LlamaIndex Webinar: Make RAG Production-Ready

Рет қаралды 17,274

11 ай бұрын

If you’re building LLM apps, you may already know that RAG is easy to setup but hard to iterate + make prod-ready.
In this webinar, we host a panel of experts to discuss the ways you can make RAG production-ready:
- Bob (co-founder/CEO, Weaviate)
- Max (co-founder/CEO, sid.ai)
- Tuana (dev rel, Haystack)

Пікірлер: 14

@petersolimine5175 11 ай бұрын

this is much needed content. Please keep it coming!

@JanaBergant123 11 ай бұрын

Great video, very informative :) I'd love to see more.

@user-ho3zw1jv7s 11 ай бұрын

keep them coming guys!

@awakenwithoutcoffee 2 ай бұрын

Great talk where I picked up some valuable insights: - "The idea chunking size really depends on your application usage e.g. an email summarizer would probably work better with smaller chunks (per email) while a translating RAG Agent could be chunked based on pages/chapters." Question: Seeing as this talk is almost 1 year old, what have been the most important additions or workflow changes since then ?

@kevon217 10 ай бұрын

Great discussion and tips.

@toluladeademisoye 9 ай бұрын

Great video

@AnonymousIguana 11 ай бұрын

Great content, very much apprieciated :). However, Max, these prominent "emmm" interrputions were quite distracting.

@sushantdahal2354 8 ай бұрын

I caught that too and definitely found it distracting as well but I dont think that was intentional from him

@GigaFro 7 ай бұрын

Didn't take away from the intelligent and well informed perspective he provided though! :)

@AnonymousIguana 7 ай бұрын

@@GigaFro agreed :)

@AnonymousIguana 11 ай бұрын

At 36:30 Bob says about solutions for high LLM utilization rate, but I don't quite get it. Could someone elaborate what he meant by this spark connector etc.?

@luromillion 10 ай бұрын

Spark connectors are used in large ingestion pipelines.. spark for big data, connector for Spark to DB (JDBC, Redshift, etc.) In context he is saying that people build some small PoC that works really well, but when they convert to production ready giant datasets, the costs / time / infra explode.

@Mr01jain 5 ай бұрын

51:10 Talking more from structured database perspective, how to handle metadata/ metadata filtering if my metadata is too large to be accommodated in the context? Think of a scenario where I am working with thousands of tables in a relational database.