LlamaIndex Webinar: LLM Challenges in Production (w/ Mayo Oshin, AI Jason, Dylan from Starmorph)

No video

LlamaIndex Webinar: LLM Challenges in Production (w/ Mayo Oshin, AI Jason, Dylan from Starmorph)

Рет қаралды 3,201

Күн бұрын

In this webinar, we gather some leading AI educators / content creators to talk about LLM Challenges in Production.
Our cast includes:
- Mayo Oshin (@mayowaoshin, www.siennaiana...)
- AI Jason (@jasonzhou1993, / @aijasonz )
- Dylan / Starmorph (@StarmorphAI, / @starmorph )
We talk about the broad spectrum of LLM use cases and some of the biggest challenges that developers are facing in building/productionizing LLM apps, and how to solve them.
NOTE: apologies for the low resolution, we'll look into a better recording option than Zoom going forward.
00:00 - Introduction
04:12 - LLM Use Cases
14:00 - LLM Pain Points and Challenges (Overview)
28:50 - Retrieval Pain Points
39:00 - Tools that an AI Engineer should have
43:54 - Benchmarks
50:00 - Tools Wishlist
56:07 - Are LLMs Overhyped

Пікірлер: 4

@kevon217 11 ай бұрын

Another great webinar. Super helpful to hear from people on the front lines.

@geoffreyanderson4719 10 ай бұрын

Mayo cited as another major pain point, and so did others, that tools change so fast you can't keep up. Seems to me like a large LLM is the best thing known to mankind for rapid understanding of a broad range of subjects. So maybe all the tools need to be put into a neural QA system from the onset.

@geoffreyanderson4719 10 ай бұрын

Mayo wished for a tool to help manipulate parameters and variables to find optimal prompts etc. As a possible solution It turns out the FLAML kit for automating search space of input parameters of ML algos has been fused into their AUTOGEN kit for multiple agent apps.