Falst

joined 2 years ago
sorted by: new top controversial old
Cutting LLM token usage by 80% using recursive document analysis in c/programming@programming.dev
[–] Falst@lemmy.world 3 points 1 month ago

RLMs look promising.

Also stumbled upon https://github.com/wende/cicada a while ago which is definitely a different approach, context efficiency reminded me of it :)

Stop Doom Scrolling, Start Doom Coding: Build via the terminal from your phone in c/programming@programming.dev
[–] Falst@lemmy.world 2 points 1 month ago

There’s https://happy.engineering/ which is similar, less work as you don’t have to setup a VPN

Zero knowledge authentication in c/programming@programming.dev
[–] Falst@lemmy.world 2 points 11 months ago

SPAKE2+ is probably something worth researching as the server never receives the password. I believe it’s used in HomeKit and Matter as well 🙃