge regression/projection", "closed-form dual ridge on z(X), z(Y); used for forecast and reconstruction", "#a7f078", 780, 270),
397
+ ("Ridge + cosine ranking", "project one modality into another feature space, then rank candidates by cosine", "#7ae5c3", 60, 394),
398
+ ("Multi-label logistic", "sigmoid heads for object vocabulary; threshold 0.5 with top-1 fallback", "#d8f4a5", 780, 394),
399
  ]
400
  for title, desc, color, x, y in families:
401
+ parts.append(f'<rect x="{x}" y="{y}" width="660" height="100" rx="8" fill="#071207" stroke="#a7f078" stroke-opacity="0.22"/>')
402
+ parts.append(f'<text x="{x + 18}" y="{y + 33}" font-family="Inter Tight, Arial, sans-serif" font-size="18" font-weight="800" fill="{color}">{html.escape(title)}</text>')
403
+ draw_text_block(parts, x + 18, y + 60, [desc], size=13, color="#dce8d7", max_chars=76, line_h=18)
404
 
405
  card_w, card_h = 440, 248
406
  gap_x, gap_y = 30, 30
 
410
  x = start_x + col * (card_w + gap_x)
411
  y = start_y + card_row * (card_h + gap_y)
412
  color = family_colors[row["family"]]
413
+ parts.append(f'<rect x="{x}" y="{y}" width="{card_w}" height="{card_h}" rx="8" fill="#061006" stroke="#a7f078" stroke-opacity="0.24" stroke-width="2"/>')
414
  parts.append(f'<rect x="{x}" y="{y}" width="8" height="{card_h}" rx="4" fill="{color}"/>')
415
+ parts.append(f'<rect x="{x + 20}" y="{y + 18}" width="96" height="24" rx="6" fill="#071207" stroke="{color}" stroke-opacity="0.72"/>')
416
+ parts.append(f'<text x="{x + 68}" y="{y + 35}" text-anchor="middle" font-family="Space Grotesk, Arial, sans-serif" font-size="11" font-weight="800" fill="{color}">{html.escape(row["family"])}</text>')
417
+ parts.append(f'<text x="{x + 20}" y="{y + 72}" font-family="Inter Tight, Arial, sans-serif" font-size="20" font-weight="800" fill="#f4f8ef">{html.escape(row["task"])}</text>')
418
  cursor = y + 104
419
  for label in ("input", "head", "output", "metric"):
420
+ parts.append(f'<text x="{x + 20}" y="{cursor}" font-family="Space Grotesk, Arial, sans-serif" font-size="12" font-weight="800" fill="{color}">{label.upper()}</text>')
421
+ cursor = draw_text_block(parts, x + 92, cursor, [row[label]], size=13, color="#dce8d7", max_chars=41, line_h=17)
422
  cursor += 8
423
 
424
  notes = [
425
  "Interpretation: this suite tests whether each input/output contract is wired correctly before scaling to many episodes.",
426
  "Research-grade claims need held-out episode splits and stronger sequence/vision-language/robot-policy models.",
427
  ]
428
+ parts.append('<rect x="60" y="1688" width="1380" height="72" rx="8" fill="#071207" stroke="#a7f078" stroke-opacity="0.22"/>')
429
  for i, line in enumerate(notes):
430
+ parts.append(f'<text x="84" y="{1718 + i * 24}" font-family="Space Grotesk, Arial, sans-serif" font-size="15" fill="#dce8d7">{html.escape(line)}</text>')
431
  parts.append("</svg>")
432
  path.write_text("\n".join(parts), encoding="utf-8")
433
 
scripts/render_overview_figures.py CHANGED
@@ -28,33 +28,33 @@ DEFAULT_PIPELINE_OUTPUT = ASSETS / "pipeline_diagram.png"
28
  DEFAULT_ARCHITECTURE_OUTPUT = ASSETS / "task_architectures.png"
29
 
30
  PIPELINE_WIDTH = 1800
31
- PIPELINE_HEIGHT = 1000
32
  ARCHITECTURE_WIDTH = 1800
33
  ARCHITECTURE_HEIGHT = 1520
34
 
35
 
36
  COLORS = {
37
- "blue": "#1f6c9f",
38
- "teal": "#197d83",
39
- "green": "#346538",
40
- "amber": "#956400",
41
- "orange": "#b65b04",
42
- "red": "#9f2f2d",
43
- "ink": "#1f2421",
44
- "muted": "#5f625d",
45
- "line": "#e4ded4",
46
  }
47
 
48
 
49
  TASK_GROUPS = [
50
- ("Label + State", "#197d83", ["timeline_action", "timeline_subtask", "next_action"]),
51
  (
52
  "Prediction + Reconstruction",
53
- "#1f6c9f",
54
  ["hand_trajectory_forecast", "modality_reconstruction", "contact_prediction"],
55
  ),
56
- ("Grounding + Retrieval", "#956400", ["caption_grounding", "cross_modal_retrieval", "object_relevance"]),
57
- ("Temporal Diagnostics", "#9f2f2d", ["transition_detection", "temporal_order", "misalignment_detection"]),
58
  ]
59
 
60
 
@@ -168,30 +168,30 @@ def build_pipeline_html(summary: dict, base_path: Path) -> str:
168
  <meta charset="utf-8">
169
  <style>
170
  * {{ box-sizing: border-box; }}
171
- body {{ margin: 0; background: #fbfaf7; font-family: "Avenir Next", "SF Pro Display", Arial, sans-serif; }}
172
  .canvas {{
173
  position: relative;
174
  width: {PIPELINE_WIDTH}px;
175
  height: {PIPELINE_HEIGHT}px;
176
  overflow: hidden;
177
- color: #1f2421;
178
  background:
179
- linear-gradient(90deg, rgba(68,55,38,0.03) 1px, transparent 1px),
180
- linear-gradient(0deg, rgba(68,55,38,0.024) 1px, transparent 1px),
181
- #fbfaf7;
182
- background-size: 58px 58px, 58px 58px, auto;
183
  }}
184
  .base-layer {{
185
  position: absolute;
186
  inset: 0;
187
  background-size: cover;
188
  background-position: center;
189
- filter: saturate(0.95) contrast(0.98);
190
  }}
191
  .wash {{
192
  position: absolute;
193
  inset: 0;
194
- background: linear-gradient(180deg, rgba(251,250,247,0.72), rgba(251,250,247,0.9));
195
  }}
196
  .content {{
197
  position: relative;
@@ -207,7 +207,7 @@ def build_pipeline_html(summary: dict, base_path: Path) -> str:
207
  }}
208
  .kicker {{
209
  font: 700 17px "SF Mono", Menlo, monospace;
210
- color: #68665f;
211
  text-transform: uppercase;
212
  letter-spacing: 0.09em;
213
  margin-bottom: 14px;
@@ -221,7 +221,7 @@ def build_pipeline_html(summary: dict, base_path: Path) -> str:
221
  .subtitle {{
222
  margin: 18px 0 0;
223
  max-width: 1010px;
224
- color: #4d524d;
225
  font-size: 24px;
226
  line-height: 1.42;
227
  font-weight: 520;
@@ -233,23 +233,23 @@ def build_pipeline_html(summary: dict, base_path: Path) -> str:
233
  margin-top: 2px;
234
  }}
235
  .metric {{
236
- background: rgba(255,254,253,0.86);
237
- border: 1px solid #e4ded4;
238
  border-radius: 8px;
239
  padding: 13px 15px 12px;
240
- box-shadow: 0 16px 40px rgba(68,55,38,0.06);
241
  }}
242
  .metric strong {{
243
  display: block;
244
  font: 850 24px "SF Mono", Menlo, monospace;
245
- color: #1f2421;
246
  line-height: 1;
247
  font-variant-numeric: tabular-nums;
248
  }}
249
  .metric span {{
250
  display: block;
251
  margin-top: 7px;
252
- color: #696d67;
253
  font-size: 14px;
254
  font-weight: 650;
255
  }}
@@ -270,11 +270,11 @@ def build_pipeline_html(summary: dict, base_path: Path) -> str:
270
  flex: 1 1 0;
271
  height: 182px;
272
  position: relative;
273
- background: rgba(255,254,253,0.88);
274
- border: 1px solid rgba(228,222,212,0.96);
275
  border-radius: 8px;
276
  padding: 24px 24px 22px 30px;
277
- box-shadow: 0 24px 62px rgba(68,55,38,0.09);
278
  backdrop-filter: blur(12px);
279
  }}
280
  .stage::before {{
@@ -301,7 +301,7 @@ def build_pipeline_html(summary: dict, base_path: Path) -> str:
301
  margin: 0;
302
  padding: 0;
303
  list-style: none;
304
- color: #39413d;
305
  font-size: 17px;
306
  line-height: 1.48;
307
  font-weight: 560;
@@ -313,11 +313,11 @@ def build_pipeline_html(summary: dict, base_path: Path) -> str:
313
  display: grid;
314
  place-items: center;
315
  border-radius: 999px;
316
- border: 1px solid #d7d0c4;
317
- background: rgba(255,254,253,0.78);
318
- color: #7c807a;
319
  font: 850 22px "SF Mono", Menlo, monospace;
320
- box-shadow: 0 14px 34px rgba(68,55,38,0.06);
321
  }}
322
  .audit {{
323
  position: absolute;
@@ -328,14 +328,14 @@ def build_pipeline_html(summary: dict, base_path: Path) -> str:
328
  grid-template-columns: 190px 1fr;
329
  gap: 26px;
330
  align-items: center;
331
- background: rgba(255,254,253,0.88);
332
- border: 1px solid #e4ded4;
333
  border-radius: 8px;
334
  padding: 24px 28px;
335
- box-shadow: 0 22px 52px rgba(68,55,38,0.08);
336
  }}
337
  .audit strong {{
338
- color: #1f2421;
339
  font-size: 23px;
340
  line-height: 1.1;
341
  }}
@@ -343,7 +343,7 @@ def build_pipeline_html(summary: dict, base_path: Path) -> str:
343
  margin: 0;
344
  padding: 0;
345
  list-style: none;
346
- color: #40463f;
347
  font-size: 17px;
348
  line-height: 1.55;
349
  font-weight: 560;
@@ -445,30 +445,30 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
445
  <meta charset="utf-8">
446
  <style>
447
  * {{ box-sizing: border-box; }}
448
- body {{ margin: 0; background: #fbfaf7; font-family: "Avenir Next", "SF Pro Display", Arial, sans-serif; }}
449
  .canvas {{
450
  position: relative;
451
  width: {ARCHITECTURE_WIDTH}px;
452
  height: {ARCHITECTURE_HEIGHT}px;
453
  overflow: hidden;
454
- color: #1f2421;
455
  background:
456
- linear-gradient(90deg, rgba(68,55,38,0.03) 1px, transparent 1px),
457
- linear-gradient(0deg, rgba(68,55,38,0.024) 1px, transparent 1px),
458
- #fbfaf7;
459
- background-size: 58px 58px, 58px 58px, auto;
460
  }}
461
  .base-layer {{
462
  position: absolute;
463
  inset: 0;
464
  background-size: cover;
465
  background-position: center;
466
- filter: saturate(0.95) contrast(0.98);
467
  }}
468
  .wash {{
469
  position: absolute;
470
  inset: 0;
471
- background: linear-gradient(180deg, rgba(251,250,247,0.72), rgba(251,250,247,0.92));
472
  }}
473
  .content {{
474
  position: relative;
@@ -484,7 +484,7 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
484
  }}
485
  .kicker {{
486
  font: 700 16px "SF Mono", Menlo, monospace;
487
- color: #68665f;
488
  text-transform: uppercase;
489
  letter-spacing: 0.09em;
490
  margin-bottom: 13px;
@@ -498,7 +498,7 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
498
  .subtitle {{
499
  margin: 15px 0 0;
500
  max-width: 1060px;
501
- color: #4d524d;
502
  font-size: 22px;
503
  line-height: 1.42;
504
  font-weight: 520;
@@ -508,10 +508,10 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
508
  place-items: center;
509
  min-width: 188px;
510
  min-height: 112px;
511
- border: 1px solid #e4ded4;
512
  border-radius: 8px;
513
- background: rgba(255,254,253,0.86);
514
- box-shadow: 0 18px 44px rgba(68,55,38,0.07);
515
  text-align: center;
516
  }}
517
  .summary-pill strong {{
@@ -521,7 +521,7 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
521
  .summary-pill span {{
522
  display: block;
523
  margin-top: 8px;
524
- color: #6c6d68;
525
  font-size: 15px;
526
  font-weight: 700;
527
  }}
@@ -533,11 +533,11 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
533
  }}
534
  .shared article {{
535
  min-height: 110px;
536
- border: 1px solid #e4ded4;
537
  border-radius: 8px;
538
- background: rgba(255,254,253,0.88);
539
  padding: 20px 22px;
540
- box-shadow: 0 18px 44px rgba(68,55,38,0.06);
541
  }}
542
  .shared h2 {{
543
  margin: 0 0 9px;
@@ -546,7 +546,7 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
546
  }}
547
  .shared p {{
548
  margin: 0;
549
- color: #4c534f;
550
  font-size: 16px;
551
  line-height: 1.38;
552
  font-weight: 560;
@@ -559,11 +559,11 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
559
  }}
560
  .family {{
561
  min-height: 124px;
562
- border: 1px solid #e4ded4;
563
  border-radius: 8px;
564
- background: rgba(255,254,253,0.82);
565
  padding: 20px 20px 18px;
566
- box-shadow: 0 16px 40px rgba(68,55,38,0.055);
567
  }}
568
  .family h3 {{
569
  margin: 0 0 10px;
@@ -573,7 +573,7 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
573
  }}
574
  .family p {{
575
  margin: 0;
576
- color: #4d534f;
577
  font-size: 15px;
578
  line-height: 1.42;
579
  font-weight: 560;
@@ -584,11 +584,11 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
584
  gap: 20px;
585
  }}
586
  .task-group {{
587
- border: 1px solid rgba(228,222,212,0.96);
588
  border-radius: 8px;
589
- background: rgba(255,254,253,0.74);
590
  padding: 18px;
591
- box-shadow: 0 22px 54px rgba(68,55,38,0.07);
592
  backdrop-filter: blur(10px);
593
  }}
594
  .group-head {{
@@ -616,9 +616,9 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
616
  .task-card {{
617
  min-height: 244px;
618
  position: relative;
619
- border: 1px solid color-mix(in srgb, var(--accent), #ffffff 68%);
620
  border-radius: 8px;
621
- background: rgba(255,254,253,0.92);
622
  padding: 17px 18px 16px;
623
  overflow: hidden;
624
  }}
@@ -639,11 +639,11 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
639
  font: 850 11px "SF Mono", Menlo, monospace;
640
  text-transform: uppercase;
641
  letter-spacing: 0.03em;
642
- background: rgba(255,254,253,0.72);
643
  }}
644
  .task-card h3 {{
645
  margin: 13px 0 12px;
646
- color: #161a17;
647
  font-size: 21px;
648
  line-height: 1.08;
649
  overflow-wrap: anywhere;
@@ -653,7 +653,7 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
653
  grid-template-columns: 54px 1fr;
654
  gap: 5px 9px;
655
  margin: 0;
656
- color: #3d453f;
657
  font-size: 13px;
658
  line-height: 1.32;
659
  font-weight: 560;
@@ -671,13 +671,13 @@ def build_architecture_html(summary: dict, base_path: Path) -> str:
671
  gap: 12px;
672
  align-items: center;
673
  margin-top: 12px;
674
- border-top: 1px solid #eee9e1;
675
  padding-top: 12px;
676
  font-size: 13px;
677
  font-weight: 700;
678
  }}
679
  .metric-line span {{
680
- color: #6c6d68;
681
  font: 850 11px "SF Mono", Menlo, monospace;
682
  text-transform: uppercase;
683
  }}
 
28
  DEFAULT_ARCHITECTURE_OUTPUT = ASSETS / "task_architectures.png"
29
 
30
  PIPELINE_WIDTH = 1800
31
+ PIPELINE_HEIGHT = 1120
32
  ARCHITECTURE_WIDTH = 1800
33
  ARCHITECTURE_HEIGHT = 1520
34
 
35
 
36
  COLORS = {
37
+ "blue": "#9bdfff",
38
+ "teal": "#7ae5c3",
39
+ "green": "#a7f078",
40
+ "amber": "#d8f4a5",
41
+ "orange": "#b7ff91",
42
+ "red": "#ff8f7a",
43
+ "ink": "#f4f8ef",
44
+ "muted": "#a5afa2",
45
+ "line": "#2b4428",
46
  }
47
 
48
 
49
  TASK_GROUPS = [
50
+ ("Label + State", "#9bdfff", ["timeline_action", "timeline_subtask", "next_action"]),
51
  (
52
  "Prediction + Reconstruction",
53
+ "#a7f078",
54
  ["hand_trajectory_forecast", "modality_reconstruction", "contact_prediction"],
55
  ),
56
+ ("Grounding + Retrieval", "#7ae5c3", ["caption_grounding", "cross_modal_retrieval", "object_relevance"]),
57
+ ("Temporal Diagnostics", "#d8f4a5", ["transition_detection", "temporal_order", "misalignment_detection"]),
58
  ]
59
 
60
 
 
168
  <meta charset="utf-8">
169
  <style>
170
  * {{ box-sizing: border-box; }}
171
+ body {{ margin: 0; background: #020502; font-family: "Inter Tight", "Space Grotesk", Arial, sans-serif; }}
172
  .canvas {{
173
  position: relative;
174
  width: {PIPELINE_WIDTH}px;
175
  height: {PIPELINE_HEIGHT}px;
176
  overflow: hidden;
177
+ color: #f4f8ef;
178
  background:
179
+ radial-gradient(circle at 78% 24%, rgba(167,240,120,0.18), transparent 24%),
180
+ radial-gradient(circle, rgba(167,240,120,0.16) 1px, transparent 2px),
181
+ #020502;
182
+ background-size: auto, 18px 18px, auto;
183
  }}
184
  .base-layer {{
185
  position: absolute;
186
  inset: 0;
187
  background-size: cover;
188
  background-position: center;
189
+ filter: saturate(1.08) contrast(1.05) brightness(0.48);
190
  }}
191
  .wash {{
192
  position: absolute;
193
  inset: 0;
194
+ background: linear-gradient(180deg, rgba(2,5,2,0.76), rgba(2,5,2,0.94));
195
  }}
196
  .content {{
197
  position: relative;
 
207
  }}
208
  .kicker {{
209
  font: 700 17px "SF Mono", Menlo, monospace;
210
+ color: #a7f078;
211
  text-transform: uppercase;
212
  letter-spacing: 0.09em;
213
  margin-bottom: 14px;
 
221
  .subtitle {{
222
  margin: 18px 0 0;
223
  max-width: 1010px;
224
+ color: #dce8d7;
225
  font-size: 24px;
226
  line-height: 1.42;
227
  font-weight: 520;
 
233
  margin-top: 2px;
234
  }}
235
  .metric {{
236
+ background: rgba(7,18,7,0.86);
237
+ border: 1px solid rgba(167,240,120,0.26);
238
  border-radius: 8px;
239
  padding: 13px 15px 12px;
240
+ box-shadow: 0 16px 44px rgba(0,0,0,0.42);
241
  }}
242
  .metric strong {{
243
  display: block;
244
  font: 850 24px "SF Mono", Menlo, monospace;
245
+ color: #f4f8ef;
246
  line-height: 1;
247
  font-variant-numeric: tabular-nums;
248
  }}
249
  .metric span {{
250
  display: block;
251
  margin-top: 7px;
252
+ color: #a5afa2;
253
  font-size: 14px;
254
  font-weight: 650;
255
  }}
 
