frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Ask HN: Does the tech industry require deterministic ML training?

4•Swapnopam•5h ago
Hey, I was wondering whether the tech industries require deterministic ML training or is non-determinism acceptable? As in, I know in case of LLMs, small float differences while training are generally tolerated as the model converges to similar quality regardless. But in more specific domains like medical device certification or autonomous vehicle safety validation or financial model auditing, I feel like whether two training runs producing bit identical models would be critical whether two training runs producing bit-identical models is critical, on the same hardware, let alone across heterogeneous hardware. I have read how in certain cases like HFT or Quant, people would just use integers to make deterministic actions, wondering if its the same case here too?

Claude Code getting "API Error: 529 Overloaded"

3•croemer•1h ago•2 comments

Ask HN: Why is every company encorporating AI everywhere?

7•vanessa1211•2h ago•6 comments

Opus Cost = Minimum Wage in South Africa

4•bridgettegraham•4h ago•1 comments

Ask HN: Where do those LLM-generated outreach emails come from?

3•nicbou•4h ago•1 comments

Ask HN: Any New Computer Ideas?

4•robalni•6h ago•3 comments

Ask HN: Does the tech industry require deterministic ML training?

4•Swapnopam•5h ago•0 comments

CstWeave – On-device, privacy-first Chrome extension for tab region GIFs

2•fujidevv•5h ago•0 comments

Ask HN: What are the most promising RL fields for a new master student?

33•zecice•23h ago•14 comments

Ask HN: Are we having a substantial increase of "Show HN" since LLM aided coding

7•mdp2021•6h ago•4 comments

Ask HN: Is Zig worth it over Rust in 2026?

5•soywod•12h ago•4 comments

Ask HN: What livestream do you keep open in a tab?

50•missmoss•1d ago•18 comments

Ask HN: How do you figure out what you're passionate about?

6•charliebwrites•15h ago•10 comments

Ask HN: Protecting your Sites/Services from Unwanted Traffic?

12•prologic•1d ago•8 comments

Ask HN: What's the best hands-on path to learn ML inference infrastructure?

7•censor5•16h ago•3 comments

Ask HN: Are you using Rust on embedded devices yet? If not, why?

9•mempirate•1d ago•4 comments

BSD seq(1) on Linux

3•1vuio0pswjnm7•17h ago•0 comments

Robotic Adaptation Suffers from Data Collection Bottleneck

2•aunmesh•18h ago•0 comments

Ask HN: Anyone suffering from the tyranny of the Waiting Equation?

2•akersten•18h ago•2 comments

Ask HN: Do you delete your browsing history?

4•tempestn•19h ago•7 comments

Tell HN: Namecheap gave my account to an unverified third party

493•Thrashed•3d ago•175 comments

Ask HN: Which Jabber clients support SCRAM+ and XEP-0474

12•Bender•1d ago•0 comments

Ask HN: Is an in-page speed reader useful?

2•jvbotelho•20h ago•5 comments

Connection-Agnostic Presence Tracking for Stateless Distributed Back Ends

5•duckydude20•1d ago•0 comments

Ask HN: I spent personal money on game dev. How to turn stalled project around?

2•pixelbyindex•21h ago•9 comments

Timed YouTube video transcripts in .txt and .pdf

7•cristyg0101•1d ago•4 comments

Ask HN: Graal VM Relevance

5•sid_op•23h ago•0 comments

What happens behind the scenes when we change effort for same LLM models?

13•tbharath•1d ago•8 comments

Ask HN: Something fishy is going on with Google feed on my Android

7•motbus3•1d ago•8 comments

Who's making money from using/exploiting AI?

11•jonthepirate•1d ago•5 comments

Ask HN: HotPin – lossless 120B MoE inference on 24GB RAM (CPU, 50 loc)

10•LozzKappa•1d ago•0 comments