cinnabrad/ai-edge-gallery
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
Fork of Google AI Edge Gallery with telemetry stripped
337 commits 1 branch 0 tags 10 MiB
  • Kotlin 91%
  • HTML 8.4%
  • JavaScript 0.5%
Find a file
Exact
Rudy 8003fd8b01
All checks were successful
Build Android APK / Build Android APK (push) Successful in 3m11s
Details
Make Prompt Lab system prompt editable 
- Add curSystemPrompt state and allowEditingSystemPrompt=true
- Changes apply on model reload (switch models / restart)
- AI Chat, Ask Image, Ask Audio already had real-time reset support
2026-04-05 21:36:09 -04:00
.forgejo/workflows Fix: Use comfychair workflow pattern - cache fix + debug build 2026-04-04 17:24:07 -04:00
.github Update template 2026-04-01 17:13:52 -07:00
Android Make Prompt Lab system prompt editable 2026-04-05 21:36:09 -04:00
model_allowlists update allowlist and README 2026-04-02 09:01:32 -07:00
skills Update README.md 2026-04-02 09:34:42 -07:00
.gitignore
Bug_Reporting_Guide.md
CONTRIBUTING.md
DEVELOPMENT.md
Function_Calling_Guide.md Create Function_Calling_Guide.md 2026-02-24 21:15:44 -08:00
LICENSE
model_allowlist.json
README.md Add README documenting fork changes 2026-04-04 16:06:59 -04:00

README.md

Bun Gallery

Fork of Google AI Edge Gallery with telemetry stripped and rebranded.

Changes from Upstream

  • Removed Firebase telemetry: Analytics and Firebase Cloud Messaging removed
  • Removed Firebase permissions: No more tracking or push notification permissions
  • Rebranded: com.google.ai.edge.gallery → ca.bun.gallery
  • App name: Google AI Edge Gallery → Bun Gallery
  • Stable telemetry: All firebaseAnalytics calls stubbed to null

What's Kept

  • LiteRT-LM runtime (0.10.0) for on-device LLM inference
  • CPU/GPU/NPU backend support
  • Gemma model support (downloads from HuggingFace)
  • All UI features: Agent Skills, AI Chat, Ask Image, Audio Scribe, Prompt Lab, Mobile Actions, Tiny Garden

Building

cd Android/src
./gradlew assembleRelease

APK output: app/build/outputs/apk/release/app-release.apk

Adding Custom Models

  1. Convert model to LiteRT-LM format (.lite)
  2. Update ModelAllowlist.kt or create a custom config
  3. Point to your HuggingFace repo or local file

License

Apache License 2.0 - same as upstream

Upstream

This fork prioritizes privacy and removes all Google telemetry while maintaining full functionality for running on-device AI models.