Does fine tuning really improve anything above just pure RAG approaches for usee... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		aliljet 18 days ago \| parent \| context \| favorite \| on: Qwen3.5 Fine-Tuning Guide Does fine tuning really improve anything above just pure RAG approaches for usee cases that involve tons of direct document context?

44za12 18 days ago | [–]

Specialised models easily beat SOTA, case in point: https://nehmeailabs.com/flashcheck

NitpickLawyer 18 days ago | | [–]

Remember how the tab-next-action model from Cursor was all the rage ~2 years ago when they launched it? That was a fine-tune of a ~70b model (they kinda alluded to this in a podcast).

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact