131–140 of 922 posts
I recently made a little tool for people interested in running local LLMs to figure out if their hardware is able to run an LLM in GPU memory. https://canirunthisllm.com/
We’re building a new git collaboration platform on top of atproto! Here’s a sneak preview (best viewed on desktop for now; the UI is mostly WIP): https://tangled.sh/@tangled.sh/core It’s going to be fully decentralised from day 1—we borrowed the PDS model from Bluesky to allow users to run their own “knots” to self-host their git repositories. https://www.tldraw.com/s/v2_c_E1XeFuW0tGbDxqnhDiLRT?d=v-286....
That's super nice. Can I eventually replace my centralized forge ?
Was looking for an iOS app to always see my age in days on the lock screen. Didn‘t find one, so I first created a shortcut which would change my lock screen background image each night and overlay the number of days on it. This didn’t feel integrated enough and could fail if the phone was off, so I started looking into Swift and created my first app [1] with added features like contact import and notifications for ot…
nice job , I think I can do something like that on with my linux desktop. Great idea , it can give a sense of how many days how many hours I have lived on this earth and maybe even by a average time , show how much time is left (yes it won't be predictable , but I also don't want to procastinate thinking there is a tomorrow , I think I like steve jobs quote in the manner that he said live your life as if its the last…
https://screenmemory.app is my current project as of a year or so. Records your screen continuously and lets you look back at it through a GUI. I use it myself to recap days or weeks at work, mostly.
Two new projects of note this month, one specific to Finnish language learners, and one that is probably useful for language learners in general:
* https://github.com/hiAndrewQuinn/tsk - A Finnish pocket dictionary with a TUI interface. This is the first nontrivial thing I've built in Go, by which I meant I had to implement and tweak a randomly pruning trie by hand to get the performance characteristics I wanted (it wasn't actually that bad). I chose Go mostly because of the fantastic cross-compilation story.
* https://github.com/hiAndrewQuinn/audio2anki - This Python program wraps around `yt-dlp` and `whisper` to create Anki decks for listening practice. This should work for any (monolingual) video in any language. There are many such projects on GitHub, I'm aware, but it was surprisingly hard to find any that actually wrapped around Whisper instead of needing an SRT, VTT, etc file to come from somewhere else. In that sense mine is a "one command" solution - just provide the YouTube link and go. It does not provide a translation for those subtitles yet; in keeping with the all-in-one approach, I'm thinking I might wrap around LLaMa 3 to let the user specify that we should also --translate-to {en,es,eo, etc} if desired. For now my reading skills are advanced enough that I don't need that.