Google Introduces Gemma 2 with 27 Billion Parameters

The new AI model promises enhanced performance and efficiency, set to launch in June.

Gemma 2 is optimized to run on Nvidia's next-gen GPUs or a single TPU in Vertex AI.
The model features 27 billion parameters, a significant upgrade from its predecessors.
Google claims Gemma 2 outperforms larger models in quality and efficiency.
PaliGemma, a vision-language model for image tasks, is also part of the Gemma family.
Google's Responsible Generative AI Toolkit now includes the LLM Comparator for model evaluations.