RAG Eval Notebok

[hardikjshah]

🚀 Describe the new functionality needed

Notebook flow showcasing the e2e flow of how we can do evals for RAG agents.
This will merge a lot of recent work around -- eval apis, agent updates, RAG updates, ReACT agents, etc

• Local file with some benchmarking data
• Upload file using /files
• Make a dataset out of it
• Make a benchmark ( Register this dataset via /benchmarks )
• Run RAG agent Eval /eval apis
  • This RAG agent should have all the core changes around no adhoc RAG calling, tool_option, system_prompt, etc
• Make changes to the Agent prompt or give extra tool and re-eval
• [stretch] Use docling for indexing / chunking to showcase improvements

💡 Why is this needed? What if we don't build it?

This showcases a first use case of the e2e SDLC of a RAG agent.

Other thoughts

No response