Leveraging large-scale models for efficient learning

From the creator

Struggling to improve your model's performance without increasing its size, even after exhaustive data scraping and architectural tweaks? If you have access to a larger, well-performing model and the resources to train it, knowledge distillation could be the answer. This technique leverages a strong model to guide the training of a smaller one, leading to improved learning and faster convergence. This presentation offers a brief overview of knowledge distillation and its key benefits. Subscribe to Google for Developers → https://goo.gle/developers Speakers: Morgane Rivière Products Mentioned: Gemma

Published 2025-04-02 · Watch on YouTube ↗
Availability and playback restrictions are controlled by the original channel.