Hacker News: lalassu

New comment by lalassu in "Computer scientist Yann LeCun: 'Intelligence is about learning'"

lalassu — Fri, 02 Jan 2026 17:48:38 +0000

New comment by lalassu in "VectorChord 1.0: Vector Search on Postgres, 100x Faster Indexing than pgvector"

lalassu — Thu, 04 Dec 2025 13:32:56 +0000

The obvious question would be how does it compare to dedicated vector databases (qdrant, pinecone, chroma, etc..). What are the tradeoffs?

New comment by lalassu in "Prompt Injection via Poetry"

lalassu — Wed, 03 Dec 2025 18:14:31 +0000

Can someone explains why does that work?

I mean you can't social engineer a human using poetry? Why does it work for LLMs? Is it an artefact of their architecture or how these guardrails are implemented?

New comment by lalassu in "The era of AI slop cleanup has begun"

lalassu — Tue, 02 Dec 2025 16:21:55 +0000

I think many people are discovering again that the cost of software maintenance is much higher than the cost of building it the first time.

New comment by lalassu in "Octoverse: A new developer joins GitHub every second, AI leads TypeScript to #1"

lalassu — Tue, 02 Dec 2025 16:16:39 +0000

> 180 million-plus developers now work and build on GitHub

I did not realise the market for devtools is this huge. That explains why Cursor succeeded so much. That's 2% of humanity, no?

New comment by lalassu in "Mistral 3 family of models released"

lalassu — Tue, 02 Dec 2025 16:12:18 +0000

It's sad that they only compare to open weight models. I feel most users don't care much about OSS/not OSS. The value proposition is the quality of the generation for some use case.

I guess it says a bit about the state of European AI

New comment by lalassu in "Cara Hunter on the deepfake video that nearly ended her political career"

lalassu — Mon, 01 Dec 2025 19:56:52 +0000

I think the future of videos as a source of truth is bleak.

You can't trust anything that you can see in a video. And you can deny any video.

I think this will make an interesting post-truth world. What will happen when we can't almost prove anything. Eye witnesses are untrust worthy. Videos ditto. Cryptographic proof? Will politician sign their videos and messages with their public keys?

New comment by lalassu in "DeepSeek-v3.2: Pushing the frontier of open large language models [pdf]"

lalassu — Mon, 01 Dec 2025 19:53:48 +0000

Disclaimer: I did not test this yet.

I don't want to make big generalizations. But one thing I noticed with chinese models, especially Kimi, is that it does very well on benchmarks, but fails on vibe testing. It feels a little bit over-fitting to the benchmark and less to the use cases.

I hope it's not the same here.

New comment by lalassu in "Theranos founder Elizabeth Holmes' prison sentence keeps getting shorter"

lalassu — Mon, 01 Dec 2025 19:51:12 +0000

There is a very active account in twitter claiming to be her: https://x.com/ElizabethHolmes

I wonder what is the story behind it?