frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Ask HN: When is fine-tuning a small LLM worth it?

4•kooldeep7•56m ago
I'd be interested in hearing about your experiences. What kind of task did you use it for, what model did you train, and what were the results?

Feel free to share examples of what you've tried.

Comments

minimaxir•41m ago
It's worth it if a) you have a decent sample size of data for your problem and b) you have a cost-effective infra to host it.

Notably the latter is more of the bottleneck, particularly with the price race-to-zero with models such as GPT-6 Luna.

verdverm•28m ago
There are allegedly multi-tenant LoRA offerings in the works which would change the hosting-pricing constraints considerably. Keeping my fingers crossed they materialize
carlos_rpn•15m ago
Maybe when you want to learn the process for when/if you ever need to fine-tune a larger model?

It's faster to make iterate when you're toying around with a 1B model than a 27B one.

The UV index is not the warm sensation of sunlight on bare skin

https://blog.asciitweezers.com/the-uv-index-is-not-the-warm-sensation-of-sunlight-on-bare-skin/
3•evakhoury•1m ago•0 comments

Relaticle, a self-hosted CRM where AI updates data only after your approval

https://relaticle.com
2•manukminasyan•1m ago•0 comments

No Sloptober

https://no-sloptober.com/
2•jeremiahlee•2m ago•0 comments

Show HN: NetM8 OS

https://netm8.com/
1•goodknightleo•4m ago•0 comments

Show HN: Typed polyglot compiler for CLI/API/MCP generation

https://github.com/morloc-project/morloc
1•arendsee•5m ago•0 comments

Tinyjs – desktop apps for macOS, Windows, and Linux in ~6 MB

https://tinyjs.app
1•josephscott•7m ago•0 comments

I beat all the popular harnesses on FrontierHarness Eval

https://github.com/tontinton/maki
2•tontinton•7m ago•0 comments

Microsoft killed FoxPro in 2007. Anyway, here's FoxPro revived

https://foxscript.org/
1•boredjohnny•7m ago•0 comments

ArchiveBox v0.9 released with new native apps and 50 new plugins

https://docs.sweeting.me/s/archivebox-v0.9-announcement
1•nikisweeting•8m ago•1 comments

Kotlin 2026: Layoffs, AI, Google – Is the Golden Age Over? Jake Wharton Explains [video]

https://www.youtube.com/watch?v=9WjPdgOK-68
1•cstever•11m ago•0 comments

Why AI Can't Replace You with Cory Doctorow [YouTube] [video]

https://www.youtube.com/watch?v=VU3gABvwZCM
2•cstever•12m ago•0 comments

Open-Weight AI Models Seize Token Lead, but Proprietary Still Make the Money

https://techstrong.ai/generative-ai/open-weight-ai-models-seize-token-lead-but-proprietary-system...
1•CrankyBear•12m ago•0 comments

Hive at v5: The Swarm Grew Up, and It Is Learning to Leave the Dashboard

https://hivecommons.substack.com/p/hive-at-v5-the-swarm-grew-up-and
1•jcastro•12m ago•0 comments

DoorDash reaches $131.5M settlement with NYC for failing to pay workers

https://documentedny.com/2026/09/22/mamdani-delivery-workers-doordash-labor-settlement/
2•paimapi•12m ago•0 comments

Apple's new watches are always listening

https://theconversation.com/apples-new-watches-are-always-listening-292117
2•sajithdilshan•13m ago•0 comments

Enough Reason to Act: AI safety needs policy not dismissal

https://substack.norabble.com/p/enough-reason-to-act
1•nedruod•15m ago•0 comments

Show HN: Video clip maker/editor right in the browser (supports YouTube)

https://heahy.com/studio
1•ignasheahy•17m ago•0 comments

Show HN: Sandopolis – A portable multi-system Sega emulator

1•habedi0•20m ago•0 comments

Show HN: JevEval, evals using Jev-as-a-judge

https://deepeval.com/blog/introducing-jev-as-a-judge
1•jeffreyip•20m ago•0 comments

Datapages v0.10.0 Beta Release

https://github.com/romshark/datapages/releases/tag/v0.10.0
4•romshark•21m ago•2 comments

Kremlin-backed forgery scheme moved $6.9B through global banks

https://www.ft.com/content/3848a4ad-1426-4352-ba86-ef1279c4ffd1
3•tagyro•25m ago•0 comments

Common denominators of organisms with exceptional longevity

https://www.aging-us.com/article/206419/text
1•tima101•27m ago•0 comments

Motorola Signature 27 officially gets a North American launch

https://9to5google.com/2026/09/22/motorola-signature-27-north-america-launch-confirmed/
1•Cider9986•30m ago•0 comments

LLM Ass Bench

https://www.assbench.com/
5•fragmede•33m ago•3 comments

Show HN: ToolHub – Hierarchical nav for 30 MCP tools without context bloat

https://github.com/Talos-popcorn/toolhub
1•soidkeo•33m ago•0 comments

New trends in global card fraud

https://stripe.com/blog/new-trends-in-global-card-fraud-how-3d-secure-and-regional-mandates-are-a...
2•rndsignals•35m ago•0 comments

Show HN: With generators/DI, LazyPromise has become a tiny alternative to Effect

https://lazypromise.com/
1•ivan7237d•35m ago•0 comments

Making Concurrent Hardware Verification Sequential (2025)

https://dl.acm.org/doi/full/10.1145/3729331
1•imakwana•35m ago•0 comments

BLUF (Communication)

https://en.wikipedia.org/wiki/BLUF_(communication)
3•gurjeet•37m ago•0 comments

Longtime SUSE staff asked if they'd opt for 'voluntary separation'

https://www.theregister.com/software/2026/09/22/longtime-suse-staff-asked-if-theyd-opt-for-volunt...
2•romaniitedomum•38m ago•0 comments