270
  flex: 1 1 0;
271
  height: 182px;
272
  position: relative;
273
+ background: rgba(7,18,7,0.86);
274
+ border: 1px solid rgba(167,240,120,0.24);
275
  border-radius: 8px;
276
  padding: 24px 24px 22px 30px;
277
+ box-shadow: 0 24px 62px rgba(0,0,0,0.40);
278
  backdrop-filter: blur(12px);
279
  }}
280
  .stage::before {{
 
301
  margin: 0;
302
  padding: 0;
303
  list-style: none;
304
+ color: #dce8d7;
305
  font-size: 17px;
306
  line-height: 1.48;
307
  font-weight: 560;
 
313
  display: grid;
314
  place-items: center;
315
  border-radius: 999px;
316
+ border: 1px solid rgba(167,240,120,0.26);
317
+ background: rgba(7,18,7,0.78);
318
+ color: #a7f078;
319
  font: 850 22px "SF Mono", Menlo, monospace;
320
+ box-shadow: 0 14px 34px rgba(0,0,0,0.36);
321
  }}
322
  .audit {{
323
  position: absolute;
 
328
  grid-template-columns: 190px 1fr;
329
  gap: 26px;
330
  align-items: center;
331
+ background: rgba(7,18,7,0.88);
332
+ border: 1px solid rgba(167,240,120,0.24);
333
  border-radius: 8px;
334
  padding: 24px 28px;
335
+ box-shadow: 0 22px 52px rgba(0,0,0,0.42);
336
  }}
337
  .audit strong {{
338
+ color: #f4f8ef;
339
  font-size: 23px;
340
  line-height: 1.1;
341
  }}
 
343
  margin: 0;
344
  padding: 0;
345
  list-style: none;
346
+ color: #dce8d7;
347
  font-size: 17px;
348
  line-height: 1.55;
349
  font-weight: 560;
 
445
  <meta charset="utf-8">
446
  <style>
447
  * {{ box-sizing: border-box; }}
448
+ body {{ margin: 0; background: #020502; font-family: "Inter Tight", "Space Grotesk", Arial, sans-serif; }}
449
  .canvas {{
450
  position: relative;
451
  width: {ARCHITECTURE_WIDTH}px;
452
  height: {ARCHITECTURE_HEIGHT}px;
453
  overflow: hidden;
454
+ color: #f4f8ef;
455
  background:
456
+ radial-gradient(circle at 76% 18%, rgba(167,240,120,0.16), transparent 24%),
457
+ radial-gradient(circle, rgba(167,240,120,0.13) 1px, transparent 2px),
458
+ #020502;
459
+ background-size: auto, 18px 18px, auto;
460
  }}
461
  .base-layer {{
462
  position: absolute;
463
  inset: 0;
464
  background-size: cover;
465
  background-position: center;
466
+ filter: saturate(1.08) contrast(1.05) brightness(0.48);
467
  }}
468
  .wash {{
469
  position: absolute;
470
  inset: 0;
471
+ background: linear-gradient(180deg, rgba(2,5,2,0.76), rgba(2,5,2,0.94));
472
  }}
473
  .content {{
474
  position: relative;
 
484
  }}
485
  .kicker {{
486
  font: 700 16px "SF Mono", Menlo, monospace;
487
+ color: #a7f078;
488
  text-transform: uppercase;
489
  letter-spacing: 0.09em;
490
  margin-bottom: 13px;
 
498
  .subtitle {{
499
  margin: 15px 0 0;
500
  max-width: 1060px;
501
+ color: #dce8d7;
502
  font-size: 22px;
503
  line-height: 1.42;
504
  font-weight: 520;
 
508
  place-items: center;
509
  min-width: 188px;
510
  min-height: 112px;
511
+ border: 1px solid rgba(167,240,120,0.26);
512
  border-radius: 8px;
513
+ background: rgba(7,18,7,0.86);
514
+ box-shadow: 0 18px 44px rgba(0,0,0,0.42);
515
  text-align: center;
516
  }}
517
  .summary-pill strong {{
 
521
  .summary-pill span {{
522
  display: block;
523
  margin-top: 8px;
524
+ color: #a5afa2;
525
  font-size: 15px;
526
  font-weight: 700;
527
  }}
 
533
  }}
534
  .shared article {{
535
  min-height: 110px;
536
+ border: 1px solid rgba(167,240,120,0.24);
537
  border-radius: 8px;
538
+ background: rgba(7,18,7,0.86);
539
  padding: 20px 22px;
540
+ box-shadow: 0 18px 44px rgba(0,0,0,0.36);
541
  }}
542
  .shared h2 {{
543
  margin: 0 0 9px;
 
546
  }}
547
  .shared p {{
548
  margin: 0;
549
+ color: #dce8d7;
550
  font-size: 16px;
551
  line-height: 1.38;
552
  font-weight: 560;
 
559
  }}
560
  .family {{
561
  min-height: 124px;
562
+ border: 1px solid rgba(167,240,120,0.24);
563
  border-radius: 8px;
564
+ background: rgba(7,18,7,0.82);
565
  padding: 20px 20px 18px;
566
+ box-shadow: 0 16px 40px rgba(0,0,0,0.34);
567
  }}
568
  .family h3 {{
569
  margin: 0 0 10px;
 
573
  }}
574
  .family p {{
575
  margin: 0;
576
+ color: #dce8d7;
577
  font-size: 15px;
578
  line-height: 1.42;
579
  font-weight: 560;
 
584
  gap: 20px;
585
  }}
586
  .task-group {{
587
+ border: 1px solid rgba(167,240,120,0.22);
588
  border-radius: 8px;
589
+ background: rgba(7,18,7,0.74);
590
  padding: 18px;
591
+ box-shadow: 0 22px 54px rgba(0,0,0,0.42);
592
  backdrop-filter: blur(10px);
593
  }}
594
  .group-head {{
 
616
  .task-card {{
617
  min-height: 244px;
618
  position: relative;
619
+ border: 1px solid color-mix(in srgb, var(--accent), #020502 66%);
620
  border-radius: 8px;
621
+ background: rgba(7,18,7,0.92);
622
  padding: 17px 18px 16px;
623
  overflow: hidden;
624
  }}
 
639
  font: 850 11px "SF Mono", Menlo, monospace;
640
  text-transform: uppercase;
641
  letter-spacing: 0.03em;
642
+ background: rgba(7,18,7,0.72);
643
  }}
644
  .task-card h3 {{
645
  margin: 13px 0 12px;
646
+ color: #f4f8ef;
647
  font-size: 21px;
648
  line-height: 1.08;
649
  overflow-wrap: anywhere;
 
653
  grid-template-columns: 54px 1fr;
654
  gap: 5px 9px;
655
  margin: 0;
656
+ color: #dce8d7;
657
  font-size: 13px;
658
  line-height: 1.32;
659
  font-weight: 560;
 
671
  gap: 12px;
672
  align-items: center;
673
  margin-top: 12px;
674
+ border-top: 1px solid rgba(167,240,120,0.16);
675
  padding-top: 12px;
676
  font-size: 13px;
677
  font-weight: 700;
678
  }}
679
  .metric-line span {{
680
+ color: #a5afa2;
681
  font: 850 11px "SF Mono", Menlo, monospace;
682
  text-transform: uppercase;
683
  }}
scripts/render_task_suite_infographic.py CHANGED
@@ -34,8 +34,8 @@ GROUPS = [
34
  {
35
  "name": "Label + State",
36
  "tone": "teal",
37
- "color": "#197d83",
38
- "soft": "#e8f4f3",
39
  "tasks": [
40
  ("timeline_action", "supervised"),
41
  ("timeline_subtask", "supervised"),
@@ -45,8 +45,8 @@ GROUPS = [
45
  {
46
  "name": "Prediction + Reconstruction",
47
  "tone": "blue",
48
- "color": "#1f6c9f",
49
- "soft": "#e8f1fb",
50
  "tasks": [
51
  ("hand_trajectory_forecast", "forecast"),
52
  ("modality_reconstruction", "forecast"),
@@ -56,8 +56,8 @@ GROUPS = [
56
  {
57
  "name": "Grounding + Retrieval",
58
  "tone": "amber",
59
- "color": "#9b6516",
60
- "soft": "#fbf3df",
61
  "tasks": [
62
  ("caption_grounding", "retrieval"),
63
  ("cross_modal_retrieval", "retrieval"),
@@ -67,8 +67,8 @@ GROUPS = [
67
  {
68
  "name": "Temporal Diagnostics",
69
  "tone": "red",
70
- "color": "#b0443e",
71
- "soft": "#fdeceb",
72
  "tasks": [
73
  ("transition_detection", "diagnostic"),
74
  ("temporal_order", "diagnostic"),
@@ -107,7 +107,7 @@ def image_data_uri(image, fmt: str = "PNG", quality: int = 92) -> str:
107
  return f"data:image/{mime};base64,{encoded}"
108
 
109
 
110
- def make_canvas(size=(THUMB_WIDTH, THUMB_HEIGHT), color=(255, 254, 253)):
111
  from PIL import Image
112
 
113
  return Image.new("RGB", size, color)
@@ -138,7 +138,7 @@ def read_video_frame(video_path: Path, frame_index: int = 2400):
138
  return Image.fromarray(frame)
139
 
140
 
141
- def draw_label(draw, xy, text, fill=(31, 36, 33), size=18):
142
  from PIL import ImageFont
143
 
