Arcuru

How much money did you give the Roo and Cline developers?

From the website, this is a "well funded" startup that is giving away $20 of credits for free to people who use it. They are literally forking 2 OSS projects and turning them into a funded startup, seemingly without giving anything back.

Is there an image of the shirt?

I assume that embedded Instagram post is a pic of the shirt, but I don't have Instagram so it's a big blank box with a link to "View this post on Instagram".

It's ironic that on this post about a large corporation abusing its power, they are requiring the user to use Instagram to get the full context..

Does anyone know how making the models multimodal impacts their text capabilities? The article is claiming this achieves good performance on pure text as well, but I'm curious if there is any analysis on how much impact it usually has.

I've seen some people claim it should make the models better at text, but I find that a little difficult to believe without data.

I've been labeling LLMS as "teensy", "smol", "mid", "biggg", "yuuge". I've been struggling to figure out where to place the lines between them though.

I am currently in the process of switching from using Obsidian to manage local markdown notes/files to using Notion for at least my personal project tracking. I have 2 reasons.

1. Notions "Databases" are easier to use, edit, and manage than any similar setup I could figure out locally. And they have inbuilt integrations so you can sync things like Github/Gitlab/Jira, etc directly into your docs. I highly recommend setting up a personal project inside Notion to try it out.

2. Less important but still useful, the collaboration is good to use with my partner. We can have shared household notes/saved info/tasks/etc all in one place very easily using it.

The submission is a link to the daily Matt Levine column where he typically rants his opinions on Finance.

I had a similar thought recently[0], though I was thinking about it more from the copyright side. I think there needs to be some kind of additional tax taken from AI profits, possibly justified by copyrights, because otherwise the economic impact might end up being nuts.

[0] https://jackson.dev/post/generative-ai-and-copyright/

And since this thread is talking about git hooks, there's also this project for managing git hooks inside Nix https://github.com/cachix/git-hooks.nix

I think they're referencing "Thinking, Fast and Slow" - https://en.wikipedia.org/wiki/Thinking,_Fast_and_Slow

"The book's main thesis is a differentiation between two modes of thought: "System 1" is fast, instinctive and emotional; "System 2" is slower, more deliberative, and more logical. "

> 5. Therefore all new development should be microservices written from scratch and interacting via cleanly defined APIs.

Not necessarily. You can get the same benefits you described in (1)-(3) by using clearly defined modules in your codebase, they don't need to be separate microservices.

> Much better :) But what about C? Let’s try it:

> I tried both versions (modulo 2 and bitwise AND) and got the same result. I think the optimizer recognizes modulo 2 and converts it to bitwise AND.

Yes, even without specifying optimizations - https://godbolt.org/z/9se9c6qKT

You can see that the output of the compiler is identical whether you use `i%2 == 0` or `(i&1) == 0`. The bitwise AND is instruction 12 in the output.

Using -O3 like in the post actually compiles to SIMD instructions on x86-64 - https://godbolt.org/z/dWbcK947G

Where I live we eliminated allowing tipped workers to be paid under minimum wage, and minimum wage is now at least $15/hour.

So yes, now I tip people a little less because I know they aren't being paid $2+tips/hour. That was the entire point of eliminating that weird loophole.

Just a minor clarification, DeepSeek's pricing for this model is temporary to match their previous model. They announced [1] that it will be the following after February 8:

DeepSeek - 0.27$ per million tokens input, 1.10$ million tokens output (66 tokens per/s)

Still much cheaper than the others though for input pricing.

[1] https://api-docs.deepseek.com/news/news1226#-api-pricing-upd...

I bought and use Comic Code [1] and use it as my main coding font. I actually really like it.

I work from home and only my wife has noticed...she thinks I'm crazy.

[1] https://tosche.net/fonts/comic-code

I'm not sure I am understanding where exactly this slots in, but isn't this an embedding model? Shouldn't they be comparing it to a service like Voyage AI?

