Find answers from the community

Updated 3 months ago

Quick question: Is there a public

Quick question: Is there a public roadmap for llama-index as a framework? Or are we adding more features as we discover new use cases? I would be interested in the general future direction of the framework, what are we trying to achieve, let's say for 2024?

4 comments

LLogan M

The space evolves so quickly that it can be hard to specify exact features

I think in general, the overall goals are

More "deployment" focused examples and fixes (i.e. memory usage, load times, better storage, etc.)
Better support/examples for "large dataset" type problems
Improved open-source LLM support (specifically with prompts, structured data)
Code refactors (knowledge graph, tree index)
more/improved async support
batch support

SSan Nguyen

Yeah, I agree. This space is moving so fast. 😄. I hope next year, open source LLM will finally catch up or at least reliably functional on crucial features like tool calling, agent etc.

SSan Nguyen

@Logan M Also a bit related to this topic. Do you have any plan to release v1, what features would you expect before the v1 release?

LLogan M

I think for me, V1 means stability. So I think V1 is more about "vibes" than actual features haha

For example

Most LLMs work fine out of the box with core features
I want to better oragnize/package the never ending integrations
remove experimental or dead features from the core repo
deploying to aws, render, etc. and setting up a fastapi service is a breeze

All that would make me more comfortable labelling something as V1 🙂

Add a reply