aichannel

- Ran the same prompt on DeepSeek R1 and O1
- DeepSeek gave me a better response and better code

🇨🇳

Hi! My name is Daniel. I am building bountycaster.xyz

Ex Phantom, 0x Project engineer.

From Italy originally, currently live in NYC

someone on twitter suggested using R1 instead of RAG and now I'm testing it

Multiplying onchain economies. Working on interfaces /comint and intelligence @tenfold
Past: AI and crypto rubinovitz.com

RAG can be used with R1, isn't it mainly a concept around retrieval?

I don’t do RAG anymore, get 10x the result just spinning up a pipeline and feed all content to Deepseek. And yes it scales to over 10K docs. 

RAG is anti pattern.

homanp

https://x.com/pelaseyed/status/1882471632129994914/?s=12&rw_tt_thread=True

huge context windows have tradeoffs like aren’t always cost feasible (observation) and obscured prompts (opinion)

until the next big breakthrough this is mostly more conjecture imo