- https://docs.voyageai.com/docs/embeddings

True, finding one solution is easy but finding all the solutions can be a fun little optimization challenge.

I made a repo many years ago with a bunch of grab bag solutions for comparisons [1]; from dumb brute force to DLX (Knuth's Dancing Links) and a multithreaded bitwise backtracking algorithm. And one where I just hardcoded the answers because all the counts up to 27 are known.

So I'm all for just jumping to the existing known solutions, but it seems like the OP is having fun while they learn a little bit. They seem to just be a college freshman.

[1] - https://github.com/arcuru/nqueens

That looks awesome. You didn't mention it but uithub.com also has an API, I can definitely see myself using this for a new tool.

I’m not the author but which SEA are you referring to? Southeast Asia? Southeast Africa? I assume you’re not talking about Seattle.

It’s important to be specific when you’re trying to localize things.

https://en.wikipedia.org/wiki/Sea_(disambiguation)

From the first paragraph of TFA:

> The latest recall—the wedge wagon’s sixth this year—requires shop time, not an over-the-air (OTA) update.

I hate bash, but it's always the first thing I start scripting in when I need to glue some things together. I tend to go in the following order as complexity and a need for robustness increases.

1. Bash (or ysh, which I do prefer but isn't as available)

2. Python, using the `sh` package as needed. It's been great for upgrading shell scripts, and `sh` is really nice to use.

3. Rust, usually. That's my personal language of choice for "real" projects.

I have a number of projects at each stage actually, though most of the things I'd call a real "project" with it's own repo end up as Rust projects.

The Boeing reboot is necessary because of an integer overflow that occurs once the aircraft is on for 248 days: https://www.i-programmer.info/news/149-security/8548-reboot-...

> This condition is caused by a software counter internal to the GCUs that will overflow after 248 days of continuous power. We are issuing this AD to prevent loss of all AC electrical power, which could result in loss of control of the airplane.

> A simple guess suggests the the problem is a signed 32-bit overflow as 231 is the number of seconds in 248 days multiplied by 100, i.e. a counter in hundredths of of a second.

Companies tend to focus on the overall % profit margin for a product. If a higher percentage of sales are for a discounted (edu) SKU with lower margins, they will tend to raise the price of the product to hit their desired profit margin.

e.g. If a company was selling a product at $1000, and wanted to offer a 20% discount for EDU that would be bought by 50% of the market, they would need to raise the price by about 10% to keep the same margin. If only 20% of the market bought the discounted SKU, they could keep the same margins with only a 5% (?) increase in price to the rest of the SKUs.

They're faster? They must have been working on performance then, maybe I should take another look at how that project has been maturing.

When I last looked 'dd' was significantly slower, though I did make it a bit closer a while back - https://jackson.dev/post/rust-coreutils-dd/

A lot of the Rust coreutils were written by people learning the language, and the code is often _more_ complicated than the corresponding code for the Unix utils. They don't seem to get enough experienced programmers fixing them up or usage for me to actually recommend anybody use them, but maybe that's changing.

Element X doesn't even support Threads or Spaces still, the two things that are absolutely required to organize discussions.

I don't think there's a setting for "only allow people from X server", but you can make the room private and only invite people from that particular server.

I use Chaz to just talk to any LLM from Matrix, my normal chat client.

https://github.com/arcuru/chaz

I access them by running the models in Ollama (on my own hardware), and then using my app Chaz[1] to access it through my normal Matrix client.

[1] - https://github.com/arcuru/chaz

There’s also Comic Code, which is a new font drawn in the same style. https://tosche.net/fonts/comic-code

Wow, and their current "pre-sales" price is $280. No wonder they don't have any dates on that page, or link to the Github repo that hasn't been updated in over a year.

https://github.com/accelerant-dev/implement-rougedb

No I can't. Because I use fish shell. `cat` can also be a single character if you alias it :)