Demo: Gemma on-device with MediaPipe and TensorFlow Lite
summary
Runs Gemma 2B on-device across web, Android and iOS using MediaPipe LLM Inference and TensorFlow Lite; delivers a live web demo and sample apps on GitHub. Code is minimal: load the web SDK and Wasm, point at a model URL, set temperature/topK/maxTokens, then call LLM inference with a prompt and a token-stream listener. Shows text generation, rewrites, classification and document Q&A working locally, highlights offline privacy and current limit of 2B models (not 7B yet).
AI-written summary · about
related talks
14mGemma on mobile and web. Best and worst practices
15mRunning Google's Gemma LLMs in the browser with MediaPipe Web
01h 21mTLMs: Tiny LLMs and Agents on Edge Devices with LiteRT-LM
42mOn-Device Machine Learning with Google AI Edge
21mFrom 46% to 90%: Fine-Tuning Tiny LLMs for On-Device Agents
46mFrom ML to LLM: on-device AI in the browser
31mFrom ML to LLM on device AI in the browser
39mFlutter Project with Gemma: A State-of-the-Art Open Model