Pushing the capabilities of Gemma 3 via distillation and RL fine-tuning

youtube
Pushing the capabilities of Gemma 3 via distillation and RL fine-tuning Specialized capabilities (e.g. math abilities, coding, multilinguality, tool use...) are key areas of improvement in post-training. In this talk we explore a novel strategy involving large-scale distillation and RL finetuning to push specialized capabilities in LMs while still improving their generality. Subscribe to Google for Developers → Speakers: Johan Ferret Products Mentioned: Gemma
  2025/04/02      youtube

Our Tag

最近投稿されたプログラミング学習動画

Let’s rewind it back to Google IO ‘24

Google

Get ready for #GoogleIO May 20-21, where...

  2025/04/03

Architecting for Multi-Cloud: AWS and Beyond with PwC

Amazon
cloud

This webinar explores the fundamentals o...

  2025/04/03

Can you beat our time solving the green world? #GoogleIO

iot
Google

Think you’ve mastered the #GoogleIO puzz...

  2025/04/03

Pushing the capabilities of Gemma 3 via distillation and RL fine-tunin

Specialized capabilities (e.g. math abil...

  2025/04/02

Welcome to the Gemmaverse

The Gemma family of open models keeps ev...

  2025/04/02

Gemma on mobile and web. Best and worst practices

モバイル

Come learn new methods and best practice...

  2025/04/02

ShieldGemma 2 – Developing safe and responsible AI for images

Responsible AI development in multimodal...

  2025/04/02

Agentic AI vs Generative AI | Agentic AI vs Generative AI Explained |

🔥Generative AI Course: Masters Program: ...

  2025/04/02

Modern Observability and Event Driven Architectures - Martin Thwaites

This talk was recorded at NDC London in ...

  2025/04/02

Advanced Cloud Native Development with .NET Aspire - Scott Hunter & Ma

cloud

This talk was recorded at NDC London in ...

  2025/04/02