Subscribe NOW

Enter your email address:

Text Message our CEO:

650-283-8008

or on twitter

Free Resources

Click Here to learn more

In The Media

OpenAI’s new assistants RAG and Raj Singh’s Take on Dev Tools LangChain and LlamaIndex

by Larry Chiang on April 17, 2024

Details on OpenAI’s new assistants RAG
Broaches well into vectorstore territory
Thoughts:
✅ Default chunk overlap of 50%, super interesting
✅ Metadata filtering, super interesting how this dips into vectorstore territory
✅ Unsure about what chunking method they use – 800 tokens character splitter? Recursive? semantic?
✅ Support for CSV retrieval, this is hard! Cool
✅ Hints towards alternative methods routing for Q&A types (summarization)
Retrieval set up:
* Query expansion (optimize user queries)
* Multi-query (one query > multiple searches)
* Hybrid search (keyword + semantic)
* Reranking
Fun to see what they included in version one to anticipate what GPT-5 will cover

 
 
BGy5XYVQ_normal.jpg spacer.png
Greg Kamradt
⁦‪@GregKamradt‬⁩
logo_twitter-1497383721365.png
spacer_464x1-1582829598167.png
Details on ⁦‪@OpenAI‬⁩‘s new assistants RAG

*Hard* creep into vectorstore territory

Thoughts:
* Default chunk overlap of 50%, super interesting
* Metadata filtering, super interesting how this dips into vectorstore territory
* Unsure about what chunking method they use – 800 tokens… pic.twitter.com/MaMtZ0h9Lr

 
4/17/24, 1:41 PM
 
 

image1.pngimage2.pngimage3.pngimage4.pngimage5.pngimage6.pngimage7.pngimage8.pngimage9.pngimage10.pngimage11.pngimage12.pngimage13.png

WordPressed from my personal iPhone, 650-283-8008, number that Steve Jobs texted me on

https://www.YouTube.com/watch?v=ejeIz4EhoJ0

Leave a Comment

Previous post:

Next post: