GoogleDeepMind Releases DataGemma and Gemma 2 to Enhance LLM Accuracy with Real-World Data

Sep 12, 2024, 01:14 PM

Google has released DataGemma, a series of open models designed to enhance the factual accuracy of large language models (LLMs) by grounding them with real-world data from Data Commons. DataGemma employs Retrieval Interleaved Generation (RIG) and Retrieval Augmented Generation (RAG) techniques to reduce LLM hallucinations. The models, including Gemma 2, are fine-tuned for these approaches to incorporate public statistical data, thereby improving the reliability and responsibility of AI outputs. Data Commons is a publicly available knowledge graph containing over 240 billion data points across hundreds of thousands of statistical variables, sourced from trusted organizations like the UN and WHO. The release was announced by GoogleDeepMind.