Notebooks for Gemma models and variants.
| Notebook Name | Description |
|---|---|
| Agentic_RAG.ipynb | Build an Agentic RAG system that intelligently decides when to call functions, uses a Qdrant based RAG pipeline, and falls back to Google Search. Trace and monitor with OPIK. |
| Image_Segmentation.ipynb | Image Segmentation Task with Gemma 4 |
| On_Device_AI.ipynb | Build a fully on-device RAG pipeline using Lite-RT and Qdrant Edge |
| RAG_with_EmbeddingGemma.ipynb | Build simple RAG with EmbeddingGemma |
| Translator_of_Old_Korean_Literature.ipynb | Fine-tuning Gemma to translate old Korean literature |
Explore Gemma's capabilities across different modalities:
| Notebook Name | Description |
|---|---|
| Text - Basic | Basic text generation and prompting with Gemma 4. |
| Text - Function Calling | Leverage Gemma 4 for tool use and function calling. |
| Vision - Image | Visual understanding and captioning with Gemma 4. |
| Vision - Video | Video understanding and analysis with Gemma 4. |
| Audio | Explore audio processing and understanding. |
| Thinking | Reasoning capabilities. |
Examples of fine-tuning Gemma models:
| Notebook Name | Description |
|---|---|
| Text Fine-tuning with QLoRA | Efficiently fine-tune Gemma 4 for text tasks using QLoRA. |
| Vision Fine-tuning with QLoRA | Efficiently fine-tune Gemma 4 for vision tasks using QLoRA. |