144
  try:
@@ -158,7 +158,7 @@ def video_thumb(sample_dir: Path) -> str:
158
  canvas.paste(fish, (0, 0))
159
  canvas.paste(stereo, (226, 0))
160
  draw = ImageDraw.Draw(canvas, "RGBA")
161
- draw.rounded_rectangle((188, 0, 232, THUMB_HEIGHT), radius=0, fill=(251, 250, 247, 235))
162
  draw_label(draw, (194, 16), "fisheye", fill=(255, 255, 255), size=14)
163
  draw_label(draw, (240, 16), "stereo", fill=(255, 255, 255), size=14)
164
  return image_data_uri(canvas, "JPEG")
@@ -168,11 +168,11 @@ def colorize(values):
168
  import numpy as np
169
 
170
  stops = np.array([
171
- [26, 35, 126],
172
- [36, 123, 160],
173
- [68, 170, 122],
174
- [238, 190, 76],
175
- [197, 79, 51],
176
  ], dtype=np.float32)
177
  x = np.clip(values, 0, 1)
178
  scaled = x * (len(stops) - 1)
@@ -200,8 +200,8 @@ def depth_thumb(h5) -> str:
200
  canvas.paste(depth, (0, 0))
201
  canvas.paste(conf_img, (216, 0))
202
  draw = ImageDraw.Draw(canvas, "RGBA")
203
- draw.rounded_rectangle((0, 0, 116, 28), radius=6, fill=(31, 36, 33, 150))
204
- draw.rounded_rectangle((216, 0, 350, 28), radius=6, fill=(31, 36, 33, 150))
205
  draw_label(draw, (10, 6), "depth", fill=(255, 255, 255), size=14)
206
  draw_label(draw, (226, 6), "confidence", fill=(255, 255, 255), size=14)
207
  return image_data_uri(canvas, "JPEG")
@@ -248,19 +248,19 @@ def audio_thumb(sample_dir: Path) -> str:
248
  for i, value in enumerate(rms):
249
  x = 18 + i / max(bins - 1, 1) * (THUMB_WIDTH - 36)
250
  h = 8 + np.clip(value * 86, 0, 86)
251
- draw.line((x, 126, x, 126 - h), fill=(31, 108, 159, 150), width=2)
252
  points = []
253
  for i, value in enumerate(waveform):
254
  x = 18 + i / max(bins - 1, 1) * (THUMB_WIDTH - 36)
255
  y = 74 - np.clip(value, -1, 1) * 42
256
  points.append((x, y))
257
- draw.line(points, fill=(155, 101, 22, 210), width=2)
258
  except Exception:
259
  for i in range(48):
260
  x = 22 + i * 8
261
  h = 16 + (i % 7) * 7
262
- draw.rounded_rectangle((x, 128 - h, x + 4, 128), radius=2, fill=(31, 108, 159, 150))
263
- draw_label(draw, (16, 12), "AAC audio waveform", fill=(31, 36, 33), size=17)
264
  return image_data_uri(canvas, "PNG")
265
 
266
 
@@ -299,8 +299,8 @@ def slam_thumb(h5) -> str:
299
  traj = np.array(h5["slam/trans_xyz"][:2450:36], dtype=np.float64)
300
  traj_xy = normalize_points(traj[:, [0, 2, 1]], THUMB_WIDTH, THUMB_HEIGHT)
301
  for a, b in zip(traj_xy[:-1], traj_xy[1:]):
302
- draw.line((a[0], a[1], b[0], b[1]), fill=(31, 108, 159, 190), width=2)
303
- draw_label(draw, (16, 14), "camera pose + SLAM map", fill=(31, 36, 33), size=17)
304
  return image_data_uri(canvas, "PNG")
305
 
306
 
@@ -314,10 +314,10 @@ def imu_thumb(h5) -> str:
314
  accel = np.array(h5["imu/accel_xyz"][max(0, key_idx - 220): key_idx + 220], dtype=np.float64)
315
  gyro = np.array(h5["imu/gyro_xyz"][max(0, key_idx - 220): key_idx + 220], dtype=np.float64)
316
  series = [accel[:, 0], accel[:, 1], accel[:, 2], gyro[:, 0], gyro[:, 1], gyro[:, 2]]
317
- colors = [(31, 108, 159), (52, 101, 56), (176, 68, 62), (155, 101, 22), (46, 119, 117), (96, 109, 128)]
318
  for row in range(4):
319
  y = 26 + row * 33
320
- draw.line((18, y, THUMB_WIDTH - 18, y), fill=(228, 222, 212, 180), width=1)
321
  for values, color in zip(series, colors):
322
  values = values[:420]
323
  if len(values) < 2:
@@ -330,7 +330,7 @@ def imu_thumb(h5) -> str:
330
  y = 138 - np.clip(v, 0, 1) * 112
331
  pts.append((x, y))
332
  draw.line(pts, fill=color + (200,), width=2)
333
- draw_label(draw, (16, 12), "inertial accel / gyro", fill=(31, 36, 33), size=17)
334
  return image_data_uri(canvas, "PNG")
335
 
336
 
@@ -357,17 +357,17 @@ def mocap_thumb(h5) -> str:
357
 
358
  body_xy = project(body, 18, 165)
359
  for x, y in body_xy:
360
- draw.ellipse((x - 2.4, y - 2.4, x + 2.4, y + 2.4), fill=(52, 101, 56, 175))
361
  for a, b in zip(body_xy[:-1], body_xy[1:]):
362
- draw.line((a[0], a[1], b[0], b[1]), fill=(52, 101, 56, 70), width=1)
363
 
364
- for points, x_offset, color in [(left, 218, (31, 108, 159)), (right, 314, (155, 101, 22))]:
365
  xy = project(points, x_offset, 82)
366
  for a, b in HAND_EDGES:
367
  draw.line((xy[a][0], xy[a][1], xy[b][0], xy[b][1]), fill=color + (180,), width=2)
368
  for x, y in xy:
369
  draw.ellipse((x - 2.4, y - 2.4, x + 2.4, y + 2.4), fill=color + (220,))
370
- draw_label(draw, (16, 12), "body + hand mocap", fill=(31, 36, 33), size=17)
371
  return image_data_uri(canvas, "PNG")
372
 
373
 
@@ -383,18 +383,18 @@ def text_thumb(h5) -> str:
383
  actions = [a.get("label", "") for a in segment.get("Current Action", [])][:2]
384
  canvas = make_canvas()
385
  draw = ImageDraw.Draw(canvas, "RGBA")
386
- draw_label(draw, (16, 13), "language annotation", fill=(31, 36, 33), size=17)
387
  y = 46
388
  for label in objects:
389
- draw.rounded_rectangle((16, y, 16 + 20 + len(label) * 8, y + 24), radius=6, fill=(251, 243, 219, 230), outline=(226, 200, 144, 255))
390
- draw_label(draw, (26, y + 5), label, fill=(83, 74, 56), size=12)
391
  y += 30
392
  x = 184
393
  y = 48
394
  for action in actions:
395
  wrapped = action[:32] + ("..." if len(action) > 32 else "")
396
- draw.rounded_rectangle((x, y, THUMB_WIDTH - 16, y + 36), radius=7, fill=(232, 244, 243, 230), outline=(169, 204, 202, 255))
397
- draw_label(draw, (x + 10, y + 10), wrapped, fill=(31, 36, 33), size=12)
398
  y += 44
399
  return image_data_uri(canvas, "PNG")
400
 
