Hacker News: dust42

New comment by dust42 in "DeepSeek V4—almost on the frontier"

dust42 — Sat, 02 May 2026 17:53:55 +0000

For many models the performance of llama.cpp on Mac is 20-40% lower than MLX. Did you try MLX? At least on HF there are MLX 2-bit quants. Unfortunately I have only 64GB, so I can't test it.

New comment by dust42 in "DeepSeek V4 – almost on the frontier"

dust42 — Sat, 02 May 2026 17:45:45 +0000

The output of any LLM is always 100% hallucination by principle. On top of that, most benchmarks are at best an approximation of LLM quality. Your use case decides which one to use. That said, I haven't tested v4 yet but the old 3.2 is still a decent model. And concerning use cases, I had coding problems that Opus couldn't solve but a local 35B model did.

All the talk about frontier and SOTA is do dig deeper and deeper into the pockets of VCs and finally do an IPO.

New comment by dust42 in "A Periodic Map of Cheese"

dust42 — Tue, 21 Apr 2026 18:20:14 +0000

That website is so low effort that 2s is actually long to figure it out. Very sure that it is robot upvoted.

Edit: I live in the cheese triangle, France - Switzerland - Italy.

New comment by dust42 in "Tindie store under "scheduled maintenance" for days"

dust42 — Tue, 21 Apr 2026 17:48:32 +0000

Tinder for indie (hardware) devs and their customers. I.e. a webshop for indie devs who sell small series of niche hardware.

New comment by dust42 in "Recommended GPU Repairshop in Europe (Germany)"

dust42 — Tue, 21 Apr 2026 17:30:30 +0000

Ask on mikrocontroller.net. There will definitely be people who know who in Germany is doing it or maybe even offer to do it for you. One of the last old fashioned forums on the net. German speaking but nowadays that shouldn't be a problem.

New comment by dust42 in "Qwen3.6-35B-A3B: Agentic coding power, now open to all"

dust42 — Fri, 17 Apr 2026 12:57:55 +0000

I am using it with pi agent and I have stopped renting tokens. Much better for me than Claude Code, on M1 Max 64GB. This model with oMLX is at 16k context PP 919.9 tok/s and TG 54.7 tok/s. You have to manage the context but the better you manage context the more focused the output is. I use it without thinking.

New comment by dust42 in "Everything we like is a psyop?"

dust42 — Fri, 17 Apr 2026 06:41:58 +0000

I once worked for a company that bought a spot in the evening news (french TF1). It worked that way: a french minister was visiting a fair and coming to stop in front of the booth and getting a product demo. And that ended up in the evening news. Since that time I kind of watch the news with different eyes.

New comment by dust42 in "Germany suspends military approval for long stays abroad for men under 45"

dust42 — Thu, 16 Apr 2026 08:58:10 +0000

The law is still there though and thus its application can be resumed at any time without need of parliament approval.

New comment by dust42 in "Google Gemma 4 Runs Natively on iPhone with Full Offline AI Inference"

dust42 — Wed, 15 Apr 2026 19:24:34 +0000

There is a project on github named ANEMLL. Was discussed here a month ago, running LLMs on iPhone - https://news.ycombinator.com/item?id=47490070

New comment by dust42 in "Elevated errors on Claude.ai, API, Claude Code"

dust42 — Wed, 15 Apr 2026 14:59:01 +0000

Well, at least we know by now that Mythos is a mythos.

New comment by dust42 in "Reallocating $100/Month Claude Code Spend to Zed and OpenRouter"

dust42 — Thu, 09 Apr 2026 12:24:27 +0000

I really love it. The simplicity is key. The first play project I made with it was a public transport map with GTFS data - click on a stop and get the routes and the timetables for the stop and the surrounding ones. I used Qwen3.5-35B on Mac M1 Max with oMLX. It wrote 98% of the code with very little interaction from me. And very useful is the /tree feature to go back in history when the model is on a wrong track or my instructions where not good enough. I usually work in a two path approach: first let the model explore what it needs to fulfill the task and write it into CONTEXT.md (or any other name to your liking). Then restart the session with the CONTEXT.md. That way you are always nicely operating in 5-15k context, i.e. all is very fast. Create an account for pi (or docker) and make sure it can't walk into other directories - it has bash access. Add the browser-tools to the skills and load them when useful: https://github.com/badlogic/pi-skills

No need for database MCP, I use postgres and tell it to use psql.

