punnerud

Yes

"""One of their better releases that seemingly focused more on quality and refinements than new-features.""" - Compared to a lot of others, this one have introduced more bugs than I have ever experienced in more than 10 years.

I have stopped submitting bugs because of the tedious bug reporting app, only submit really bad ones. Can't they give some helping functions to finding the correct categories based on problem description?

The program this fits into was not started by us nor was it proposed by a Large Language Model. (…) We took their work as a starting point, using Large Language Models to push their program to completion.

Thats how most people use LLMs? If I was back in my student days working in Navier-Stokes, I guess I would also punch at blow ups. The number of students doing this at the same time, posting open efforts to GitHub then retraining of the models. If there is solutions to the problem, it’s a real possibility that it was not a result of this effort?

Using a large amount of tokens is not a good augment that it’s not likely others have done the same. Good questions is the difference between $10 and $10M in token usage to solve a problem.

The demo use WASM/browser when available, but also supports running only on ESP32. Figured out I could offload and give the users a better experience when available.

Thanks for noticing the Tailscale demo. Created a Tailscale-Rust client that supports NAT-traversial, encryption etc. and manage to run on ESP32 (together with the local LLM). Just the Tailscale part can be found here: https://github.com/punnerud/tailscale-mpe-rust

A trick to fit language models of this size it to remove the word to embeddings from the NN, and have a database to look up a dictionary of words and their embeddings. This way the LLL only need the "core" and you do the reverse for loopup again (kind of Text->RAG->LLM->RAG->Text). Have an example here: https://punnerud.github.io/pyspell/

The example have a limited language of around 1000 words, but make it possible to do Python (like) programming with LLM on an ESP32.

My prediction is that HuggingFace will take up the challenge with OpenRouter (now Stripe) and ollama.com (Facebook)

Already using it as the primary DB in multiple off my apps and dev pipelines. I don't trust my own "slop", but the all the tests from SQLite, Python, PHP etc agains the DB

You can think further, you can build a no-server server; Example let Nginx be running, then forward requests to the database, running the code, query data and responding then shutting down again. Need the service to do something regularly, just trigger that part of the database through crontab.

And yes you can let authorized users add custom code, make it run in the database like a micro-micro-vm and do stuff on triggers etc in the database as well.

This makes it even more powerful, because users can switch code during runtime and you can have multiple users "programming" against the same database at the same time working against the same data.

I build a SQLite clone in Rust with this analogy in mind, and unlike SQLite it support multiple writers by giving them separate segment in a file. Calles the programming "language" PySpell, it converts Python into AST then into RUST with a lot of cool features you can enable when the database and programming language is living in the same place.

http://github.com/punnerud/mpedb

Added, thanks

Checked and this 100% compatible SQLite3 database (also C-API) did not contain the bug: https://github.com/punnerud/mpedb (Disclaimer: My own project)

And this statement is wrong in the article: “ Because SQLite is a single-writer database with serialisable transactions, our transaction history was completely linear and deterministic. (This wouldn’t be true in a multi-writer database like Postgres or MySQL.)”

Actually possible in mpedb to replay multi-writer, and actually better than SQLite3. Try to reply now() in a statement, that is not deterministic in SQLite but is in MPEdb.

Wow, had to check some of his other repos; his the one behind dump1090

What’s in that “Search tool”, could actually be the same solution presented in a different way

What a great idea! Anyone that want to join up in Oslo / Norway?

I can use the work office, have more then enough coffee, tea etc. with space to around 150 people.

https://forms.gle/2pDEkQTYihQ4txK29

(Copy pasted the form. Why change a winning team)

Sorry, that’s correct. The other one is an internal one not released yet that also have a full Tailscale server written in Rust.

I made this for ESP32 in Rust giving full Tailscale experience with NAT-traversal https://github.com/punnerud/mpe-tunnel

Not often I see companies referring to HN, thanks “ We quietly released the open-source Codex Security CLI, but Hacker News found it before we had a chance to share it here…”

https://x.com/openai/status/2082263717916586117?s=46&t=mnfnj...

I was thinking image analysis from satellite images, but of course start with what’s annotated (OSM)

No you don’t need an external one, it’s actually possible to sample parts of high bandwidth using a lower bandwidth because of the small buffer in most SDR. The problem is that you can’t freeze the buffer so have to gather multiple “sweeps” get parts of ADSB stitched together.

Maybe I post the Rust code after some cleanup. Now I am busy build a new database from scratch to compete with them all https://github.com/punnerud/mpedb

Actually possible to receive and decode ADS-B on ESP32, but MLAT is failing (tried hard but not found a trick yet).

Biggest problem isn’t that it’s not powerful enough, but the USB port bandwidth don’t match most SDR.

Using some clever tricks you can receive parts of the message on a slower USB and be able to decode ADS-B

Nice job! When I zoom in on Google earth you can easily see if it’s actually a golf course or not, would been interesting to use this to draw up the areas with polygons and some more details.

Then you could start providing a lot more details about each one automatically.

Tested with incognito, works fine without login using the link

Why not just show the real link to ?

You know that a lot of acquisitions is equity hire. How do you get engineers to be excited about fixing something low level in Facebook, a hard task; you don't. Better to wrap those problems in green field projects and have engineers excited, and adopt that tech back into Facebook. Google done this forever (DeepMind). Seems like the work of a genius?

Zuckerbergs comment from 2010: «We have not once bought a company for the company. We buy companies to get excellent people.»

Thanks for the feedback, will fix the local storage part

Thank you, tried to make it intuitive and fun to play

Can we get a super fast way to update DNS with lower cache, so dynamic IP updated through API works. This is one of the limiting factors.

Cloudflare works but the cache give downtime after every IP-switch

If you have enough examples you can train an AI on your preferences, then use that distilled AI as a unit test. Don’t combine multiple into one AI. If they don’t agree you want it to fail so you can decide and retrain the tests.

Why not have 5,6 rings at different levels and do it live in 3D?

Luckily I have a project with 200k lines of code, generating substantial revenue. Asked Fabel to run continuously through all features with up to 50 parallel agents, fix bugs, log improvements etc (just to max out tokens before the week reset)