Hacker News: meffmadd

New comment by meffmadd in "Beijing is looking at curbing overseas access to China's top AI models"

meffmadd — Tue, 07 Jul 2026 15:04:06 +0000

This doesn’t hold up imo. Your phone analogy fits the description of a harness. That is the thing you interact with. The LLM is more like the processor in that phone. As long as it gets the work done, I will choose the cheapest/fastest because I can switch it out in a matter of seconds anyways.

New comment by meffmadd in "Why are large language models so terrible at video games?"

meffmadd — Mon, 01 Jun 2026 12:48:11 +0000

That is very true but I was surprised by how clear the “signal” was. Only Gemini really confidently solved all levels. But yeah the goal is now to include harder levels as well!

New comment by meffmadd in "Why are large language models so terrible at video games?"

meffmadd — Mon, 01 Jun 2026 12:09:02 +0000

Yes that was the post that inspired me to build this.

While I did implement a more comprehensive harness with path finding tools etc. the models themselves have improved significantly.

New comment by meffmadd in "Why are large language models so terrible at video games?"

meffmadd — Mon, 01 Jun 2026 11:13:35 +0000

I found LLMs to be surprisingly good at puzzle games like Baba Is You: https://meffmadd.github.io/samplesurium/posts/baba_is_agent/

New comment by meffmadd in "Crystals found inside wreckage from the first nuclear bomb test"

meffmadd — Mon, 18 May 2026 13:24:36 +0000

…or basically the plot of twin peaks

Can LLMs Play Baba Is You?

meffmadd — Fri, 15 May 2026 15:57:01 +0000

Article URL: https://meffmadd.github.io/samplesurium/posts/baba_is_agent/

Comments URL: https://news.ycombinator.com/item?id=48150223

Points: 5

# Comments: 0

New comment by meffmadd in "Claude Code's source code has been leaked via a map file in their NPM registry"

meffmadd — Tue, 31 Mar 2026 16:48:35 +0000

I think this view assumes no human will/should ever read the code. This is considered bad practice because someone else will not understand the code as well whether written by a human or agent. Unless 0% human oversight is needed anymore agents should still code like us.

New comment by meffmadd in "Palantir extends reach into British state as gets access to sensitive FCA data"

meffmadd — Sun, 22 Mar 2026 21:23:57 +0000

Genuine question to people more knowledgeable: Why are politicians/technocrats doing this?

Also generally speaking e.g. in relation to chat control and so on. Do they think this is what the people actually want because of lobbying or are they aware and believe they know better? Is it literally just corruption? Or are there actual benefits and we are just in the HN bubble where most people think its a bad idea?

New comment by meffmadd in "The First Fully General Computer Action Model"

meffmadd — Thu, 26 Feb 2026 08:49:57 +0000

I am amazed that the IDM is able to produce enough high quality annotations for the downstream FDM to work, even matching the ground truth contractor annotations!

New comment by meffmadd in "US orders diplomats to fight data sovereignty initiatives"

meffmadd — Wed, 25 Feb 2026 16:13:39 +0000

As an EU citizen I really hope we can gain some meaningful distance to the US asap. I hope my leaders feel the same. And if everything works out I think this will be great for the EU.

This is really some sort of diplomatic Streisand effect. If the US would not have been so aggressive and just string us along they could have continued to feed us their slop indefinitely without us noticing.

New comment by meffmadd in "Terence Tao, at 8 years old (1984) [pdf]"

meffmadd — Tue, 24 Feb 2026 12:45:16 +0000

I really wonder what motivates a seven year old to persistently work on that „one thing“ and not get distracted/bored. I guess he knew he was special?

New comment by meffmadd in "Qwen3.5: Towards Native Multimodal Agents"

meffmadd — Mon, 16 Feb 2026 14:26:43 +0000

Have you ever used an open model for a bit? I am not saying they are not benchmaxxing but they really do work well and are only getting better.

New comment by meffmadd in "Resizing windows on macOS Tahoe – the saga continues"

meffmadd — Fri, 13 Feb 2026 17:39:17 +0000

Here is what I don’t get tho: you have UX designers/engineers creating a new interface. What do you tell them? Just to do whatever? They probably spent months designing the new interface but why not fix this? They must have seen it is unusable…

New comment by meffmadd in "GLM-5: Targeting complex systems engineering and long-horizon agentic tasks"

meffmadd — Wed, 11 Feb 2026 17:24:59 +0000

It will be tough to run on our 4x H200 node… I wish they stayed around the 350B range. MLA will reduce KV cache usage but I don’t think the reduction will be significant enough.

Ask HN: What is the most complicated Algorithm you came up with yourself?

meffmadd — Fri, 06 Feb 2026 10:38:02 +0000

Hi HN, what was the most complex and sophisticated algorithm you came up with yourself while working on a problem?

Comments URL: https://news.ycombinator.com/item?id=46911249

Points: 3

# Comments: 9

New comment by meffmadd in "Show HN: 100 Million splats, a whole town, rendered in M2 MacBook Air"

meffmadd — Mon, 15 Dec 2025 20:47:13 +0000

From what I have heard „Bodycam“ uses scans of actual locations for its maps.

New comment by meffmadd in "Netflix to Acquire Warner Bros"

meffmadd — Fri, 05 Dec 2025 17:33:06 +0000

Yeah true, but also this is a bit like saying the lock screen of your phone should not become a "one stop shop" for all push notifications. I actually do not own an Apple TV but I just imaging you have a list of shows from different streaming providers on the "home screen" (like it is on my PS4). And on a technical level it is just an API you integrate with (same as push notifications), which helps UX.

New comment by meffmadd in "Netflix to Acquire Warner Bros"

meffmadd — Fri, 05 Dec 2025 13:18:40 +0000

Wow that is quite anti-consumer! Surely a monopoly on streaming will help them realize this. /s

How Attention Got So Efficient [GQA/MLA/DSA] [video]

meffmadd — Wed, 03 Dec 2025 19:20:40 +0000

Article URL: https://www.youtube.com/watch?v=Y-o545eYjXM

Comments URL: https://news.ycombinator.com/item?id=46138735

Points: 2

# Comments: 0

New comment by meffmadd in "NIST's DeepSeek "evaluation" is a hit piece"

meffmadd — Sun, 05 Oct 2025 21:33:40 +0000

As an EU citizen hosting LLMs for researchers and staff at the university I work at, this is hits home. Without Chinese models we could not do what we do right now. IMO, in the EU (and anywhere else for that matter), we should be grateful for the Chinese labs to release these models with such permissive licenses. Without them the options would be bleak. Sometimes we would get some non-frontier model „as a treat“ and if you would like something more powerful the US labs would suggest your country pay some hundred millions for an NVIDIA data center and the only EU option is to still pay them a license fee to host on your own hardware (afaik) while they protect all the expertise. Meanwhile DeepSeek has a week where they post the „secret sauce“ to host their model more efficiently, which helped open-source projects like vLLM (which we use) to improve.