Evaluate LLM to use

[phin01]

Current candidate Mistral https://mistral.ai/

RAG - LlamaIndex https://github.com/mistralai/cookbook/blob/main/llamaindex_agentic_rag.ipynb

Can be tested out locally and seems to be a good choice for answering questions based on multiple documents

https://www.llamaindex.ai/blog/evaluating-the-ideal-chunk-size-for-a-rag-system-using-llamaindex-6207e5d3fec5

https://www.restack.io/docs/llamaindex-knowledge-llamaindex-json-parser

https://docs.llamaindex.ai/en/stable/examples/query_engine/JSONalyze_query_engine/

[phin01]

Check Meta's Llama3 https://llama.meta.com/docs/get-started/

Check OLamma https://ollama.com/