@@ -569,11 +569,11 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
569
  margin: 0;
570
  width: {CANVAS_WIDTH}px;
571
  height: {CANVAS_HEIGHT}px;
572
- background: #fbfaf7;
573
  }}
574
  body {{
575
- font-family: "Avenir Next", "SF Pro Display", ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif;
576
- color: #1f2421;
577
  text-rendering: optimizeLegibility;
578
  }}
579
  .canvas {{
@@ -583,12 +583,10 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
583
  overflow: hidden;
584
  padding: 54px 64px 44px;
585
  background:
586
- radial-gradient(circle at 9% 6%, rgba(31,108,159,0.13), transparent 20%),
587
- radial-gradient(circle at 90% 9%, rgba(155,101,22,0.10), transparent 22%),
588
- linear-gradient(90deg, rgba(68,55,38,0.035) 1px, transparent 1px),
589
- linear-gradient(0deg, rgba(68,55,38,0.027) 1px, transparent 1px),
590
- #fbfaf7;
591
- background-size: auto, auto, 54px 54px, 54px 54px, auto;
592
  }}
593
  .image-background {{
594
  position: absolute;
@@ -596,8 +594,8 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
596
  background-position: center;
597
  background-repeat: no-repeat;
598
  background-size: cover;
599
- opacity: 0.30;
600
- filter: saturate(0.85) contrast(0.98);
601
  }}
602
  .content {{
603
  position: relative;
@@ -609,13 +607,13 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
609
  gap: 44px;
610
  align-items: end;
611
  padding-bottom: 30px;
612
- border-bottom: 1px solid #e4ded4;
613
  }}
614
  .kicker {{
615
  display: inline-flex;
616
  align-items: center;
617
  gap: 12px;
618
- color: #5f625d;
619
  font-family: "SF Mono", "JetBrains Mono", ui-monospace, monospace;
620
  font-size: 15px;
621
  text-transform: uppercase;
@@ -625,7 +623,7 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
625
  content: "";
626
  width: 44px;
627
  height: 1px;
628
- background: #1f2421;
629
  }}
630
  h1 {{
631
  margin: 18px 0 0;
@@ -637,7 +635,7 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
637
  .subtitle {{
638
  margin: 18px 0 0;
639
  max-width: 900px;
640
- color: #5f625d;
641
  font-size: 23px;
642
  line-height: 1.35;
643
  font-weight: 520;
@@ -650,9 +648,9 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
650
  .stat {{
651
  min-height: 78px;
652
  padding: 14px 15px;
653
- border: 1px solid #e4ded4;
654
- background: rgba(255,254,253,0.76);
655
- border-radius: 10px;
656
  }}
657
  .stat strong {{
658
  display: block;
@@ -664,7 +662,7 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
664
  .stat span {{
665
  display: block;
666
  margin-top: 8px;
667
- color: #6f716c;
668
  font-size: 13px;
669
  line-height: 1.15;
670
  }}
@@ -673,14 +671,14 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
673
  align-items: center;
674
  justify-content: space-between;
675
  margin: 28px 0 14px;
676
- color: #5f625d;
677
  font-family: "SF Mono", "JetBrains Mono", ui-monospace, monospace;
678
  font-size: 14px;
679
  text-transform: uppercase;
680
  letter-spacing: 0.08em;
681
  }}
682
  .section-label span:last-child {{
683
- color: #7e817b;
684
  text-transform: none;
685
  letter-spacing: 0;
686
  font-family: inherit;
@@ -693,16 +691,16 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
693
  .modality {{
694
  min-height: 204px;
695
  padding: 11px 12px 14px;
696
- border: 1px solid #e4ded4;
697
- background: rgba(255,254,253,0.84);
698
- border-radius: 12px;
699
  }}
700
  .modality-thumb {{
701
  height: 86px;
702
  overflow: hidden;
703
- border: 1px solid #eee9e1;
704
- border-radius: 9px;
705
- background: #f5f1e9;
706
  }}
707
  .modality-thumb img {{
708
  display: block;
@@ -716,7 +714,7 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
716
  font-variant-numeric: tabular-nums;
717
  }}
718
  .modality-index {{
719
- color: #8a8072;
720
  font-size: 12px;
721
  margin-top: 10px;
722
  }}
@@ -728,14 +726,14 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
728
  }}
729
  .modality p {{
730
  margin: 9px 0 0;
731
- color: #4f565f;
732
  font-size: 14px;
733
  font-weight: 650;
734
  }}
735
  .modality span {{
736
  display: block;
737
  margin-top: 5px;
738
- color: #7a7d77;
739
  font-size: 13px;
740
  }}
741
  .shared-band {{
@@ -745,16 +743,16 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
745
  align-items: center;
746
  margin-top: 20px;
747
  padding: 14px;
748
- border: 1px solid #e4ded4;
749
- background: rgba(245,241,233,0.82);
750
- border-radius: 12px;
751
  }}
752
  .step {{
753
  min-height: 62px;
754
  padding: 13px 15px;
755
- background: #fffefd;
756
- border: 1px solid #eee9e1;
757
- border-radius: 9px;
758
  }}
759
  .step strong {{
760
  display: block;
@@ -764,11 +762,11 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
764
  .step span {{
765
  display: block;
766
  margin-top: 5px;
767
- color: #6f716c;
768
  font-size: 13px;
769
  }}
770
  .arrow {{
771
- color: #938a7d;
772
  font-family: "SF Mono", "JetBrains Mono", ui-monospace, monospace;
773
  font-size: 22px;
774
  }}
@@ -780,9 +778,9 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
780
  }}
781
  .family {{
782
  padding: 17px;
783
- border: 1px solid color-mix(in srgb, var(--accent) 24%, #e4ded4);
784
- background: rgba(255,254,253,0.82);
785
- border-radius: 16px;
786
  }}
787
  .family-head {{
788
  display: flex;
@@ -791,7 +789,7 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
791
  gap: 16px;
792
  min-height: 78px;
793
  padding-bottom: 14px;
794
- border-bottom: 1px solid color-mix(in srgb, var(--accent) 18%, #eee9e1);
795
  }}
796
  .family-head span {{
797
  color: var(--accent);
@@ -815,9 +813,9 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
815
  .task-card {{
816
  min-height: 168px;
817
  padding: 17px 18px;
818
- border: 1px solid color-mix(in srgb, var(--accent) 22%, #e4ded4);
819
- background: linear-gradient(180deg, #fffefd, color-mix(in srgb, var(--soft) 45%, #fffefd));
820
- border-radius: 13px;
821
  }}
822
  .task-meta {{
823
  display: flex;
@@ -826,7 +824,7 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
826
  gap: 12px;
827
  }}
828
  .index {{
829
- color: #8a8072;
830
  font-size: 12px;
831
  }}
832
  .kind {{
@@ -835,9 +833,9 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
835
  height: 24px;
836
  padding: 0 9px;
837
  border-radius: 6px;
838
- border: 1px solid color-mix(in srgb, var(--accent) 30%, #ffffff);
839
  color: var(--accent);
840
- background: rgba(255,255,255,0.72);
841
  text-transform: uppercase;
842
  font-size: 11px;
843
  line-height: 1;
@@ -845,7 +843,7 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
845
  }}
846
  .task-card h3 {{
847
  margin: 12px 0 0;
848
- color: #111827;
849
  font-family: "SF Mono", "JetBrains Mono", ui-monospace, monospace;
850
  font-size: 21px;
851
  line-height: 1.18;
@@ -854,7 +852,7 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
854
  .task-card p {{
855
  margin: 11px 0 0;
856
  min-height: 39px;
857
- color: #4f565f;
858
  font-size: 15px;
859
  line-height: 1.28;
860
  font-weight: 560;
@@ -867,16 +865,16 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
867
  min-height: 32px;
868
  padding: 7px 10px;
869
  border-radius: 8px;
870
- border: 1px solid color-mix(in srgb, var(--accent) 32%, #ffffff);
871
- background: rgba(255,255,255,0.82);
872
  }}
873
  .metric.neural {{
874
  margin-left: 8px;
875
- border-color: rgba(31,36,33,0.18);
876
- background: rgba(245,241,233,0.82);
877
  }}
878
  .metric span {{
879
- color: #64748b;
880
  font-size: 13px;
881
  font-weight: 760;
882
  }}
@@ -895,17 +893,17 @@ def build_html(summary: dict, base_image: Path | None, sample_dir: Path | None)
895
  gap: 32px;
896
  margin-top: 22px;
897
  padding-top: 20px;
898
- border-top: 1px solid #e4ded4;
899
- color: #5f625d;
900
  font-size: 18px;
901
  line-height: 1.35;
902
  font-weight: 620;
903
  }}
904
  .footer code {{
905
  font-family: "SF Mono", "JetBrains Mono", ui-monospace, monospace;
906
- color: #1f2421;
907
- background: #f5f1e9;
908
- border: 1px solid #e4ded4;
909
  border-radius: 7px;
910
  padding: 6px 9px;
911
  white-space: nowrap;
 
34
  {
35
  "name": "Label + State",
36
  "tone": "teal",
37
+ "color": "#9bdfff",
38
+ "soft": "#071d20",
39
  "tasks": [
40
  ("timeline_action", "supervised"),
41
  ("timeline_subtask", "supervised"),
 
45
  {
46
  "name": "Prediction + Reconstruction",
47
  "tone": "blue",
48
+ "color": "#a7f078",
49
+ "soft": "#10210a",
50
  "tasks": [
51
  ("hand_trajectory_forecast", "forecast"),
52
  ("modality_reconstruction", "forecast"),
 
56
  {
57
  "name": "Grounding + Retrieval",
58
  "tone": "amber",
59
+ "color": "#7ae5c3",
60
+ "soft": "#092019",
61
  "tasks": [
62
  ("caption_grounding", "retrieval"),
63
  ("cross_modal_retrieval", "retrieval"),
 
67
  {
68
  "name": "Temporal Diagnostics",
69
  "tone": "red",
70
+ "color": "#d8f4a5",
71
+ "soft": "#1b210d",
72
  "tasks": [
73
  ("transition_detection", "diagnostic"),
74
  ("temporal_order", "diagnostic"),
 
107
  return f"data:image/{mime};base64,{encoded}"
108
 
109
 
110
+ def make_canvas(size=(THUMB_WIDTH, THUMB_HEIGHT), color=(2, 5, 2)):
111
  from PIL import Image
112
 
113
  return Image.new("RGB", size, color)
 
138
  return Image.fromarray(frame)
139
 
140
 
141
+ def draw_label(draw, xy, text, fill=(244, 248, 239), size=18):
142
  from PIL import ImageFont
143
 
144
  try:
 
158
  canvas.paste(fish, (0, 0))
159
  canvas.paste(stereo, (226, 0))
160
  draw = ImageDraw.Draw(canvas, "RGBA")
161
+ draw.rounded_rectangle((188, 0, 232, THUMB_HEIGHT), radius=0, fill=(2, 5, 2, 220))
162
  draw_label(draw, (194, 16), "fisheye", fill=(255, 255, 255), size=14)
163
  draw_label(draw, (240, 16), "stereo", fill=(255, 255, 255), size=14)
164
  return image_data_uri(canvas, "JPEG")
 
168
  import numpy as np
169
 
170
  stops = np.array([
171
+ [2, 5, 2],
172
+ [58, 136, 102],
173
+ [122, 229, 195],
174
+ [167, 240, 120],
175
+ [216, 244, 165],
176
  ], dtype=np.float32)
177
  x = np.clip(values, 0, 1)
178
  scaled = x * (len(stops) - 1)
 
200
  canvas.paste(depth, (0, 0))
201
  canvas.paste(conf_img, (216, 0))
202
  draw = ImageDraw.Draw(canvas, "RGBA")
203
+ draw.rounded_rectangle((0, 0, 116, 28), radius=6, fill=(2, 5, 2, 178))
204
+ draw.rounded_rectangle((216, 0, 350, 28), radius=6, fill=(2, 5, 2, 178))
205
  draw_label(draw, (10, 6), "depth", fill=(255, 255, 255), size=14)
206
  draw_label(draw, (226, 6), "confidence", fill=(255, 255, 255), size=14)
207
  return image_data_uri(canvas, "JPEG")
 
