Google announced the release of its Gemma models, a series of lightweight open source models. They are built on the same research and technology that power their flagship Gemini models.
They released model weights in two sizes: Gemma 2B and Gemma 7B. Each size is released with pre-trained and instruction-tuned variants. They’re designed to run directly on developer-grade laptop and desktop computers, and are ready-to-use with Colab and Kaggle notebooks. They’ve also been designed for easy integration with popular tools like Hugging Face and MaxText.