dhbradshaw

The generated latex pdf!

I think evals are the key here. If your fable system fails them, it's a bad system for your use case. If not, compare cost with other systems that also succeed.

To me this doesn't seem like a disaster but just the kind of thing that happens as you role out a service and expose it to new challenges.

Presumably they haven't had the chance to do a lot of flood training but now they have that chance.

The huge advantage they have over people in general is that ideally if they figure this out then it will stay figured out. Then they can slowly role out and watch for the next hitches from new situations.

I hadn't even thought about these -- inspecting and deleting from the TUI makes a lot of sense.

Thank you!

Good question. I haven't done any work yet to make this fast so it might not be too quick...

My kids and I have been playing this for about 20 years. It's worked on Linux, Mac and Windows and has never stopped working.

In the meantime so many other favorite games have disappeared or become obsolete.

There's no absolute reason great games can't be as immortal as chess. Maybe Wesnoth can be.

A couple of months back I was looking for a fun quick project and also was having space issues on my Mac.

So github copilot and I wrote a little disk usage exploration tool in rust with very few dependencies.

It's called "syz" and I pronounce it "size" since it's all about exploring disk usage and figuring out what to trim.

Someone I respect started using it and reached out saying he liked it so I figured I'd dust it off and share it here!

After `cargo install syz`, use the `syz` command to enter the interactive cli starting at the current directory. Then use arrow keys to navigate directories and compare their recursive sizes.

My son just started using 2B on his Android. I mentioned that it was an impressively compact model and next thing I knew he had figured out how to use it on his inexpensive 2024 Motorolla and was using it to practice reading and writing in foreign languages.

Don't love touchscreens that much.

But I did love my Toshiba Satellite. It was like writing on paper!

Down with capacitive screens and long live Active Digitizers!

Had a broken key on my m2 air that I couldn't easily fix. Took it to an Apple store and a tech worked on it for a bit and came back with it fixed.

No charge. I was pretty grateful!

- making new friends does take a massive amount of time

I think the solution to this is to enjoy the journey. There's not a line that someone needs to cross before you can enjoy spending time with them. Just reach out and learn and enjoy people from the beginning.

I really like this passage:

It is always the case that there are benefits available from relinquishing core civil liberties: allowing infringements on free speech may reduce false claims and hateful ideas; allowing searches and seizures without warrants will likely help the police catch more criminals, and do so more quickly; giving up privacy may, in fact, enhance security.

But the core premise of the West generally, and the U.S. in particular, is that those trade-offs are never worthwhile. Americans still all learn and are taught to admire the iconic (if not apocryphal) 1775 words of Patrick Henry, which came to define the core ethos of the Revolutionary War and American Founding: “Give me liberty or give me death.” It is hard to express in more definitive terms on which side of that liberty-versus-security trade-off the U.S. was intended to fall.

I've been building a flutter django app around sharing and using checklists.

I use it myself by iterating on checklists and then tracking my usage of them and recently added orgs for privately shared checklists.

So it's easy to create an org around a shared task and then create a run through that task and track.

https://checkipe.com

I've been working on a task tracking app called Checkipe (https://checkipe.com but right now it's invite only for me and family members).

It's been fun to use it as I bring it to life.

A Django backend and flutter front end makes for a pretty powerful and adaptable pair.

Often on HN the comments are a better read than the original article.

This time the article is so good -- clear, funny, succinct, accurate -- that it's a better read than the comments.

Project Euler was what I used to get comfortable with my two main languages -- first python and then rust.

I loved to solve it and then look through all the different solutions and find pretty ideas and idioms.

Sounds like we need either a viable alternative or a next thing.

The next thing will probably be AR glasses and we could use some alternatives to Meta and Google and Apple.

Gardening automation :-D

If a friend told me they liked Rust but didn't like the borrow checker, I'd probably point them to Gleam and Moonbit, which both seem awesome in their own niches.

Both have rust-like flavor and neither has a borrow checker.

SEEKING WORK Huntsville AL USA, remote work preferred

Get more done, better, and in less time.

I help companies automate critical workflows, helping them accomplish their work more quickly, reliably, enjoyably, and correctly.

All the most useful techniques are on the table including task analysis and evaluation, traditional coding, and ai.

dhbradshaw@gmail.com

One of my favorites (only great if you read the site contents):

https://how.rl.works

SEEKING WORK | REMOTE | U.S. or Canada (Located in the Huntsville AL area.)

Automate business processes. Get more done, more accurately, faster, and less painfully.

PhD in laser physics / quantum optics, startup veteran, 15 years of experience in automating systems from laser control using real time video analysis to email and spreadsheet ingestion and routing.

https://github.com/dhbradshaw

Preferred tools: user interviews, systems analysis, documentation, rust, python, typescript, postgres, aws, local llms, evals and testing.

Free consultation, satisfaction guaranteed.

Reach out to dhbradshaw at gmail!

Great for space

Quote:

The Gemma 3 models are trained with distillation and achieve superior performance to Gemma 2 for both pre-trained and instruction finetuned versions. In particular, our novel post-training recipe significantly improves the math, chat, instruction-following and multilingual abilities, making Gemma3- 4B-IT competitive with Gemma2-27B-IT and Gemma3-27B-IT comparable to Gemini-1.5-Pro across benchmarks. We release all our models to the community.

Really cool

The more leverage a piece of code has, the more good or damage it can do.

The more constraints we can place on its behavior, the harder it is to mess up.

If it's riskier code, constrain it more with better typing, testing, design, and analysis.

Constraints are to errors (including hallucinations) as water is to fire.

Interesting to see this coming out of Apple

This is really cool -- going to show it off to my team. I love the fact that you opened it up so that it will work with Jupyter notebooks as well.

Hmm, could be a good time to set up as a contractor serving Salesforce

I like the idea of having a wiki be mostly serverless -- like lambda backed by s3. A little searching found this:

https://github.com/raboof/serverless-wiki

Tried out 3B on ollama, asking questions in optics, bio, and rust.

It's super fast with a lot of knowledge, a large context and great understanding. Really impressive model.