Google for Developers’ Post

✨ Introducing DataGemma, open models that enhance LLM factuality by grounding them with real-world data from Google’s Data Commons. → https://goo.gle/4dWFyKn Learn how DataGemma uses two distinct RIG and RAG approaches to: 📖 Harness the knowledge of Data Commons 🧠 Fact check numeric data against trusted sources 🛠️ Lead the way in developing responsible AI And more for the #Gemmaverse!

🧐 Take a close look at RAG (Retrieval Augmented Generation), an established approach used by DataGemma to generate a comprehensive response utilizing Data Commons. → https://goo.gle/3TtPP8j Using the RAG approach, relevant information is retrieved from Data Commons before output text is generated. This information creates an augmented prompt used by Gemini 1.5 Pro to generate a grounded response & mitigate the chance for “hallucinations,” increasing trustworthy output.

🧑💻 Sometimes you just need to stay in your flow.

Let the Gemini API help you design. Need some inspiration? Try building the below apps using the Gemini API: https://goo.gle/3XnZ3nK

💻 Spend less time troubleshooting and more time building → https://goo.gle/3XtG8Z2 The Google Developer Program gives you the tools and resources to find the information you need in your projects, create and troubleshoot code, get your questions answered, and connect with a global community of developers.

🔎 Explore the inner workings of RIG (Retrieval-Interleaved Generation), one of the approaches used by DataGemma to double-check its work against a trusted source. → https://goo.gle/4eo54Yt RIG fine-tunes Gemma 2 to recognize statistical data within its generated responses and compares it against a trusted source. The model then uses natural language queries to "call" Google's Data Commons, verifying the data and replacing potentially inaccurate numbers in its responses, ultimately decreasing the risk of "hallucinations.”

📝 Clean code starts with clear variable names. What are your best practices for writing readable and maintainable code?