Generate answers with real data
Explore and uncover verifiable insights by simply asking questions in plain language.
DataGemma
Fine-tuned Gemma 2 models that integrate retrieval techniques to ground responses in real-world data.
Explore DataGemma
The world’s first open models designed to help address the challenges of hallucination by grounding LLMs in the vast, real-world statistical data of Google's Data Commons.
Capabilities
Evaluate AI data grounding techniques
Investigate ways to guide generative AI model output with retrieval-augmented and data-interleaved techniques.
Model variants
Retrieval-Interleaved Generation
This approach uses a variant of Gemma 2 that is fine-tuned to recognize when it needs to replace a generated number with more accurate information from Data Commons.
Retrieval-Augmented Generation
This approach uses a variant of Gemma 2 that retrieves relevant information from Data Commons and then uses that information to create an extended prompt for the Gemini 1.5 Pro model.