Reported 9 months ago
During Google I/O 2024 on June 28, 2024, Google introduced the new open-source model Gemma 2 optimized for TPU and GPU acceleration, offering twice the model performance and support for up to 270 billion parameters. It also provides a smaller version with 90 billion parameters and will release a version with 26 billion parameters for mobile devices. Gemma 2 is accessible to developers and researchers through platforms like Kaggle and Colaboratory, boasting superior performance compared to other models and emphasizing security measures in its development and application.
Source: YAHOO