Welcome to RDK S600 Developer Kit repository
8. Google Gemma3 Visual Multimodal Large Model
Gemma3
Gemma31. Model Size2. Performance3. Using Gemma33.1 Running Gemma33.2 Conducting a Dialogue3.3 Visual Functions3.4 Ending the Dialogue3.5 Chinese DialogueReferences
Demo Environment
Development Board: RDK S600 Motherboard
Gemma is a lightweight model family built by Google based on Gemini technology. Gemma 3 models are multimodal (handling text and images), have 128K context windows, and support over 140 languages.
1. Model Size
| Model | Size |
|---|---|
| gemma3:1b | 815MB |
| gemma3:4b | 3.3GB |
2. Performance
3. Using Gemma3
3.1 Running Gemma3
Use the run command to start running the model. If you haven't downloaded this model before, it will automatically pull models from the Ollam model library:
ollama run gemma3:4b
3.2 Conducting a Dialogue
xxxxxxxxxxHow to learn a programming language?
The response time depends on hardware configuration. Please be patient!
3.3 Visual Functions
xxxxxxxxxxWhat do you see in this picture? :./test_pic.png#Using ": + image path" in the dialogue allows the model to use its visual functions to interpret information from the image.
3.4 Ending the Dialogue
Use the shortcut Ctrl+d or /bye to end the dialogue!
3.5 Chinese Dialogue
For those without a Chinese input method, please refer to the Chinese input method switching tutorial.
Chinese Dialogue:
References
Ollama
Official Website: https://ollama.com/
GitHub: https://github.com/ollama/ollama
Gemma3
Ollama's corresponding model: https://ollama.com/library/gemma3