Google has released three new AI models: Gemma 2.2B, ShieldGemma, and Gemma Scope. These models are part of the larger Gemma family, which is designed for conversational AI applications. The Gemma 2.2B model has been fine-tuned on a large dataset to improve its language understanding and generation capabilities. ShieldGemma is a variant that focuses on robustness against adversarial attacks. Gemma Scope is a smaller version of the original Gemma model, optimized for low-resource devices. These models are available on the Hugging Face model hub. --- Why it matters: These new AI models matter to researchers because they offer improved language understanding and generation capabilities, which can be applied to various conversational AI applications. Engineers can also use these models as building blocks to develop more robust and efficient conversational systems. Source: https://huggingface.co/blog/gemma-july-update