feat: Add support for llamaIndex in evaluation

[suekou]

Added type checks for llamaIndex LLMs and embeddings in the evaluate function.

[jjmachan]

thanks a lot for sending a PR 🙂 - just once change though and we'll merge it in

[jjmachan]

llamaindex is actually not part of the core dependency. It's best if we keep it optional

[jjmachan]

if you move it to all that would be helpful, or maybe even a new group for llamaindex?

[suekou]

Thanks for the feedback! I've updated the PR to make llama_index optional by moving it to the [all] group.

[jjmachan]

Thanks a lot @suekou 🙂 ❤️
merging this in

[jjmachan]

btw would love to get your feedback on this docs too if you get the chance https://docs.ragas.io/en/latest/howtos/integrations/_llamaindex/

[suekou]

@jjmachan I found this to be very clear and understandable!
I noticed the URL on the page was outdated, so I submitted a PR to update it! PR

[suekou]

Looking at it again, I think it might also be worth mentioning the differences between evaluate() from from ragas import evaluate and from ragas.integrations.llama_index import evaluate. For instance, using evaluate from ragas allows for more detailed configuration options, such as in_ci and token_usage_parser, which enable a more customized evaluation.

[jjmachan]

hey @suekou thats a good point but right now evaluate in ragas.integrations.llamaindex has the same features as the top level evaluate right?

the idea was to keep both of the feature compatible since behind the scene all we are doing is running the query_engine for you

ps: Thanks a lot for the other PR too 🙂 ❤️