I’m impressed with your tech support skills!

Sorry, I am just gonne dump you some links from my bookmarks that were related and interesting to read, cause I am traveling and have to get up in a minute, but I’ve been interested in this topic for a while. All of the links discuss at least some usecases. For some reason microsoft is really into tiny models and made big breakthroughs there.

https://reddit.rtrace.io/r/LocalLLaMA/comments/1cdrw7p/what_are_the_potential_uses_of_small_less_than_3b/

https://github.com/microsoft/BitNet

https://www.microsoft.com/en-us/research/blog/phi-2-the-surprising-power-of-small-language-models/

https://news.microsoft.com/source/features/ai/the-phi-3-small-language-models-with-big-potential/

https://techcommunity.microsoft.com/blog/aiplatformblog/introducing-phi-4-microsoft’s-newest-small-language-model-specializing-in-comple/4357090

requires a very deep dive

On the contrary, in my opinion it was very intuitive. AI is also more helpful than with freecad when I ask it “how to do _ in blender/freecad?”

I’ve just created an OBS audio capture device, then opened qpwgraph and put both mic’s into it.

It should do what you want, but I am not sure that will fix the problem. Give it a shot and let us know.

I definitely recommend playing around with easyeffects a bit more as well.

didn’t know that, thank you!

Can rclone mount it transparently? I thought it is more like a one time copy / sync.

What I mean by that is that the remote storage should look like a normal directory to the rest of the system and any reads and writes should go over the network directly to the remote without occupying local disk space.

Also it seems to me that you have to write your credentials to the rclone config file, which I explicitly don’t want.

1. xpipe – I use it to SSH into any of my servers, cluster nodes or directly into docker containers without having to remember hostnames, IPs, users. It can also bring your useful scripts to said ssh session without “installing” them on the target device, which is great because you don’t have to set it up for every new server. Also the dev is a really nice guy.

2. Portmaster + SPN – I use it to route each app through different VPN paths with multihop support and per app firewall rules. (e.g. one app via Denmark, another via a random country, third app no VPN, fourth app gets no internet at all etc.) It really gives you full control over the traffic. afaik there is no other all in one app like this.

3. wdfs - It’s an old project that is patched by this random github user. It’s the only way I found to mount a webDAV storage cleanly into a directory from a bash script without fucking with my fstab or being root or giving specific privileges to my user. I mount it from a bash script because that way I can use KDE wallet to store the credentials instead of having a plain text file somewhere on my fs, the script waits until the wallet is unlocked, then reads the credentials from it and mounts the webDAV to a path in my home. That is more accessible to apps and other scripts (e.g. recent files) instead of doing it via Dolphin, which generates a random string in the path every time when opening network storage.

I’d love to not handhold, but where I worked that was not possible, because my task came from upper management and was to handhold juniors while guarding our production code from anything they did (which often didn’t meet company standards). The tasks the juniors were getting came from somebody else.

But also, this is not really about the style (throwing into deep waters on one extreme, babysitting on the other, and everything in between) but rather that I experienced way better results from chatGPT for 50$ / month than from a junior who had under a year of experience and costs 3000-4000$ a month. Both require some degree of attention and time investment from seniors if you want to use the resulting code in prod.

So the real goal is to pay them to get better than chatGPT and hope they stay at your company long enough to get a return out of that investment.

If you just want junior-grade code, let your seniors deploy chatGPT for a much cheaper price.

maybe drawio

I have no clue what I am looking at but it is absolutely mesmerizing.

I ran Linux with KDE on my phone for a while and it for sure needed EVEN MORE charging all the time even though most of the system is C, with a sprinkle of C++ and QT.

But that is probably due to other inefficiencies and lack of optimization (which is fine, make it work first, optimize later)

I’m using the fattest of java (Kotlin) on the fattest of frameworks (Spring boot) and it is still decently fast on a 5 year old raspberry pi. I can hit precise 50 μs timings with it.

Imagine doing it in fat python (as opposed to micropython) like all the hip kids.

They compile each benchmark solution as needed, following the CLBG guidelines, but they do not measure or report the energy consumed during the compilation step.

Time to write our own paper with regex and compiler flags.

Why?

(A super slimmed down flavour of) Java runs on fucking simcards.

WASM would be interesting as well, because lots of stuff can be compiled to it to run on the web

There is this, but I didn’t look into it yet https://www.automotivelinux.org/

And then there is stuff for motor controls, but it’s not linux (why would it be?) http://freeems.org/

Wow! I didn’t know

It does need a server though. Either the centralized official one you can selfhost one.

You’ve worked with juniors before?

Because in my experience I was constantly reading their unreadable code, then telling them why it’s wrong or bad or not fitting in a digestible manner and then waiting weeks for a refactor.

Iterate that for a month. Mentoring them took way longer than it would have taken me to write it from scratch. Not that dissimilar to trying to using AI for where it sucks (larger, a tad more complex problems).

It only makes sense if you look at it as an investment, because they will eventually improve.

a long runway that allows us to become profitable when needed

Switch to self-hosting headscale when they enshittify in an attempt to become profitable, duh