PT Prototypes
← All seeds

tinytrainer-mobile

dormant prototype

TinyTrainer Mobile

Proves the full loop: import kit -> classify locally -> correct predictions -> personalize on device -> measure improvement.

iOS/iPadOS reference app for on-device ML personalization.

Version
0.1.0
License
MIT
Kind
desktop
Languages
SwiftUI, Core ML, NaturalLanguage, ZIPFoundation
Lines of code
0
Last commit
2026-04-20 (0d ago)

Health signals

Tests: no README: yes LICENSE: yes Buildable: unknown

Patterns proven here

onDeviceMLPersonalization data-model
End-to-end pipeline for personalizing an ML model on-device using user corrections.

Why it matters

Demonstrates a complete end-to-end ML pipeline running entirely on-device, from importing a pre-trained model to personalizing it with user corrections and measuring accuracy improvements.

View source →