Given that 50% of the time, the generated code is unworkable garbage, having an AI automatically write code to create new training models will either solve all problems, or spontaneously combust into a pile of ash.
My money’s on the latter.
Does “Please shut up and get to the point!” count?
Blocked all the server domains. There are a bunch of lists out there for various TV brands.
Blocked all this crap at the network level. Don’t get any ads now.
More details here: https://www.eufymake.com/eufymake-uv-printer-e1
Not clear how much the refills will run.
Be funny if someone started a gofundme.
Was working on a simulator and needed random interaction data. Statistical randomness didn’t capture likely scenarios (bell curves and all that). Switched to LLM synthetic data generation. Seemed better, but wait… seemed off 🤔. Checked it for clustering and entropy vs human data. JFC. Waaaaaay off.
Lesson: synthetic data for training is a Bad Idea. There are no shortcuts. Humans are lovely and messy.
Big fan of Geerling, precisely because he goes down these obscure rabbit holes. Found out about Meshtastic through him, and now BPS.
There are lots of applications to having a super-accurate time source, without having to have antennas and view of multiple satellites.
Synchronizing time is tricky. WWVB is too coarse resolution. NNTP requires access to the internet and all the inherent lags and delays. GPS was the only accurate source, but the super high resolution time signal is classified, and you are at the mercy of view of the sky. Also, signal jamming, thanks to what’s going on in Europe and Ukraine.
BPS could be a niche experiment, or a Big Deal.
Sans Serif strikes again.
The ground level weedwhacker blades could do double-duty clearing out brush (and pesky pedestrians).