NEW DataGemma-27B LLM Uncover the Truth: RIG + RAG

GoogleDeepMind just released 2 new Gemma2 fine-tuned optimized models only for access to their global Data Commons (DC) database and knowledge graph. New DataGemma 2 27B IT RIG and DataGemma 2 27B IT RAG models.

DataGemma is a series of fine-tuned Gemma 2 models used to help LLMs access and incorporate reliable public statistical data from Data Commons into their responses.

All right with authors:
DataGemma: Using real-world data to address AI hallucinations

2 new Google COLAB Notebooks:

On HuggingFace

00:00 Data Gemma 27B RIG and RAG models
00:33 Data Commons Knowledge Graph
02:39 Technical paper by Google
02:56 DataGemma 27B RIG explained
04:05 DataGemma 27B RAG explained
07:42 COLAB NB of RIG and RAG (complete code)

#airesearch
#google
#ai

[Rich_Cooking]

Great to see another vid! Keep giving us google gemini stuff I use it in production and it's the most underrated model there is, even Gemma is good

[khaos]

Thank you for sharing this! May I ask what is your personal experience with both RIG and RAG, which one of these 2 do you feel you like it's the most accurate or better

[mlcat]

Can a RIG model make not valid DC requests? For example what happens if I ask it about statistics about alien encounters and there is no such data?