PaliGemma, Gemma 2, and Responsible AI Upgrades

Gemma is being used to create useful and approachable AI products like Octopus v2, an on-device action model, and Navarasa, a multilingual Indic variation

Gemma open models are lightweight, cutting-edge, and built with the same technology and research as Gemini models

PaliGemma, a powerful open vision-language model (VLM), and Gemma 2, Google Cloud is excited to announce their ambitions to expand the Gemma family

PaliGemma uses open components like the Gemma language model and SigLIP vision model to deliver class-leading fine-tune performance

Gemma 2 fits on less than half the compute of similar models because to its effective design

The 27B model may run on a single TPU host in Vertex AI or NVIDIA GPUs, making deployment more economical and accessible to more users

Google Cloud is offering the LLM Comparator open source as part of its Responsible Generative AI Toolkit to let developers evaluate models more thoroughly