On way to #AAS245 with a lot of heartbreak for our friends in LA. Looking forward to figuring out how the Astro community can best support them. Meanwhile we have this fundraiser and are working w/ NASA + Caltech/Carnegie friends on how best to distribute. 🔭 www.gofundme.com/f/support-na...

Fuckyeah!!!

This is pure (evil) genius

If you subscribe to Microsoft 365 Personal, the 40% price increase is a con to upsell you AI. Your old plan and price is still there called "Classic" - but only if you try to cancel.

Microsoft should be ashamed if not prosecuted.

Just one example of this program's capabilities -

SDR# driving one SDR (HackRF One) at 20 Msps (18.993 MHz usable B/W)

6 independent, 35-kHz "Sharp Slicer" USB receivers spread across that B/W

Each receiver feeding its own, indepent FT8 decoder

All running on a ($89 used) i5-7600T mini-PC

Can't wait to try it.
Such an amazing program!

New book for holidays

Scrapling : Undetectable, Lightning-Fast, and Adaptive Web Scraping for Python : github.com/D4Vinci/Scra... credits @D4Vinci1

The difference between hackers and Engineers is principally this -

Engineers assess requirements to derive implementations

Hackers assess implementations to derive functionality
-
Implementations invariably enable more functionality and alternative processing flows than required

Prompt Engineering against an LLM is more about reverse engineering the LLM's "implementation" than it is about human language and verbal communication

Don't treat LLMs as conversationalists

Treat them as an R/E-target that must be carefully explored

If "everyone" is supposedly abandoning Twitter and moving to BlueSky "because of how bad Twitter is now", why are so many people posting the same political bullshit on BlueSky that supposedly drove them off Twitter?

Don't repeat your mistakes; learn from them

One wild-ass LLM prompt that I came up with recently is -

"Generate a hierarchical taxonomy for <insert_target_here>"

I use this for hardware devices I am doing research on

This prompt, alone, repeated 10 - 20 times takes me from zero to no-marketing-hype understanding in just a few minutes

Treating Large Language Model (LLM) Prompt Engineering like a reverse-engineering challenge has some merit

Treat the LLM as a "hostile witness" that you have to "catch" via Q&A + logic

Yet another "app" loaded, account created

Let the (latest) deluge begin!