248
  for i, value in enumerate(rms):
249
  x = 18 + i / max(bins - 1, 1) * (THUMB_WIDTH - 36)
250
  h = 8 + np.clip(value * 86, 0, 86)
251
+ draw.line((x, 126, x, 126 - h), fill=(167, 240, 120, 170), width=2)
252
  points = []
253
  for i, value in enumerate(waveform):
254
  x = 18 + i / max(bins - 1, 1) * (THUMB_WIDTH - 36)
255
  y = 74 - np.clip(value, -1, 1) * 42
256
  points.append((x, y))
257
+ draw.line(points, fill=(122, 229, 195, 220), width=2)
258
  except Exception:
259
  for i in range(48):
260
  x = 22 + i * 8
261
  h = 16 + (i % 7) * 7
262
+ draw.rounded_rectangle((x, 128 - h, x + 4, 128), radius=2, fill=(167, 240, 120, 170))
263
+ draw_label(draw, (16, 12), "AAC audio waveform", fill=(244, 248, 239), size=17)
264
  return image_data_uri(canvas, "PNG")
265
 
266
 
 
299
  traj = np.array(h5["slam/trans_xyz"][:2450:36], dtype=np.float64)
300
  traj_xy = normalize_points(traj[:, [0, 2, 1]], THUMB_WIDTH, THUMB_HEIGHT)
301
  for a, b in zip(traj_xy[:-1], traj_xy[1:]):
302
+ draw.line((a[0], a[1], b[0], b[1]), fill=(167, 240, 120, 205), width=2)
303
+ draw_label(draw, (16, 14), "camera pose + SLAM map", fill=(244, 248, 239), size=17)
304
  return image_data_uri(canvas, "PNG")
305
 
306
 
 
314
  accel = np.array(h5["imu/accel_xyz"][max(0, key_idx - 220): key_idx + 220], dtype=np.float64)
315
  gyro = np.array(h5["imu/gyro_xyz"][max(0, key_idx - 220): key_idx + 220], dtype=np.float64)
316
  series = [accel[:, 0], accel[:, 1], accel[:, 2], gyro[:, 0], gyro[:, 1], gyro[:, 2]]
317
+ colors = [(167, 240, 120), (122, 229, 195), (155, 223, 255), (216, 244, 165), (244, 248, 239), (165, 175, 162)]
318
  for row in range(4):
319
  y = 26 + row * 33
320
+ draw.line((18, y, THUMB_WIDTH - 18, y), fill=(167, 240, 120, 48), width=1)
321
  for values, color in zip(series, colors):
322
  values = values[:420]
323
  if len(values) < 2:
 
330
  y = 138 - np.clip(v, 0, 1) * 112
331
  pts.append((x, y))
332
  draw.line(pts, fill=color + (200,), width=2)
333
+ draw_label(draw, (16, 12), "inertial accel / gyro", fill=(244, 248, 239), size=17)
334
  return image_data_uri(canvas, "PNG")
335
 
336
 
 
357
 
358
  body_xy = project(body, 18, 165)
359
  for x, y in body_xy:
360
+ draw.ellipse((x - 2.4, y - 2.4, x + 2.4, y + 2.4), fill=(167, 240, 120, 185))
361
  for a, b in zip(body_xy[:-1], body_xy[1:]):
362
+ draw.line((a[0], a[1], b[0], b[1]), fill=(167, 240, 120, 82), width=1)
363
 
364
+ for points, x_offset, color in [(left, 218, (122, 229, 195)), (right, 314, (216, 244, 165))]:
365
  xy = project(points, x_offset, 82)
366
  for a, b in HAND_EDGES:
367
  draw.line((xy[a][0], xy[a][1], xy[b][0], xy[b][1]), fill=color + (180,), width=2)
368
  for x, y in xy:
369
  draw.ellipse((x - 2.4, y - 2.4, x + 2.4, y + 2.4), fill=color + (220,))
370
+ draw_label(draw, (16, 12), "body + hand mocap", fill=(244, 248, 239), size=17)
371
  return image_data_uri(canvas, "PNG")
372
 
373
 
 
383
  actions = [a.get("label", "") for a in segment.get("Current Action", [])][:2]
384
  canvas = make_canvas()
385
  draw = ImageDraw.Draw(canvas, "RGBA")
386
+ draw_label(draw, (16, 13), "language annotation", fill=(244, 248, 239), size=17)
387
  y = 46
388
  for label in objects:
389
+ draw.rounded_rectangle((16, y, 16 + 20 + len(label) * 8, y + 24), radius=6, fill=(7, 18, 7, 235), outline=(167, 240, 120, 170))
390
+ draw_label(draw, (26, y + 5), label, fill=(244, 248, 239), size=12)
391
  y += 30
392
  x = 184
393
  y = 48
394
  for action in actions:
395
  wrapped = action[:32] + ("..." if len(action) > 32 else "")
396
+ draw.rounded_rectangle((x, y, THUMB_WIDTH - 16, y + 36), radius=7, fill=(7, 18, 7, 235), outline=(122, 229, 195, 170))
397
+ draw_label(draw, (x + 10, y + 10), wrapped, fill=(244, 248, 239), size=12)
398
  y += 44
399
  return image_data_uri(canvas, "PNG")
400
 
 
569
  margin: 0;
570
  width: {CANVAS_WIDTH}px;
571
  height: {CANVAS_HEIGHT}px;
572
+ background: #020502;
573
  }}
574
  body {{
575
+ font-family: "Inter Tight", "Space Grotesk", ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif;
576
+ color: #f4f8ef;
577
  text-rendering: optimizeLegibility;
578
  }}
579
  .canvas {{
 
583
  overflow: hidden;
584
  padding: 54px 64px 44px;
585
  background:
586
+ radial-gradient(circle at 72% 10%, rgba(167,240,120,0.18), transparent 24%),
587
+ radial-gradient(circle at 20% 28%, rgba(255,255,255,0.10) 1px, transparent 2px),
588
+ #020502;
589
+ background-size: auto, 18px 18px, auto;
 
 
590
  }}
591
  .image-background {{
592
  position: absolute;
 
594
  background-position: center;
595
  background-repeat: no-repeat;
596
  background-size: cover;
597
+ opacity: 0.36;
598
+ filter: saturate(1.05) contrast(1.08) brightness(0.42);
599
  }}
600
  .content {{
601
  position: relative;
 
607
  gap: 44px;
608
  align-items: end;
609
  padding-bottom: 30px;
610
+ border-bottom: 1px solid rgba(167,240,120,0.20);
611
  }}
612
  .kicker {{
613
  display: inline-flex;
614
  align-items: center;
615
  gap: 12px;
616
+ color: #a7f078;
617
  font-family: "SF Mono", "JetBrains Mono", ui-monospace, monospace;
618
  font-size: 15px;
619
  text-transform: uppercase;
 
623
  content: "";
624
  width: 44px;
625
  height: 1px;
626
+ background: #a7f078;
627
  }}
628
  h1 {{
629
  margin: 18px 0 0;
 
635
  .subtitle {{
636
  margin: 18px 0 0;
637
  max-width: 900px;
638
+ color: #dce8d7;
639
  font-size: 23px;
640
  line-height: 1.35;
641
  font-weight: 520;
 
648
  .stat {{
649
  min-height: 78px;
650
  padding: 14px 15px;
651
+ border: 1px solid rgba(167,240,120,0.24);
652
+ background: rgba(7,18,7,0.80);
653
+ border-radius: 8px;
654
  }}
655
  .stat strong {{
656
  display: block;
 
662
  .stat span {{
663
  display: block;
664
  margin-top: 8px;
665
+ color: #a5afa2;
666
  font-size: 13px;
667
  line-height: 1.15;
668
  }}
 
671
  align-items: center;
672
  justify-content: space-between;
673
  margin: 28px 0 14px;
674
+ color: #a5afa2;
675
  font-family: "SF Mono", "JetBrains Mono", ui-monospace, monospace;
676
  font-size: 14px;
677
  text-transform: uppercase;
678
  letter-spacing: 0.08em;
679
  }}
680
  .section-label span:last-child {{
681
+ color: #dce8d7;
682
  text-transform: none;
683
  letter-spacing: 0;
684
  font-family: inherit;
 
691
  .modality {{
692
  min-height: 204px;
693
  padding: 11px 12px 14px;
694
+ border: 1px solid rgba(167,240,120,0.22);
695
+ background: rgba(7,18,7,0.84);
696
+ border-radius: 8px;
697
  }}
698
  .modality-thumb {{
699
  height: 86px;
700
  overflow: hidden;
701
+ border: 1px solid rgba(167,240,120,0.16);
702
+ border-radius: 8px;
703
+ background: #020502;
704
  }}
705
  .modality-thumb img {{
706
  display: block;
 
714
  font-variant-numeric: tabular-nums;
715
  }}
716
  .modality-index {{
717
+ color: #a5afa2;
718
  font-size: 12px;
719
  margin-top: 10px;
720
  }}
 
726
  }}
727
  .modality p {{
728
  margin: 9px 0 0;
729
+ color: #dce8d7;
730
  font-size: 14px;
731
  font-weight: 650;
732
  }}
733
  .modality span {{
734
  display: block;
735
  margin-top: 5px;
736
+ color: #a5afa2;
737
  font-size: 13px;
738
  }}
739
  .shared-band {{
 
743
  align-items: center;
744
  margin-top: 20px;
745
  padding: 14px;
746
+ border: 1px solid rgba(167,240,120,0.22);
747
+ background: rgba(7,18,7,0.72);
748
+ border-radius: 8px;
749
  }}
750
  .step {{
751
  min-height: 62px;
752
  padding: 13px 15px;
753
+ background: rgba(7,18,7,0.92);
754
+ border: 1px solid rgba(167,240,120,0.16);
755
+ border-radius: 8px;
756
  }}
757
  .step strong {{
758
  display: block;
 
