SciGemma: Fine-tune and deploy Gemma LLM on Android using MediaPipe

SciGemma is an an Android app that leverages the Gemma model's on-device capabilities. Developed by fine-tuning the Gemma 2B-it model on a custom dataset of Science textbooks and deployed on Android using MediaPipe LLM Inference task, it offers students to convenience of revising their course materials anytime, anywhere, and offline.

The project pipeline covers end-to-end aspects right from preparing a custom dataset, fine-tuning the Gemma model, and deploy it on a mobile device using the MediaPipe LLM Inference task. We've curated these steps in a series of three blogs.

Part 1: Preparing custom dataset

Part 2: Fine-tune Gemma 2B-it model

Part 3: Deploying Gemma on Android

GitHub repository:

Try out the model quickly on Hugging Face Spaces:

If you liked our project, don't forget to leave a star.

#Gemma #GemmaSprint #GenerativeAI #GoogleDeveloperExperts #MachineLearning #Android #MediaPipe #LargeLanguageModels #LLM

[PrasanthPamula]

Can you briefly explain how to achive it.