🚨 Major Scoop:
OpenAI has just expanded internal testing of GPT Astra, codenamed "mozaik-alpha-fdm", signifying release might be near
AND of course like previous times, we have the FIRST EVER public outputs of it for y'all😉
Both outputs are zero-shot on Max effort. Frontend *might* have finally been fixed, and the model gives a lot of attention to details
@bcherny there's this controversial "Setup" hook which ought to be working yet it's not there in the documentation and not working on win / mac environments either.
yet 2.1.10 release mentioned it in the changeset. can you shed some light on it?!
Instead of using a global fixed chunk size for RAG, try splitting based on the semantics of the text ✂️💡
@GregKamradt proposed a super simple method to split long documents based on embedding similarity between sentences, with an auto-tuned threshold.
If there’s sufficiently low similarity between two sentences, then there’s a break and you can split!
We’re excited to implement this as a LlamaPack 🦙📦 using @llama_index abstractions, so you can use this to power advanced RAG pipelines.
Semantic Chunking LlamaPack: https://t.co/o1cYrq1KxR
Notebook: https://t.co/QxeKncUYZZ
System Design Blueprint: The Ultimate Guide.
Hope this checklist is useful to guide your discussions during the interview process.
This briefly touches on:
- LB
- Gateway
- Communication
- CDN
- Database
- Cache
- MQ
- ID Generation
- Scalability
- Availability
- More
A picture of Ukrainian resilience - growing cabbage and pumpkins around a destroyed Russian tank, the tank itself used as a compost pit. There’s even a flower bed next to the old turret! Having been through so much, people are finding hope in these daily acts of care for the land