Gemma 7B IT

Google's 7B-parameter Instruction-Tuned (IT) model: a lightweight, state-of-the-art LLM optimized for chat and on-device deployment.

This is Gemma 7B IT (Instruction-Tuned): a high-performance, open-weight language model from Google, built on the same research as the Gemini family. The model features 7 billion parameters and is a text-to-text, decoder-only architecture. Its instruction-tuned variant excels at conversational use, question answering, and reasoning tasks, demonstrating strong benchmarks in MMLU and HellaSwag. Crucially, its lightweight design allows for efficient deployment on resource-constrained environments (laptops, desktops, or private cloud infrastructure), democratizing access to powerful AI capabilities. The 1.1 update further refines its performance via a novel RLHF method, specifically enhancing factuality and multi-turn conversation quality.

https://huggingface.co/google/gemma-1.1-7b-it