news newest ask show jobs

Open Source @Github

fp.

Beyond Context Windows: Evaluating Long-Term Memory for AI Agents

https://twitter.com/AgentMemoryL/status/2101312784688726331

1•IreneAI•4m ago•0 comments

A Summer of AI Optimization

https://lemire.me/blog/2026/09/22/a-summer-of-ai-optimization/

1•chmaynard•7m ago•0 comments

Be the first human to review your own code

https://abhinavr.me/posts/be-the-first-human-to-review-your-own-code/

1•avr5500•9m ago•0 comments

AMD has now officially joined the trillion-dollar club

https://companiesmarketcap.com/amd/marketcap/

2•AbuAssar•19m ago•1 comments

Axon Training Pod (ATP) VR for training police officers

https://www.axon.com/products/axon-training-pod

2•nephihaha•20m ago•0 comments

GeoRisk Trace – An evidence-gated agent that abstains on unsupported risk paths

https://github.com/ViviLoveAI/GeoRiskAgent

1•ViviLoveAI•21m ago•0 comments

When Pixels Collide – The Story of Reddit's Place (2017)

https://web.archive.org/web/20170409173246/http://sudoscript.com/reddit-place/

1•pmlnr•21m ago•0 comments

Speechify for Cartesia Users

https://chromewebstore.google.com/detail/lisen-free-read-aloud-wit/jjglbebakkebkafjilbfnejmgpdphbhp

1•theaugustin•23m ago•0 comments

Windows Exploitation Techniques: Dangling COM Object Registrations

https://projectzero.google/2026/09/windows-dangling-com.html

1•mfrw•23m ago•0 comments

What you can build with TypeSafe AI Jev

https://vercel.com/i/jev-use-cases

1•flashbrew•28m ago•0 comments

Deploy Drunk

https://korshakov.com/posts/deploy-drunk

1•denysvitali•30m ago•0 comments

Genesis – The Bible verse that far-right Americans misinterpret to defend Israel

https://www.trtworld.com/article/5081f8ade0cf

3•timeoperator•30m ago•0 comments

Keep Using Your Intel Mac in 2025?

https://aslihan.live/posts/why-intel-mac-in2025/

2•aslihana•33m ago•2 comments

AMD briefly joins the $1T club

https://www.theregister.com/systems/2026/09/21/perpetual-underdog-amd-nips-at-nvidias-heels-as-it...

3•mdp2021•35m ago•0 comments

Can gzip be a language model?

https://nathan.rs/posts/gzip-lm/

1•networked•36m ago•0 comments

Daily Geography Adventures

https://maptap.gg/

1•yitchelle•38m ago•0 comments

Mirror Node Reconnaissance

https://sgaillabs.com

1•sgaillabs•41m ago•0 comments

How AI Creates a Capability Mirage

https://sloanreview.mit.edu/article/how-ai-creates-a-capability-mirage/

1•ventana•42m ago•1 comments

PyTorch Written in C

https://github.com/thevoxium/bare-lm

1•docuru•42m ago•0 comments

Maki, the efficient coder

https://maki.sh/

1•abc42•43m ago•1 comments

Orchestrat

https://orchestrat.dev

1•armanghev•46m ago•0 comments

A TypeScript Server Just Outran Rust's Hyper

https://geastack.com/blog-typescript-server-outran-hyper

1•k3030•49m ago•0 comments

Rust code faster than SOTA libraries by asking agents to make the code faster

https://minimaxir.com/2026/09/agentic-iteration/

2•sdoering•53m ago•0 comments

Pedagogy in the Times of AI [video]

https://www.youtube.com/watch?v=N2a1J0UPeL4

1•tarunvelli•53m ago•0 comments

Firmware Freedom

https://playtaurus.com/blog/firmware-freedom

9•arbayi•56m ago•3 comments

Lex Friedman Brings Back Strategery

https://daringfireball.net/linked/2026/09/21/strategery-friedman

1•Tomte•1h ago•0 comments

How HN: ScribeToAny – Free Audio and Video to Text Transcription

https://scribetoany.com/

1•hypercube67•1h ago•0 comments

Flomo: A two-person notes app from China that bought a product from ByteDance

https://www.sinosphere.app/p/flomo-two-people-one-text-box-and

1•Nyloc1•1h ago•0 comments

Why AI resembles a charismatic religious leader

https://theconversation.com/why-ai-resembles-a-charismatic-religious-leader-291831

7•billybuckwheat•1h ago•1 comments

I made ByeBlocked: hide blocked users

https://github.com/8ug8ird/ByeBlocked

1•8ug8irdchoom•1h ago•0 comments

Open in hackernews

I solved almost all of free problems on LeetCode using AI

https://old.reddit.com/r/leetcode/comments/1krg7by/i_solved_almost_all_of_free_problems_on_leetcode/

3•tevlon•1y ago

Comments

cratermoon•1y ago

I'm sure all the free problems from LeetCode are in the training sets, so I don't think this is much of a flex.

tevlon•1y ago

i was thinking the same. And i am sure it is in the training set, but i couldn't solve 10 hard problems. Also, most of the hard problems, the LLMs weren't able to solve in one shot. How do you explain this?

cratermoon•1y ago

There are many more correct answers to the easy problems than to the hard ones. Thus the LLM is more likely to hit upon the correct completion for them.

tevlon•1y ago

makes sense

whobre•1y ago

Alright, but how can this exercise help them pass an interview? The point of leetcode is to memorize the solutions so one can do well in an interview; yes, I agree it’s ridiculous.