_cs2017_'s favorites | Hacker News

1.		When AI writes the software, who verifies it? (leodemoura.github.io)
		305 points by todsacerdoti 16 days ago \| 299 comments
2.		Don't become an engineering manager (manager.dev)
		396 points by flail 16 days ago \| 269 comments
3.		Why No AI Games? (franklantz.substack.com)
		72 points by pavel_lishin 16 days ago \| 85 comments
4.		Semantic ablation: Why AI writing is generic and boring (theregister.com)
		285 points by benji8000 30 days ago \| 204 comments
5.		OpenAI should build Slack (latent.space)
		252 points by swyx 33 days ago \| 327 comments
6.		uBlock filter list to hide all YouTube Shorts (github.com/i5heu)
		1172 points by i5heu 33 days ago \| 344 comments
7.		Cowork: Claude Code for the rest of your work (claude.com)
		1298 points by adocomplete 66 days ago \| 565 comments
8.		TimeCapsuleLLM: LLM trained only on data from 1800-1875 (github.com/haykgrigo3)
		737 points by admp 66 days ago \| 314 comments
9.		Insights into Claude Opus 4.5 from Pokémon (lesswrong.com)
		123 points by surprisetalk 72 days ago \| 23 comments
10.		Play Aardwolf MUD (aardwolf.com)
		182 points by caminanteblanco 71 days ago \| 102 comments
11.		Agent design is still hard (pocoo.org)
		426 points by the_mitsuhiko 3 months ago \| 258 comments
12.		The Continual Learning Problem (jessylin.com)
		102 points by Bogdanp 4 months ago \| 8 comments
13.		Kafka is Fast – I'll use Postgres (topicpartition.io)
		561 points by enether 4 months ago \| 401 comments
14.		BERT is just a single text diffusion step (nathan.rs)
		455 points by nathan-barry 5 months ago \| 110 comments
15.		SWE-Grep and SWE-Grep-Mini: RL for Fast Multi-Turn Context Retrieval (cognition.ai)
		97 points by meetpateltech 5 months ago \| 31 comments
16.		Writing an LLM from scratch, part 22 – training our LLM (gilesthomas.com)
		254 points by gpjt 5 months ago \| 10 comments
17.		Show HN: I invented a new generative model and got accepted to ICLR (discrete-distribution-networks.github.io)
		656 points by diyer22 5 months ago \| 91 comments
18.		A small number of samples can poison LLMs of any size (anthropic.com)
		1202 points by meetpateltech 5 months ago \| 439 comments
19.		Reasoning LLMs are wandering solution explorers (arxiv.org)
		90 points by Surreal4434 5 months ago \| 98 comments
20.		Building the heap: racking 30 petabytes of hard drives for pretraining (si.inc)
		412 points by nee1r 5 months ago \| 274 comments
21.		We reverse-engineered Flash Attention 4 (modal.com)
		134 points by birdculture 5 months ago \| 48 comments
22.		Claude’s memory architecture is the opposite of ChatGPT’s (shloked.com)
		448 points by shloked 6 months ago \| 236 comments
23.		Le Chat: Custom MCP Connectors, Memories (mistral.ai)
		398 points by Anon84 6 months ago \| 165 comments
24.		A PM's Guide to AI Agent Architecture (productcurious.com)
		208 points by umangsehgal93 6 months ago \| 62 comments
25.		Physics of badminton's new killer spin serve (arstechnica.com)
		119 points by amichail 7 months ago \| 16 comments
26.		Dispelling misconceptions about RLHF (aerial-toothpaste-34a.notion.site)
		120 points by fpgaminer 7 months ago \| 32 comments
27.		Diffusion language models are super data learners (jinjieni.notion.site)
		218 points by babelfish 7 months ago \| 16 comments
28.		My Lethal Trifecta talk at the Bay Area AI Security Meetup (simonwillison.net)
		430 points by vismit2000 7 months ago \| 115 comments
29.		How attention sinks keep language models stable (hanlab.mit.edu)
		219 points by pr337h4m 7 months ago \| 36 comments
30.		Gemini 2.5 Deep Think (blog.google)
		461 points by meetpateltech 7 months ago \| 249 comments
		More