762
  .step span {{
763
  display: block;
764
  margin-top: 5px;
765
+ color: #a5afa2;
766
  font-size: 13px;
767
  }}
768
  .arrow {{
769
+ color: #a7f078;
770
  font-family: "SF Mono", "JetBrains Mono", ui-monospace, monospace;
771
  font-size: 22px;
772
  }}
 
778
  }}
779
  .family {{
780
  padding: 17px;
781
+ border: 1px solid color-mix(in srgb, var(--accent) 28%, #020502);
782
+ background: rgba(7,18,7,0.82);
783
+ border-radius: 8px;
784
  }}
785
  .family-head {{
786
  display: flex;
 
789
  gap: 16px;
790
  min-height: 78px;
791
  padding-bottom: 14px;
792
+ border-bottom: 1px solid color-mix(in srgb, var(--accent) 24%, #020502);
793
  }}
794
  .family-head span {{
795
  color: var(--accent);
 
813
  .task-card {{
814
  min-height: 168px;
815
  padding: 17px 18px;
816
+ border: 1px solid color-mix(in srgb, var(--accent) 28%, #020502);
817
+ background: linear-gradient(180deg, rgba(10,24,10,0.96), color-mix(in srgb, var(--soft) 24%, #071207));
818
+ border-radius: 8px;
819
  }}
820
  .task-meta {{
821
  display: flex;
 
824
  gap: 12px;
825
  }}
826
  .index {{
827
+ color: #a5afa2;
828
  font-size: 12px;
829
  }}
830
  .kind {{
 
833
  height: 24px;
834
  padding: 0 9px;
835
  border-radius: 6px;
836
+ border: 1px solid color-mix(in srgb, var(--accent) 40%, #020502);
837
  color: var(--accent);
838
+ background: rgba(2,5,2,0.48);
839
  text-transform: uppercase;
840
  font-size: 11px;
841
  line-height: 1;
 
843
  }}
844
  .task-card h3 {{
845
  margin: 12px 0 0;
846
+ color: #f4f8ef;
847
  font-family: "SF Mono", "JetBrains Mono", ui-monospace, monospace;
848
  font-size: 21px;
849
  line-height: 1.18;
 
852
  .task-card p {{
853
  margin: 11px 0 0;
854
  min-height: 39px;
855
+ color: #dce8d7;
856
  font-size: 15px;
857
  line-height: 1.28;
858
  font-weight: 560;
 
865
  min-height: 32px;
866
  padding: 7px 10px;
867
  border-radius: 8px;
868
+ border: 1px solid color-mix(in srgb, var(--accent) 42%, #020502);
869
+ background: rgba(2,5,2,0.42);
870
  }}
871
  .metric.neural {{
872
  margin-left: 8px;
873
+ border-color: rgba(255,255,255,0.20);
874
+ background: rgba(255,255,255,0.08);
875
  }}
876
  .metric span {{
877
+ color: #a5afa2;
878
  font-size: 13px;
879
  font-weight: 760;
880
  }}
 
893
  gap: 32px;
894
  margin-top: 22px;
895
  padding-top: 20px;
896
+ border-top: 1px solid rgba(167,240,120,0.20);
897
+ color: #a5afa2;
898
  font-size: 18px;
899
  line-height: 1.35;
900
  font-weight: 620;
901
  }}
902
  .footer code {{
903
  font-family: "SF Mono", "JetBrains Mono", ui-monospace, monospace;
904
+ color: #020502;
905
+ background: #a7f078;
906
+ border: 1px solid #a7f078;
907
  border-radius: 7px;
908
  padding: 6px 9px;
909
  white-space: nowrap;
scripts/research_direction_extension_tasks.py CHANGED
@@ -691,7 +691,7 @@ def write_markdown(payload: dict[str, Any]) -> None:
691
  (OUT_DIR / "research_direction_extension_summary.md").write_text("\n".join(lines).rstrip() + "\n", encoding="utf-8")
692
 
693
 
694
- def svg_text(x: int, y: int, text: str, size: int = 16, weight: int = 500, color: str = "#1f2421") -> str:
695
  return (
696
  f'<text x="{x}" y="{y}" font-size="{size}" font-weight="{weight}" '
697
  f'fill="{color}">{html.escape(text)}</text>'
@@ -702,13 +702,13 @@ def write_svg(payload: dict[str, Any]) -> None:
702
  CHARTS.mkdir(parents=True, exist_ok=True)
703
  width = 1420
704
  height = 920
705
- colors = {"A": "#1f6c9f", "B": "#2e7775", "C": "#956400", "D": "#346538"}
706
  svg: list[str] = [
707
  f'<svg xmlns="http://www.w3.org/2000/svg" width="{width}" height="{height}" viewBox="0 0 {width} {height}">',
708
- '<rect width="1420" height="920" fill="#fbfaf7"/>',
709
- '<rect x="28" y="28" width="1364" height="864" rx="18" fill="#fffefd" stroke="#e4ded4"/>',
710
  svg_text(66, 88, "Ropedia Xperience-10M: four direction extension probes", 32, 760),
711
- svg_text(66, 122, "Data-backed from the same 1,161-window public sample feature tensor; extension probes, not full direction claims.", 17, 500, "#6f716c"),
712
  ]
713
  x0 = 66
714
  y0 = 166
@@ -728,15 +728,15 @@ def write_svg(payload: dict[str, Any]) -> None:
728
  metric = spec["metric_name"]
729
  svg.extend(
730
  [
731
- f'<rect x="{x}" y="{y}" width="{card_w}" height="{card_h}" rx="10" fill="#fbfaf7" stroke="#e4ded4"/>',
732
  f'<rect x="{x}" y="{y}" width="10" height="{card_h}" rx="5" fill="{color}"/>',
733
  f'<circle cx="{x + 42}" cy="{y + 40}" r="24" fill="{color}" opacity="0.14"/>',
734
  svg_text(x + 32, y + 48, spec["direction"], 21, 760, color),
735
  svg_text(x + 76, y + 35, spec["name"], 20, 760),
736
- svg_text(x + 76, y + 62, spec["direction_name"], 13, 650, "#6f716c"),
737
- svg_text(x + 76, y + 94, f"Minimal: {fmt_metric(min_v, spec['metric_key'])} {metric}", 16, 700, "#1f2421"),
738
- svg_text(x + 300, y + 94, f"Neural MLP: {fmt_metric(nn_v, spec['metric_key'])} {metric}", 16, 700, "#1f2421"),
739
- svg_text(x + 76, y + 125, spec["output"], 13, 500, "#4e514c"),
740
  ]
741
  )
742
  min_score = choose_score(task, result["minimal"])
@@ -744,22 +744,22 @@ def write_svg(payload: dict[str, Any]) -> None:
744
  bar_x = x + 76
745
  bar_y = y + 138
746
  bar_w = 440
747
- svg.append(f'<rect x="{bar_x}" y="{bar_y}" width="{bar_w}" height="8" rx="4" fill="#eee8df"/>')
748
  svg.append(f'<rect x="{bar_x}" y="{bar_y}" width="{max(4, min(bar_w, bar_w * min_score)):.1f}" height="8" rx="4" fill="{color}" opacity="0.72"/>')
749
- svg.append(f'<rect x="{bar_x}" y="{bar_y + 12}" width="{bar_w}" height="8" rx="4" fill="#eee8df"/>')
750
- svg.append(f'<rect x="{bar_x}" y="{bar_y + 12}" width="{max(4, min(bar_w, bar_w * nn_score)):.1f}" height="8" rx="4" fill="#1f2421" opacity="0.72"/>')
751
 
752
  legend_y = 570
753
  svg.extend(
754
  [
755
  svg_text(66, legend_y, "How to read this", 24, 760),
756
- svg_text(66, legend_y + 34, "Each card adds one concrete task to a research direction using existing sample modalities.", 16, 500, "#4e514c"),
757
- svg_text(66, legend_y + 62, "Colored bar: minimal baseline normalized score. Dark bar: neural MLP normalized score. Lower-is-better MAE is shown as 1 - MAE for bar length only.", 16, 500, "#4e514c"),
758
- '<line x1="66" y1="675" x2="1354" y2="675" stroke="#e4ded4"/>',
759
  svg_text(66, 724, "Implementation boundary", 22, 760),
760
- svg_text(66, 758, "A: motion-energy proxy, not a full human body model. B: view-feature retrieval, not neural rendering.", 16, 500, "#4e514c"),
761
- svg_text(66, 786, "C: phase-progress regression, not open-world intent. D: ego-motion forecast, not a persistent map.", 16, 500, "#4e514c"),
762
- svg_text(66, 835, "All metrics are computed from held-out chronological windows of the same public sample episode.", 16, 700, "#1f2421"),
763
  ]
764
  )
765
  svg.append("</svg>")
 
691
  (OUT_DIR / "research_direction_extension_summary.md").write_text("\n".join(lines).rstrip() + "\n", encoding="utf-8")
692
 
693
 
694
+ def svg_text(x: int, y: int, text: str, size: int = 16, weight: int = 500, color: str = "#f4f8ef") -> str:
695
  return (
696
  f'<text x="{x}" y="{y}" font-size="{size}" font-weight="{weight}" '
697
  f'fill="{color}">{html.escape(text)}</text>'
 
702
  CHARTS.mkdir(parents=True, exist_ok=True)
703
  width = 1420
704
  height = 920
705
+ colors = {"A": "#a7f078", "B": "#7ae5c3", "C": "#d8f4a5", "D": "#9bdfff"}
706
  svg: list[str] = [
707
  f'<svg xmlns="http://www.w3.org/2000/svg" width="{width}" height="{height}" viewBox="0 0 {width} {height}">',
708
+ '<rect width="1420" height="920" fill="#020502"/>',
709
+ '<rect x="28" y="28" width="1364" height="864" rx="18" fill="#050905" stroke="#a7f078" stroke-opacity="0.24"/>',
710
  svg_text(66, 88, "Ropedia Xperience-10M: four direction extension probes", 32, 760),
711
+ svg_text(66, 122, "Data-backed from the same 1,161-window public sample feature tensor; extension probes, not full direction claims.", 17, 500, "#a5afa2"),
712
  ]
713
  x0 = 66
714
  y0 = 166
 
