LlamaParse: super-charging parsing of complex documents

LlamaParse is a free service from LlamaIndex that provides industry-leading parsing of complex documents. In this example we explore parsing instructions, which allow you to dramatically transform documents during the parsing process to make them more easily understood by LLMs. It's a big step towards LlamaIndex's mission of connecting your data to LLMs.

[HadiAlhassan-bxeh]

What about using an llm other than OpenAi?

I am trying to use cohere's command-R llm but i haven't found any documentation/ tutorial that shows this?

[daviscraig]

When you use the node parser with an instruction does it still maintain some sort of metadata about which original page it got the info from so you can check the source?

[nicolassuarez]

Outstanding! But how to get the metadata? Thanks!

[googleyoutubechannel]

Are you all still working on this or did you give up? I noticed llamaparse massively underdelivered on its promises initially but hoped y'all would keep working on it, have you given up or?

[danishsoomro]

So LLamaParse is just adding a prompt and asking GPT funciton to use embededdings.

[jaivalani]

How does Instructions parsing works, does it uses the LLMs ?

[shingenma]

From what I understood it uses an llm. Is it still safe to use it on sensitive documents? I'd like to propose it to a customer, and I need to answer this question.