Skip to content

Add model-eval Tools, theme customizer, and Labs UI polish #9

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Nostromo-618 merged 3 commits into from
Aug 8, 2026
Merged

Add model-eval Tools, theme customizer, and Labs UI polish #9

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
f6a99d1
Declare Node >=20.9.0 to match sharp@0.35 engines.
Nostromo-618 Aug 8, 2026
a775f09
Add model-eval tools, theme customizer with vdl storage, rotating hom…
Nostromo-618 Aug 8, 2026
96d55f0
Fix think-tag streaming sanitize and home-quote salt; document eval C…
Nostromo-618 Aug 8, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 6 additions & 2 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,8 @@ Current component versions:
| Component | Version | Module |
|------|------|------|
| `vdl-neptune-search` | `0.0.3` | [`neptune-search.js`](./neptune-search.js) |
| `vdl-ai-chat` | `0.0.5` | [`ai-chat.js`](./ai-chat.js) |
| `vdl-ai-chat` | `0.0.9` | [`ai-chat.js`](./ai-chat.js) |
| `vdl-model-eval` | `0.0.1` | [`model-eval.js`](./model-eval.js) |

### Shared Guardrails Modules

Expand Down Expand Up @@ -112,6 +113,8 @@ Then open:
- `http://localhost:3000/demo/neptune-demo.html` for the focused Neptune demo page
- `http://localhost:3000/demo/ai-chat-demo.html` for the focused AI Chat demo page

Theme controls live in the navbar (`VdThemeSwitcher` + `VdThemeCustomizer` with palette selector hidden — Open Color only). Labs remaps vd3 theme `localStorage` keys to a `vdl-` prefix (`vdl-theme-preference`, `vdl-palette`, …) so preferences do not collide with Vanduo docs on the shared GitHub Pages origin. Global defaults for new visitors (when no `vdl-*` preference is stored): font Open Sans (`open-sans`), neutral Neutral (`neutral`), radius `0.25`, Open Color palette. Existing `vdl-*` preferences are preserved.

Production build / preview:

```bash
Expand Down Expand Up @@ -148,7 +151,8 @@ pnpm dev

### Notes

- Defaults to **Gemma 4 E2B**, with Gemma 4 E4B as the quality tier; optional small/fast models (SmolLM2, Qwen2.5, Llama 3.2, Qwen Coder) remain available.
- Defaults to **Gemma 4 E2B** (LiteRT web-official); Tiny is **Qwen3 0.6B WebLLM** (SmolLM2 removed). LiteRT Qwen3/Ministral kept as documented spikes (PrefillDecode load blocked in current LiteRT-LM.js).
- Local eval helper: `pnpm model-eval` → Tools page `#tools/model-eval` ([doc/vdl-model-eval.md](./doc/vdl-model-eval.md)).
- Detects runtime hardware capabilities (WebGPU + `shader-f16`) and shows system compatibility info in setup UI.
- Automatically applies compatible fallback variants on lower-capability devices when needed (optional built-ins).
- Model download is user-triggered and cached by the browser.
Expand Down
Loading
Loading