The two Gemma 2 model members reflect the current AI landscape, with technological innovations allowing for smaller and lighter models to tackle different user requests. Google is giving developers a choice in how they might want to use these models—either on-device or through the cloud. As the models are open-sourced, they can be easily customized and integrated into various projects. It remains to be seen if the existing Gemma variants—CodeGemma, RecurrentGemma and PaliGemma—will benefit from these two Gemma 2 models.
Key takeaways:
- Google's Gemma 2, an open lightweight model series, will be available to researchers and developers through Vertex AI starting next month.
- Initially, Gemma 2 only contained a 27-billion parameter member, but Google surprised by also including a 9-billion one.
- The Gemma 2 model is designed to run on Nvidia’s latest GPUs or a single TPU host in Vertex AI, targeting developers who want to incorporate AI into their apps or edge devices.
- Being open-sourced, Gemma 2 can be easily customized and integrated into various projects that Google may not usually foresee.