whisperx does this all quite well and can be run with `uvx whisperx`

https://github.com/m-bain/whisperX

Yup. Framework 13 AMD with Fedora 41, gnome or sway. Works great.

I daily drive an AMD Framework 13 with Fedora. I do light software dev, writing, some gaming, and run a variety of local LLMs without pain. I'm happy with the machine and Framework's support, and I do like their vision of repairability/upgradability and I've taken advantage of that already. On the downside, it just clearly isn't as good of a machine for the cost compared to what you could get -- several aspects are just middle-of-the-road: the webcam, speakers and the battery life stand out for me as not great. The AMD chip is also new enough that I still run into glitches with kernel updates on occasion, and power efficiency was initially much worse but has been improving.

In the past, my jam has always been to buy used Thinkpads and run them into the ground. If my Framework got stolen today, I don't think buying another Framework would be a slam-dunk decision for me but I'd still consider it.

tl;dr: happy enough with the laptop and Framework support; hopeful that the dream of repairability/upgradeability pays off; but it does feel like a bit of a compromise on quality.

p.s. I'm quite sensitive to noise and really did not want a machine that ran its fan all the time; the AMD Framework 13 is pretty good for this unless gaming or under some other high load and then it's a rushing sound not a whine.

confabulation

I used to volunteer with Software Carpentry to teach and create lessons. Wonderful project, and so, so needed as many researchers (eg grad students) never get to properly learn fundamental data crunching skills and instead bumble along with poorly crafted spreadsheets and lots of cutting and pasting.

As someone who has seen what "research quality" software looks like, I can see the need for this

I recall this being essentially the inspiration for the naming of the Wake-Sleep algorithm of Boltzmann machines.

In the paper: "It is worth noting that the proposal of a "wake/sleep" specific algorithm for unsupervised learning of generative models based on feedback from stochastic stimulation goes back 25 years (Hinton et al., 1995)"

~/tmp - scratch folder for WIP stuff I don't care if I lose

~/dat - downloads and other temporary data

~/vir - virtual machines

~/arc - everything backup worthy

~/arc/dot - dotfiles, in "stow" format to easily set up on a new machine

~/arc/prj - projects, mostly software. Usually symlink to ~/prj

~/arc/doc - documents, organized by life-stage, project, company (e.g. "finances", "$company")

~/arc/img - pictures/videos

~/arc/dat - backup worthy software, keys, and other "data"

~/arc/res - research projects. Could go under prj/ I suppose.

~/arc/eml - IMAP email snapshot

I use Zotero and Mendeley (historical) for PDFs. I have a folder in ~/arc/dat/books for ebooks and the like.

> Just a few years ago, paramedics in Germany were routinely (and probably not entirely legally) using WhatsApp for communicating with the hospital

Here in Canada, WhatsApp is routinely used in the hospitals for doctors to communicate with each other about patient care. We aren't supposed to send identifiable information but... it happens all the time, of course. And for the same reason: no sanctioned system to replace it.

Does anyone have a recipe for using pihole via OpenVPN using docker?

I'm using this[1] but I'm surprised there isn't something more official/baked.

[1] https://github.com/mr-bolle/docker-openvpn-pihole

I just set this up yesterday, was quite easy.

https://hub.docker.com/r/linuxserver/openvpn-as/

source code: https://github.com/m1el/woscope/

The repo was only the source for the line drawing code.