A truly multilingual Gemma 3

Channel:
United States Google for Developers
Subscribers:
2,520,000
Published on ● Video Link: https://www.youtube.com/watch?v=Sgmye159YQI


Duration: 0:00
250 views
9

Building multilingual AI applications is crucial for reaching global audiences, and varied language proficiency remains a top developer priority. In this talk, we’ll explore the training strategies applied to make Gemma 3 fluent in over 100 languages, and showcase its state-of-the-art performance on multilingual tasks.

Subscribe to Google for Developers → https://goo.gle/developers

Speakers: Adi Mayrav Gilady
Products Mentioned: Gemma



Other Videos By Google for Developers


2025-04-03Let’s rewind it back to Google IO ‘24
2025-04-02Can you beat our time solving the green world? #GoogleIO
2025-04-02Pushing the capabilities of Gemma 3 via distillation and RL fine-tuning
2025-04-02Welcome to the Gemmaverse
2025-04-02Gemma on mobile and web. Best and worst practices
2025-04-02ShieldGemma 2 – Developing safe and responsible AI for images
2025-04-02Integrating the Gemma model family in Transformers
2025-04-02RLHF for finer alignment with Gemma 3
2025-04-02Open models and open compression
2025-04-02SynthID – Watermarking and identifying AI-generated text
2025-04-02A truly multilingual Gemma 3
2025-04-02PaliGemma – Making Gemma 2 see by adding a vision encoder
2025-04-02Gemma 3 architecture and design
2025-04-02LeRobot – Lowering the entry barrier to AI for robotics
2025-04-01Panel: The importance of open models
2025-04-01Building intelligent agents with Gemma 3
2025-04-01Leveraging large-scale models for efficient learning
2025-04-01Live demo: Gemma 3 on Jetson Orin Nano
2025-04-01Using and fine-tuning Gemma 3
2025-04-01Deepdive into Gemma 3
2025-04-01Welcome to Gemma 3