Hacker News: snats

New comment by snats in "Show HN: Duplicate 3 layers in a 24B LLM, logical deduction .22→.76. No training"

snats — Thu, 19 Mar 2026 02:59:15 +0000

you can also have removed layers of models and keep the same score in benchmarks [1].

i feel that sometimes a lot of the layers might just be redundant and are not fully needed once a model is trained.

[1] https://snats.xyz/pages/articles/pruningg.html

Erebor: New bank by Palmer Luck [pdf]

snats — Sun, 22 Feb 2026 09:20:25 +0000

Article URL: https://www.occ.gov/topics/charters-and-licensing/digital-assets-licensing-applications/erebor-bank.pdf

Comments URL: https://news.ycombinator.com/item?id=47109595

Points: 2

# Comments: 1

Famous Disease

snats — Sun, 08 Feb 2026 20:32:22 +0000

Article URL: https://weblog.snats.xyz/posts/2025/12/12/

Comments URL: https://news.ycombinator.com/item?id=46938230

Points: 2

# Comments: 0

New comment by snats in "Ask HN: Share your personal website"

snats — Wed, 14 Jan 2026 20:57:02 +0000

personal website: snats.xyz weblog: weblog.snats.xyz

Famous Disease

snats — Mon, 22 Dec 2025 00:43:53 +0000

Article URL: https://weblog.snats.xyz/posts/2025/12/12/

Comments URL: https://news.ycombinator.com/item?id=46350162

Points: 3

# Comments: 0

Famous Disease

snats — Thu, 18 Dec 2025 15:45:57 +0000

Article URL: https://weblog.snats.xyz/posts/2025/12/12/

Comments URL: https://news.ycombinator.com/item?id=46314113

Points: 2

# Comments: 1

Flow-Lenia: Towards open-ended evolution in cellular automata

snats — Wed, 03 Dec 2025 18:16:20 +0000

Article URL: https://arxiv.org/abs/2212.07906

Comments URL: https://news.ycombinator.com/item?id=46137927

Points: 4

# Comments: 0

New comment by snats in "Just how resilient are large language models?"

snats — Sun, 28 Sep 2025 02:45:34 +0000

I also did a couple of experiments with pruning LLMs[1] using genetic algorithms and you can just keep removing a surprising amount of layers in big models before they start to have a stroke.

[1]https://snats.xyz/pages/articles/pruningg.html

Machines of Entretainment

snats — Thu, 25 Sep 2025 19:04:53 +0000

Article URL: https://weblog.snats.xyz/posts/2025/09/24/

Comments URL: https://news.ycombinator.com/item?id=45377387

Points: 1

# Comments: 0

New comment by snats in "I want an iPhone Mini-sized Android phone (2022)"

snats — Thu, 17 Jul 2025 00:50:06 +0000

i get it. i want one of those. the problem is that most cellphones are not actual cellphones, they are entertainment machines. they are a pocket tv / social media feed place. most usage for my normal friends is for that.

Edison a Dictator and a Pharmacologist

snats — Fri, 11 Jul 2025 19:07:44 +0000

Article URL: https://substack.com/home/post/p-167761857

Comments URL: https://news.ycombinator.com/item?id=44535840

Points: 2

# Comments: 0

New comment by snats in "Ask HN: What are you working on? (April 2025)"

snats — Sun, 27 Apr 2025 23:19:53 +0000

I am working on the https://moviemovie.club/about, it's a tiny website about film review.

It works like a run club, where you have to make a review first to see other people's reviews.

I am currently implementing watchlists, comments and a mural to make it feel a bit less lonely. Right now I like the UI but it feels to lonely.

New comment by snats in "Ask HN: Any insider takes on Yann LeCun's push against current architectures?"

snats — Fri, 14 Mar 2025 21:20:25 +0000

Not an insider but imo the work on diffusion language models like LLaDA is really exciting. It's pretty obvious that LLMs are good but they are pretty slow. And in a world where people want agents you want a lot of the time something that might not be that smart but is capable of going really fast + searches fast. You only need to solve search in a specific domain for most agents. You don't need to solve the entire knowledge of human history in a single set of weights

New comment by snats in "'A lot worse than expected': AI Pac-Man clones, reviewed"

snats — Fri, 14 Mar 2025 16:51:53 +0000

It's pretty funny to test in-distribution for AI models. But they fail horribly once you push them a bit[1].

I recently made LLMs play Minesweeper and ALL LLMs that I tested had a pretty bad win to loose ratio. Like the only model that won more than 3 times was R1 (mind you there were 50 games).

[1] https://snats.xyz/pages/articles/minesweeper_bench.html