728
  metric = spec["metric_name"]
729
  svg.extend(
730
  [
731
+ f'<rect x="{x}" y="{y}" width="{card_w}" height="{card_h}" rx="10" fill="#071207" stroke="#a7f078" stroke-opacity="0.22"/>',
732
  f'<rect x="{x}" y="{y}" width="10" height="{card_h}" rx="5" fill="{color}"/>',
733
  f'<circle cx="{x + 42}" cy="{y + 40}" r="24" fill="{color}" opacity="0.14"/>',
734
  svg_text(x + 32, y + 48, spec["direction"], 21, 760, color),
735
  svg_text(x + 76, y + 35, spec["name"], 20, 760),
736
+ svg_text(x + 76, y + 62, spec["direction_name"], 13, 650, "#a5afa2"),
737
+ svg_text(x + 76, y + 94, f"Minimal: {fmt_metric(min_v, spec['metric_key'])} {metric}", 16, 700, "#f4f8ef"),
738
+ svg_text(x + 300, y + 94, f"Neural MLP: {fmt_metric(nn_v, spec['metric_key'])} {metric}", 16, 700, "#f4f8ef"),
739
+ svg_text(x + 76, y + 125, spec["output"], 13, 500, "#dce8d7"),
740
  ]
741
  )
742
  min_score = choose_score(task, result["minimal"])
 
744
  bar_x = x + 76
745
  bar_y = y + 138
746
  bar_w = 440
747
+ svg.append(f'<rect x="{bar_x}" y="{bar_y}" width="{bar_w}" height="8" rx="4" fill="#a7f078" opacity="0.14"/>')
748
  svg.append(f'<rect x="{bar_x}" y="{bar_y}" width="{max(4, min(bar_w, bar_w * min_score)):.1f}" height="8" rx="4" fill="{color}" opacity="0.72"/>')
749
+ svg.append(f'<rect x="{bar_x}" y="{bar_y + 12}" width="{bar_w}" height="8" rx="4" fill="#a7f078" opacity="0.14"/>')
750
+ svg.append(f'<rect x="{bar_x}" y="{bar_y + 12}" width="{max(4, min(bar_w, bar_w * nn_score)):.1f}" height="8" rx="4" fill="#ffffff" opacity="0.78"/>')
751
 
752
  legend_y = 570
753
  svg.extend(
754
  [
755
  svg_text(66, legend_y, "How to read this", 24, 760),
756
+ svg_text(66, legend_y + 34, "Each card adds one concrete task to a research direction using existing sample modalities.", 16, 500, "#dce8d7"),
757
+ svg_text(66, legend_y + 62, "Colored bar: minimal baseline normalized score. White bar: neural MLP normalized score. Lower-is-better MAE is shown as 1 - MAE for bar length only.", 16, 500, "#dce8d7"),
758
+ '<line x1="66" y1="675" x2="1354" y2="675" stroke="#a7f078" stroke-opacity="0.18"/>',
759
  svg_text(66, 724, "Implementation boundary", 22, 760),
760
+ svg_text(66, 758, "A: motion-energy proxy, not a full human body model. B: view-feature retrieval, not neural rendering.", 16, 500, "#dce8d7"),
761
+ svg_text(66, 786, "C: phase-progress regression, not open-world intent. D: ego-motion forecast, not a persistent map.", 16, 500, "#dce8d7"),
762
+ svg_text(66, 835, "All metrics are computed from held-out chronological windows of the same public sample episode.", 16, 700, "#f4f8ef"),
763
  ]
764
  )
765
  svg.append("</svg>")
scripts/research_direction_taxonomy.py CHANGED
@@ -483,7 +483,7 @@ def write_markdown(taxonomy: dict[str, Any]) -> None:
483
  )
484
 
485
 
486
- def svg_text(x: int, y: int, text: str, size: int = 16, weight: int = 500, color: str = "#16213a") -> str:
487
  return (
488
  f'<text x="{x}" y="{y}" font-size="{size}" font-weight="{weight}" '
489
  f'fill="{color}">{html.escape(text)}</text>'
@@ -496,7 +496,7 @@ def write_svg(taxonomy: dict[str, Any]) -> None:
496
  margin = 58
497
  card_w = 515
498
  card_h = 220
499
- colors = {"direct": "#1f6c9f", "proxy": "#2e7775", "diagnostic": "#956400"}
500
  cards = []
501
 
502
  for idx, (code, info) in enumerate(taxonomy["directions"].items()):
@@ -526,10 +526,10 @@ def write_svg(taxonomy: dict[str, Any]) -> None:
526
  cards.append(
527
  "\n".join(
528
  [
529
- f'<rect x="{x}" y="{y}" width="{card_w}" height="{card_h}" rx="8" fill="#ffffff" stroke="#d9e1ea"/>',
530
  svg_text(x + 24, y + 42, f"{code}. {info['name']}", 21, 700),
531
- svg_text(x + 24, y + 75, info["current_status"], 15, 700, "#566273"),
532
- svg_text(x + 24, y + 108, f"Tasks: {task_labels}", 14, 500, "#30394a"),
533
  *segments,
534
  svg_text(x + 24, y + 174, f"Direct {counts['direct']}", 14, 700, colors["direct"]),
535
  svg_text(x + 150, y + 174, f"Proxy {counts['proxy']}", 14, 700, colors["proxy"]),
@@ -548,18 +548,19 @@ def write_svg(taxonomy: dict[str, Any]) -> None:
548
  legend.extend(
549
  [
550
  f'<rect x="{lx}" y="622" width="16" height="16" rx="4" fill="{colors[key]}"/>',
551
- svg_text(lx + 24, 636, label, 14, 600, "#30394a"),
552
  ]
553
  )
554
  lx += 200
555
 
556
  svg = f"""<svg xmlns="http://www.w3.org/2000/svg" width="{width}" height="{height}" viewBox="0 0 {width} {height}" role="img" aria-label="Xperience-10M task coverage across four research directions">
557
- <rect width="100%" height="100%" fill="#f7f9fb"/>
 
558
  {svg_text(margin, 64, "Xperience-10M 12-Task Suite: Four Research Directions", 30, 800)}
559
- {svg_text(margin, 96, "One public sample episode, two baseline families, explicit direct/proxy/diagnostic coverage.", 16, 500, "#566273")}
560
  {"".join(cards)}
561
  {"".join(legend)}
562
- {svg_text(margin, 670, "Generated from results/episode_task_suite/summary_report.json and scripts/research_direction_taxonomy.py", 13, 500, "#6d7787")}
563
  </svg>
564
  """
565
  (CHARTS / "research_direction_coverage.svg").write_text(svg, encoding="utf-8")
 
483
  )
484
 
485
 
486
+ def svg_text(x: int, y: int, text: str, size: int = 16, weight: int = 500, color: str = "#f4f8ef") -> str:
487
  return (
488
  f'<text x="{x}" y="{y}" font-size="{size}" font-weight="{weight}" '
489
  f'fill="{color}">{html.escape(text)}</text>'
 
496
  margin = 58
497
  card_w = 515
498
  card_h = 220
499
+ colors = {"direct": "#a7f078", "proxy": "#7ae5c3", "diagnostic": "#d8f4a5"}
500
  cards = []
501
 
502
  for idx, (code, info) in enumerate(taxonomy["directions"].items()):
 
526
  cards.append(
527
  "\n".join(
528
  [
529
+ f'<rect x="{x}" y="{y}" width="{card_w}" height="{card_h}" rx="8" fill="#050905" stroke="#a7f078" stroke-opacity="0.24"/>',
530
  svg_text(x + 24, y + 42, f"{code}. {info['name']}", 21, 700),
531
+ svg_text(x + 24, y + 75, info["current_status"], 15, 700, "#a7f078"),
532
+ svg_text(x + 24, y + 108, f"Tasks: {task_labels}", 14, 500, "#dce8d7"),
533
  *segments,
534
  svg_text(x + 24, y + 174, f"Direct {counts['direct']}", 14, 700, colors["direct"]),
535
  svg_text(x + 150, y + 174, f"Proxy {counts['proxy']}", 14, 700, colors["proxy"]),
 
548
  legend.extend(
549
  [
550
  f'<rect x="{lx}" y="622" width="16" height="16" rx="4" fill="{colors[key]}"/>',
551
+ svg_text(lx + 24, 636, label, 14, 600, "#dce8d7"),
552
  ]
553
  )
554
  lx += 200
555
 
556
  svg = f"""<svg xmlns="http://www.w3.org/2000/svg" width="{width}" height="{height}" viewBox="0 0 {width} {height}" role="img" aria-label="Xperience-10M task coverage across four research directions">
557
+ <rect width="100%" height="100%" fill="#020502"/>
558
+ <rect x="24" y="24" width="1132" height="652" rx="20" fill="#050905" stroke="#a7f078" stroke-opacity="0.24"/>
559
  {svg_text(margin, 64, "Xperience-10M 12-Task Suite: Four Research Directions", 30, 800)}
560
+ {svg_text(margin, 96, "One public sample episode, two baseline families, explicit direct/proxy/diagnostic coverage.", 16, 500, "#a5afa2")}
561
  {"".join(cards)}
562
  {"".join(legend)}
563
+ {svg_text(margin, 670, "Generated from results/episode_task_suite/summary_report.json and scripts/research_direction_taxonomy.py", 13, 500, "#a5afa2")}
564
  </svg>
565
  """
566
  (CHARTS / "research_direction_coverage.svg").write_text(svg, encoding="utf-8")