|
when RAG was overkill
|
09-13-2026, 09:16 AM
My first guess would be this: Before rebuilding anything, I'd test one clean baseline with Phi and compare from there. Once you know the plain path works, then add the faster path back in piece by piece.
I'm happy to test one more variant if you want a second data point.
curious and caffeinated
|
|
« Next Oldest | Next Newest »
|
| Messages In This Thread |
|
when RAG was overkill - by southtinker - 09-05-2026, 07:36 AM
Re: when RAG was overkill - by kernelmike - 09-05-2026, 07:53 AM
Re: when RAG was overkill - by macropilot - 09-05-2026, 08:17 AM
Re: when RAG was overkill - by batchsize - 09-05-2026, 08:53 AM
Re: when RAG was overkill - by kernelmike - 09-14-2026, 09:34 AM
Re: when RAG was overkill - by llm_wren - 09-05-2026, 09:47 AM
Re: when RAG was overkill - by vramforge - 09-13-2026, 09:16 AM
Re: when RAG was overkill - by emberrunner - 09-08-2026, 09:56 AM
Re: when RAG was overkill - by deepseeked - 09-09-2026, 10:45 AM
Re: when RAG was overkill - by benchplease - 09-09-2026, 11:46 AM
Re: when RAG was overkill - by fastscope - 09-09-2026, 08:02 PM
Re: when RAG was overkill - by rocmstack25 - 09-12-2026, 08:29 PM
Re: when RAG was overkill - by quantnest - 09-13-2026, 01:12 AM
Re: when RAG was overkill - by macropoint - 09-15-2026, 10:14 AM
Re: when RAG was overkill - by oxideworks59 - 09-16-2026, 11:02 AM
|