Occasionally I use prettier to remove indentation - the LLM makes a lot less edit errors that way. Just add the indent back before you commit. Or tell pi to do it.

New comment by dust42 in "I've sold out"

dust42 — Wed, 08 Apr 2026 12:24:31 +0000

Context: Mario Zechner is the creator of the pi coding harness which powers OpenClaw. OpenClaw is made by Peter Steinberger, a friend of Mario Zechner. Armin is another friend who made public that OpenClaw is based on pi.

Pi itself is a minimalist coding harness with a tiny 1500 token system message and only read, edit and bash as tools. I only discovered it a few weeks ago and it is surprisingly powerful with a local Qwen3.5-35B - especially as it allows to keep the context low.

Mario's blog posts are not easily digestible (imho) until you have read a few of them but they have plenty of profound thinking. His blog is for me the first one in years where I have spent an hour to read several posts.

Mario is deeply rooted in the OSS system and basically that is what he is talking about here in this post. That said, I have no idea what earendil is doing, except that it is based on pi.

Edit: My personal take - "I've sold out" is very much Austrian style because actually it is the opposite. To quote one thing from the post:

"Then Miguel and Nat approached us. Long story short: we sold RoboVM to Xamarin. A short while later Xamarin closed-sourced our open-source RoboVM core, quickly followed by Xamarin selling to Microsoft. Then Microsoft shut down RoboVM immediately.

While there was some monetary gain, everything about this fucking sucked."

So Mario did a lot of vetting to hopefully avoid this from happening again.

New comment by dust42 in "Artemis II crew see first glimpse of far side of Moon [video]"

dust42 — Tue, 07 Apr 2026 15:07:39 +0000

The social media team.

New comment by dust42 in "Artemis II crew see first glimpse of far side of Moon [video]"

dust42 — Tue, 07 Apr 2026 15:06:52 +0000

Well, mathematically full moon is only a infinitesimally small split of a second. When I made the comment it was about 12% waxing crescent thus 88% in the dark. And actually darker than a full moon because the earth does not light up the far side.

New comment by dust42 in "Show HN: Ghost Pepper – Local hold-to-talk speech-to-text for macOS"

dust42 — Tue, 07 Apr 2026 07:13:06 +0000

When you activate it you agree that your voice input is sent to Apple. As far as I understand this project runs fully locally. Up to you to decide for whatever suits your needs best.

New comment by dust42 in "Artemis II crew see first glimpse of far side of Moon [video]"

dust42 — Sun, 05 Apr 2026 15:52:22 +0000

Yes, and right now is full moon, thus the far side is only illuminated by stars.

New comment by dust42 in "Artemis II crew see first glimpse of far side of Moon [video]"

dust42 — Sun, 05 Apr 2026 15:49:14 +0000

> It’s approximately the dark side when the moon is full, which happened two days ago.

Who downvotes that? It is true.

Edit: maybe you can illuminate why you downvote?

New comment by dust42 in "We replaced RAG with a virtual filesystem for our AI documentation assistant"

dust42 — Fri, 03 Apr 2026 18:33:08 +0000

If grep and ls do the trick, then sure you don't need RAG/embeddings. But you also don't need an LLM: a full text search in a database will be a lot more performant, faster and use less resources.

New comment by dust42 in "Nitrile and latex gloves may cause overestimation of microplastics"

dust42 — Sun, 29 Mar 2026 13:08:31 +0000

So basically the gloves that kitchen staff now must wear means we get an extra dose of micro plastics? Yikes.

New comment by dust42 in "Hold on to Your Hardware"

dust42 — Fri, 27 Mar 2026 11:11:13 +0000

Just to mention one thing, helium -which is a necessity for chip production- is a byproduct of LNG production. And 20% of that is just gone (Qatar) and the question is how long it will take to get that back. So not only a chip shortage because of AI buying chips in huge volumes but also because production will be hampered.

Tongue in cheek: we urgently need fusion power plants. For the AI and the helium.