Craig Henderson

When investors or M&A underestimate technical risk, the consequences are not abstract; they show up in missed targets, stalled growth, customer churn, and emergency engineering spend that was never priced into the deal. linkedin.com/pulse/cost-und…

🦞 BIG NEWS: We've molted! Clawdbot → Moltbot Clawd → Molty Same lobster soul, new shell. Anthropic asked us to change our name (trademark stuff), and honestly? "Molt" fits perfectly - it's what lobsters do to grow. New handle: @openclaw Same mission: AI that actually does things.

@iyoushetwt I think I'm a bit older than other people replying to you 😀

What was the first code editor you ever used? Mine was Sublime Text

Knowing how to write assembly is a skill you should learn, and these guys have a great resource for you! I've debugged 10,000 lines of assembly for every line I've ever written... but writing assembly from scratch is a core computer science skill, I believe: even if you never use it. Will you use it to write new code? Maybe not. But when you get dropped into a call stack without source code, at least you won't have to ask your Dad for help! And the next thing you should do after learning to write ASM is to get a crisp understanding of what your C code actually compiles to. Like a switch() statement is often a jump table, but if you've never debugged one...

@_trish_xD Fortran

Programming language you learned once but never touched again?

NVIDIA CEO Jensen Huang’s best career advice: “Passion isn’t enough, you’ve got to endure.”

Then x86 would be a RISC processor

@tswain555 @jasonvmiller I tried this in early 2000s and a 5+ yr developer. So-called principal rejected the code as unsafe and insisted it be wrapped in a strict to guarantee placement. I couldn’t be bothered to argue.

@jasonvmiller Laugh; it's a stack. that is next plate on the pile. Then again do you know where your stack is ^^.

Having problems with the M$ wsprintf. U suspect stack whap. so I do this. If it is getting stomped I won't see deadbeef in the debugger ^^ int canary1 = 0xDEADBEEF; wchar_t memory_name[256]; int canary2 = 0xDEADBEEF;

@Outlaw_Rabut @kurosci Not on Windows XP. We’ll stick with C++ for a bit longer

@kurosci Rust gets the job done.

Writing new stuff in C is just insanity.

@avatarsai18 @s0lst1c3 @MishManners @troyhunt x.com/brody_n77/stat…

@s0lst1c3 @MishManners @troyhunt Hello where can I see this fix?

Something super weird happening right now: just been called by several totally different media outlets in the last few minutes, all with Windows machines suddenly BSoD’ing (Blue Screen of Death). Anyone else seen this? Seems to be entering recovery mode:

@itsHesamSheikh @karpathy Translating one domain to another isn’t independent thinking, the definition of AGI

Btw writing the llm.c training code would imo be a very interesting, impressive, self-contained and very meta challenge for LLM agents. The prompt is: Take the PyTorch code train_gpt2.py And write, compile and unit test a single .c file that reproduces the training: train_gpt2.c The current models are not there, but we can check back in a year or two or so. If that worked...

Deleting 40% of LLM Layers Without Drop in Accuracy arxiv.org/html/2403.1788…

@lemire It’s not the language it’s written in that is the problem here

Microsoft should rewrite Visual Studio in Javascript. For speed.

@peterhil @BatsouElef 8088 was a math coprocessor. Maybe a 8080?

Do you remember the first computer you used? What was it? 💻

@BatsouElef I don’t remember the first I used, but this was my first - a birthday present, aged about 8 or 9. Sinclair ZX81. 1K of RAM, not external storage, plugged into tv

@olliethedev @karpathy Time will tell. I don't think you were wrong

I still remember when I studied neural nets in mid 2010's and thinking no way these will ever be good enough to match human level intelligence... Oh boy was I wrong!

Deep Neural Nets: 33 years ago and 33 years from now karpathy.github.io/2022/03/14/lec… My post from last year randomly made it to HN so resharing here too. Maybe in 2055 someone will train an improved GPT-4 on their personal computing device in ~1 min as an irrelevant fun weekend project.

@incomputable @chrisoldwood Genuine question, not wanting to sound like a troll... If you used ChatGPT, which is incapable of original thought, is this an original optimisation, or just one you've discovered now? How can you tell? If you can tell, and its the latter, then can you get a reference to source?

@chrisoldwood It's the implementation - interface stays the same. Improvements for all of lower bound, upper bound, equal range, and Boolean is/ain't.

Got some interesting results with binary search, passed them to the marketing team (i.e., ChatGPT), and... see you at NDC! ndctechtown.com/agenda/rethink…

@azeem @hamishmckenzie How many as first author?